AU715628B2 - Hierarchical communication system providing intelligent data, program and processing migration - Google Patents

Description

T w S F Ref: 371021D1

AUSTRALIA

PATENTS ACT 1990 COMPLETE SPECIFICATION FOR A STANDARD PATENT

ORIGINAL

S

S

SO

0 S *0 0@00

S

*O 0 0

*S

0O50 5* *5 5* 0 0 5* Name and Address of Applicant: Actual Inventor(s): Address for Service: Norand Corporation 550 Second Street, S.E.

Cedar Rapids Iowa 52401 UNITED STATES OF AMERICA Ronald L. Mahany, Guy 3. West, Alan G. Bunte, Arvin D.

Danlelson, Michael D. Morris and Robert C. Meler Spruson Ferguson, Patent Attorneys Level 33 St Martins Tower, 31 Market Street Sydney, New South Wales, 2000, Australia Invention Title: Hieral Data, 0 0* The following statement is a best method of performing it rchical Communication System Providing Program and Processing Migration Intelligent full description known to me/us:of this invention, including the 5845 HIERARCHICAL COMMUNICATION SYSTEM PROVIDING

INTELLIGENT

DATA, PROGRAM AND PROCESSING

MIGRATION

TECiHNICAM FIELD The present invention relates ca-enera Ily to communication networks having a pluralitv of wired and/or wireless access servers configured to provide remote processing and data storaae. More specIfically, this invention relates to the intelligen:: rigration of programs and data through a wireless and nardwireo .communication network comprised of a plurality of access servers, computers and peripherals.

10 BACKGROUND OF THIE INVENTION Multiple radio base station network'-;! have been developed to overcome a variety of problem's with single *radio base station networks such as spanning physical :radio wave penetration barriers, wasted transmission D*ower by portable computing devices, etc. However, multiple radio base station~ networks have their own inherent problems. For example, in a multiple base :station network employing a single shared channel, each base station transmission is prone co collision with neighboring base station transmissions in the overlapping coverage areas between the base stations. Therefore, it often proves undesirable for each base station to use a single or common communication channel.

-1- In contradistinction, to facilitate the roaming of portable or mobile devices from one coverage area to another, use of a common communication channel for all of the base stations is convenient. A roaming device may easily move between coverage areas without loss of connectivity to the networ:.

Such exemplary competing commonality factors have resulted in tradeoff decisions in network design. These factors become even more significant when implementing a frequency hopping spread spectrum network. Freauencv 0 0hopping is a desirable transmission technique because of its ability to combat frequency selective fading, avoid o narrowband interference and provide multiple communications channels.

Again, however, changing operating parameters between coverage areas creates difficulties for the roaming devices which move therebetween. In particular when different communication parameters are used, a eve,: portable or mobile device roaming into a new base station 6* 20 coveraqe area is not able to communicate with the new base station without obtaining and synchronizing to the S new parameters. This causes communication backlog in the network.

Computer terminals and peripheral devices are widely used. Many types of computer terminals exist which vary greatly in terms of function, power and speed. Many different types of peripheral devices also exist, such as printers, modems, graphics scanners, text scanners, code -2readers, magnetic card readers, external monitors, voice command interfaces, external storage devices, and so on.

To communicate with such peripheral devices, portable computers have been adapted to use RF (Radio Frequency) and infrared communication. Such configurations, however, do not always provide for efficient communication. For example, a cort3ble computer device may be mounted in a delivery truck: and a driver may desire to transmit data to, or rece:ve data 10 from, a host computer or peripheral device at a remote warehouse location. While permitting such trans:ssions, •wide area networks (WANs) only provide point-to-point communications, use a narrow bandwidth, and often exhibit .heavy communication tratfic. Moreover, 4ANs require rplatively higher transmission power a negative factor in the ever increasing need for power savings associated wich portable transceiving devices. As a resu.t, WANs *9 9e •are generally slow and expensive, and simpl\ do not provide an effective overall solution.

The need for portable, cr otherwise mobile, devices has led to smaller, lower power designs. Portalle @9 .computer terminals have achieved such size and power 9reductions by decreasing local processing and storage resources. In contrast, application programs are growing in size and functionality, requiring more and more processing and storage resources to operate. As a result, portable computer terminals have been effectively disabled from independently performing many needed tasks.

Others have been stretched to a nearly unacceptable limit of portability, battery life and processing and storage ability.

To address such needed tasks, remote processing and storage techniques are currently being used. For example, stationary remote host computers having superior processing and storage capability are often connected via a WAN network to a mobile computer terminal. In such configurations, whenever the mobile terminal desires access to data, it sends a request across the WAN for such data. Similarly, when it desires remote processing, the mobile computer terminal formulates a request which is sent to the host computer over the WAN. However, the mobile terminal is still required to use the relatively expensive and delayed services provided by the WAN for each such request, which often prove unacceptable for a given task.

"tftf [N:\LIBK]01132:MXL Similarly, the relaying of communications through even lower power radio networks is required in many multi-hop radio environments. Repetitive requests and associated delivery of data, program or processing resources from a source a mobile computer terminal) to a destination a host computer) takes its toll on overall network performance.

It is an object of the present invention to ameliorate one or more disadvantages of the prior art.

o• a• a.

[N:\LIBK]01132:MXL -6- SUMMARY OF THE INVENTION Therefore, the invention discloses a communication network operating between a premises and a vehicle, comprising: a data source located at a premises; a terminal device within a vehicle; a first communication link between said data source and said terminal device; a vehicular network comprising a portable computing device and said terminal device; said portable computing device and said terminal device communicating via a second communication link, said second communication link comprising a wireless communication link; and [N:\LIBK]01132:MXL -7said terminal device being configured to store data delivered from said data source, and, upon communication from said portable computing device, selectively forwarding the stored data to said portable computing device.

Before storing requested data, processing resources, or program code, an access device may consider a number of factors including the cost of re-obtaining the requested data, processing resources, or program code, the frequency that the data, processing resources, or program code is requested, the amount of its available storage capacity, and the size of the data, processing resources, or program code.

S

00o 4

S

[N:\LIBK]01132:MXL The access device may selectively delete stored data, etc. and may consider the factors listed above before doing so.

The processing device may further comprise an image database for storing signals representative of two-dimensional images. Such images are used by the processing circuit for comparison with received signals so as to aid in the code identification process. A decode algorithm might also be used, either alone or in combination with attempted identification through image database comparison.

oo I A Ao A

A

A A A A

AS

*AS•°

A.

A. I,

A

A

*AAA

A

A

*A

A

A

A

[N:\LIBK]01132:MXL Moreover, in some configurations, the terminal device also monitors the flow of data to the portable computing device, and, based on such monitoring, said terminal device selectively migrates data into local storage. Similarly, in other configurations, the terminal device also monitors the flow of program code to the portable computing device, and, based on such monitoring, said terminal device selectively migrates program code into local storage. In yet other configurations, the terminal device monitors processing requests from the portable computing device, and, based on such monitoring, said terminal device selectively migrates programming resources into local storage.

polo*, .0 C. 0* [N:\LIBK]01132:MXL BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1A is a diagrammatic illustraticn of a hierarchal communication system built in accordance with the present invention.

Fig. 1B is a diagrammatic illustration of another hierarchal communication system built in accordance with the present invention.

Fig. IC is a diagrammatic illustra:ion of still another hierarchal communication syster built in 10 accordance with the present invention.

Fig. 2 illustrates an embodiment or a bas.c access interval structure used by a hierarchical network of the present invention.

Figs. 3A and 3B illustrate the frequency of operation periodically changing corresponding to access interval boundaries in a frequency hopping communication protocol of the present invention.

Figs. 4A and 4B illustrate more than one access interval being used per hop in a frequency hopping communication protocol of the present invention.

Fig. 5A illustrates an embodiment of an access S interval used by the hierarchical network of the present invention wherein a reservation phase is Idle Sense Multiple Access.

Fig. B illustrates an embodiment of an access interval used by the hierarchical network of the present invention wherein a device response follows a reservation poll.

Fig. 6A illustrates an embodiment of an access interval used by the hierarchical network of the present invention having multiple reservation slots for transmission of a Request For Poll signal.

Fig. 6B illustrates an embodiment of an access interval used by the hierarchical network of the present inventior wherein general devices contend for channel access.

Fig. 7A illustrates sequence in an access interval 0 used by the hierarchical network of the present invention for transferring data from a remote device to a control point device.

Fig. 7B illustrates a sequence in an access interval

S..

e• used by the hierarchical network of the present invention for transferring data from a control point device to a remote device.

Fig. 8 illustrates a preferred embodiment cf an access interval used by the hieiarchical network of the o present invention.

5* Figs. 9A and B conceptually illustrate how multiple NETs may be employed in an idealized cellular-type installation according to the present invention.

Fig. 10 illustrates an access point coverage contour overlap for the multiple NETs Infrastructured Network of Fig. 1.

Fig. 11 illustrates hopping sequence reuse in a multiple NET configuration of the present invention.

-11- Fic. -2 illustrates a hierarchical inltrastru::tured networKr of the present invention wnerein a wireless -ink connects access points on separatc h ard wir-d LANs.

Fig. 13 illustrates a hierarchical infraszructured network of the present invention including i wirejes access point.

Fig. 14 illustrates conceptually acces; .o-,nt-s comnrunicating neighboring access point ~no:.tc.to facilitate roaming ot portable/mobile devices- Fig. :5 illustrates a secondary access- irnt-rval used in the MicroLAN or peripheral LAN In the nirrarcnicil *communication netwoiz according to the presen nvention.

*Fi7. 16 is a flow chart Illustrating tnc- selection of an access point by a mobile coinputinq ce for communication exchange.

Fig. 17 is a flow chart illustrating c, term~na1 *.maintaining synchronization with the network aft'er .t nas gone to sleep for several access intervals.

.Fig. 1!3 is a flow chart illustratinoi ai erminai maintaining or achieving synchronization with tne network after it has gon.i to sleep for several seconas.

5 Figs. 20A and 10B are flow charts illuistratin~g an access interval during intbound communication.

Fig. 21 illustrates a sequence in an access interval used in the hierarchical communication network of the -12present invention with Time Division Multi pie Access slots positioned at the end of the access interv.al.

Fici. 22 illustrates a sequence in an access inlterv.al used by the hierarchical network of the oresent invention with the Time Division Multiple Access slots positioned immediately following the SYNC.

Fig. 23 illustrates a -;equence in an access inter--il used by the hierarchical netwo-k of the presen: :nvent_,cr with the Time Divisirn Multiple Access slots posit~ionec 0~0**10 immediately followirng the SYNC and Reservat~on Poll.

o .0 Fig. 24 i llustrates another spauence in an accecinterval used by the hierarchical network of the present invention with the Time Division Multiple Access slots positioned immediately toj~owing the SYNC.

Fig. 25 illustrates d portion of an access _ntlerv-ii including tne preamble, SYNC and 1heser\'ation Poll Fig. 26 jilustrates the information contained in a 0 simp~e S7NC message.

27 illustrat!es the information contained a .2C sample Reservation Poll.

Fig. 28A illustrates a warehouse tnvironment incorporating a communication net-work which maintains :communication connectivity between the various network, devices according to the present invention.

Fig. 28B illustrates other features of the present invention in the use of a vehicular LAkN which is capable of detaching from the premises LAN wnen movin~g out of radio caige of the premises LAkN to perform a service, and -13reattaching to the premises LAN when moving within ranoc to automatically report on the services rendered.

Fig. 28C illustrate other features of the present invention in the use of a vehicular LAN which, when out of range of the premises LAN, is still capable gaining access to the premises LAN via radio WAN communication.

Fig. 29A is a diagrammatic illustration of the use of a peripheral LAN supporting roaming data collection by an operator according to the present invention.

10 Fig. 29B is a diagrammatic illustration of anoth r embodiment of a peripheral LAN which supports roaming S* data collection by an operator according to the present invention.

a Fig. 30 is a block diagram illustrating the functionality of RF transceivers built in accordance with the present invention.

Fig. 31 is a diagrammatic illustration of an alternate embodiment of the peripheral LAN shown in Fig.

2.

20 Fig. 32 is a block diagram illustrating a channel access algorithm used by peripheral LAN slave devices in accordance with the present invention.

Fig. 33A is a timing diagram of the protocol used according to the present invention illustrating a typical communication exchange between a peripheral LAN master device having virtually unlimited power resources and a peripheral LAN slave device.

-14- Fig. 33B is a timing diagram of the protocol used according to the present invention illustrating a typical communication exchange between a peripheral LAN master device having limited power resources and a peripheral LAN slave device.

Fig. 33C is also a timing diagram of the protocol used which illustrates a scenario wherein the peripheral LAN master device fails to service the peripheral LAN slave devices.

10 Fig. 34 is a timing diagram illustrating the peripheral LAN master device's servicing of both the higher power portion of the premises LAN as well as the lower power peripheral LAN subnetwork with a single or plural radio transceivers.

Figs. 35 and 36 are block d-agrams illustrating additional power saving features according to the present invention wherein ranging and battery parameters are used to optimally select the appropriate data rate and power level of subsequent transmissions.

20 Fig. 37 illustrates an exemplary block diagram of a radio unit capable of current participation on multiple LANs according to the present invention.

Fig. 38 illustrates an exemplary functional layout of the frequency generator of Fig. 37 according to one embodiment of the present invention.

Fig. 39 illustrates further detail of the receiver RF processing circuit of Fig. 37 according to one embodiment of the present invention.

Fig. 40 illustrates further detail of the receiver signal processing circuit of rig. 37 according to one embodiment of the present invention.

Fig. 41 illustrates further detail of the receiver signal processing circuit of Fig. 37 according to another embodiment of the present invention.

Fig. 42 illustrates further detail of the memory unit of Fig. 37 according to one embodiment of the present invention.

10 Fig. 43 illustrates a software flow chart describing the operation of the control processor in controlling the battery powered radio unit to participate on multiple LANs.

Fig. 44 is an alternate embodiment of the software 9e flow chart wherein the control processor participates on a master LAN and, when needed, on a slave LAN.

Fig. 45 illustrates another embodiment of the *0 communica' on system of the present invention as adanted for servicing a retail store environment.

S* Figs. 46a-b illustrate a further embodiment of the communication system of the present invention which illustrate the use of access servers that support local processing and provide both data and program migration.

Fig. 47a is a flow diagram which illustrates the functionality of the access servers of Figs. 46a-b in handling data, processing and direct routing requests.

Fig. 47b is a flow diagram utilized by the access servers of Figs. 46a-b to manage the migration of data -16and program code from a source storage and/or processing device toward an end-point device.

Fig. 48 is a schematic diagram of the access servers of Figs. 46a-b illustrating an exemplary circuit layout which supports the functionality described in relation to Figs. 47a-b.

Fig. 49 is a specific exemplary embodiment cf an access point in a multi-hop communication network utilized for remote processing of 2-D (two-dimension) 10 code information.

Fig. 50 is a schematic diagram similar to that shown in Fig. 48 which illustrates the circuit layout used in the access point of Fig. 49 to process the 2-D code information.

Figs. 51a-b are flow diagrams illustrating the operation of the 2-P code processing access point of Figs. 49-50.

6 Fig. 52 illustrates the structuring of 2-D code information so as to support a hierarchical recognition strategy as used by the access point of 'igs. 49-50.

S

20 Fi. 53 is a diagram illustrating an exemplary 2-D code wherein the hierarchical structure of Fig. 52 is implemented.

Fig. 54 is a flow diagram illustrating the functionality of the access point of Figs. 49-50 in carrying out the hierarchical recognition strategy of Fig. 52.

-17- DETAILED DESCRIPTION OF THE INVENTION Fig. 1A illustrates a hierarchical communication system 10 within a building in accordance with the present invention. The illustrated hierarchical communication system 10 includes a local area network (LAN) for maintaining typical communication flow within the building premises, herein refer.-ed to as a premises LAN. The premises LAN is designed to provide efficient end-to-end routing of information among hardwired and 10 wireless, stationary and roaming devices located within the hierarchical communication system The premises LAN consists of an infrastructure S. network -omprising radio base stations, wirelesz access points 15, and a data base server 16 which may be part of a more extensive, wired LAN (not shown). Herein, base stations which participate in routing and relavina data throughout the communication network are referred to as "ac-ess points." If they also participate in the storage or migration of data and program code or in local 20 proces ing, the base stations are referred to herein as "access servers." As will become apparent below, an 0 access point may be modified with additional circuitry and/or programming resources to become an access server.

Additionally, access servers and access points are both referred to herein as "access devices." The access points 15 may communicate with each other via hard-wired links, such as Ethernet, RS232, etc., or via wireless (radio frequency) links. A plurality of -18roaming terminal devices, such as a roamin-i computinq device 20, participate in the premises c f the hierarchical commu .nication network 10 to exchange information with: 1) other roaming computing devices; 2) the data base server 16; 3) other devices which might be associated with data bast server 16 (not shown) and 4) any other devices accessible via the premises I-All (not shown). A roaming computing device can be, tor example, a hand-held computer terminal or vehicle mounted computer terminal (vehicle terminal).

I SIn most circumstances, the premises LAI; provides a rather optimal solution to the communication needs of a given network. However, in some circumstances, to serve a variety of particular communication needs, the premises LAN does not offer the optimal solution. Instead of relying on the premises LAkN for such communications, when and where beneficial, alternate LANs are spontaneously created by (or with) network cOevices, such as the roaming *computing device 20, within the hierarchical communication system 10. Such spontaneously created LANs are referred to herein as spontaneous LANs. After the immediate benefits end, a task has been completed, or if the participants of the spontaneous LAN move out of range of each other, the spontaneous ILAN terminates operation.

An exemplary spontaneous LAN involves the use of peripheral devices as illustrated in Fig. Although bulk data transfer destined for a peripheral device 23, such as a printer, from the roaming computing device -19might be communicated through the premises LAI, a more dilect interconnection proves less intrusive, saves power, and offers a lower cost solution. Specifically, instead of communicating through the premise LAN, the roaming computing device 20 needing to print: 1) identifies the presence of an available printer, the peripheral device 23; 2) establishes an RF link (binds) with the peripheral device 23; 3) directly begqns transferring the bulk data for printing; and 4 i lastly, 10 when the roaming terminal finishes the transfer, the spontaneous LAN with the peripheral device 23 terminates.

A spontaneous LAN created between the computing devices and peripheral devices is herein referred to as a peripheral LAN. Other types of spontaneous LANs, such as vehicular LANs, are also possible. Embodiments described below identify vehicular LANs and wide area radio networks (WANs) which are part of the hierarchical S* communication system according to the present invention.

Although a spontaneous LAN may operate completely 00 independent of the premises LAN, it is more likely that there will be some degree of coordination between the .two. For example, while participating in the peripheral

*U

LAN, the roaming computing device 20 may terminate participation in the premises LAal, and vice versa.

Alternately, the roaming computing device 20 may only service the peripheral LAN when specific participation on the premises LAN is not required, or vice versa.

Moreover, the roaming computing device 20 may attempt to service each peripheral LAN as necessary in a balanced time-sharing fashion, placing little priority upon either LAN. Thus, based on the protocols and hardware selected, a spontaneous LAN can be configured so as to exist hierarchically above, below, at the same level, or independent of the premises LAN.

In generally, to design a given LAN configuration, only the characteristics of that LAN are considered for optimization purposes. However, in the hierarchical S 10 communication system of the present invention, the operation of other LANs must also be taken into account.

For example, because of the roaming computing devices participation in both the premises and peripheral LANs, eq the requirements and operation of the premises LAN must be taken into consideration when defining the peripheral LAN, and vice versa. Thus, the hierarchical communication system of the present invention provides a series of tightly coupled radio LANs and WANs .ith radio transceiver and communication protocol designs which take *6 20 into consideration such factors as cost, weight, power conservation, channel loading, response times, interference, communication flow, etc., as modified by a at primary factor of multiple participation.

The peripheral LAN replaces hard-wired connection between a roaming computing device and associated peripherals. In a typical configuration, a peripheral LAN will consist of one or more peripherals slaved to a single master roaming computing device, although multiple -21master roaming computing devices are cossible.

Peripheral devices may be printers, code scanners, magnetic card readers, input stylus, etc.

Each of the peripheral devices 22 has a built-in radio transceivr to communicate with the roaming computing devices 20. The roaming computing devices are configured with built-in radio transceivers capable of communicating on both the peripheral and prei:ses LAN.

The access points 15 may be configured with radio 10 transceivers only capable of communicating in the premises LAN. In alternate embodiments, as Described below, the access points 15 might instead be configured to participate on both the premises and peripheral LANs.

o* In particular, the peripheral LAN is In=ended to provide communications between two or more devices operating within near proximity, distances of a few tens of feet. The majority of constituents of the *peripheral LAN are generally devices that do not require access to resources outside their immediate Qroup, or f 20 which can suffice with indirect access throuc devices which participate outside their immediate peripheral LAN group. In contradistinction, the premises LAN is intended to provide communications between relatively many devices operating across great distances throughout a building.

The characteristics of the peripheral LAN permit the use of radio transceivers of lower cost, lower power consumption, and generally more simplistic operation than -22permitted by the premises LAN. However, the operation ot the peripheral LANl is adapted for integration with tne premises LAkN so that a radio transceiver and protocol designed for operation on the premises LAN includes features which allow concurrent or sequentially concurrent operation orn the peripheral LAN. For example, by selecting similar communication hardware characteristics and integrating protoccls, communication within the premises and peripheral LANs may be achievcd with a single radio transceiver.

*In one embodiment, radio communication throuqh the ee premises LAN, i.e. among the access points 15 3na the roaming computing device 20, utilizes relatively higher- 0 power spread-spectrum frequency-hopping -ommunication with a reservation access protocol. The reservation access protocol facilita..es frequency-hopping and **supports adaptive data rate selection. Adaptive data rate selection is based upon the quality of communication :on the premises LAN radio channel. Radio communication through the peripheral LAN utilizes a relatively lowerpower single frequency communication also with a reservation acccss protocol. As more fully described below, the coordinated use of reservation access protocols in the peripheral and premises LANs maximize information flow while minimizing conflicts between devices participating in the two LANs.

Referring to Fig. 1B, a small hierarchal communication system 30 built in accordance with the -23present invention is shown. An access point 33 anc two roaming or mobile computing devices 35 and 36 fcrm a premises LAN4 37. ihe premises LAN 37 provides for communication among the mobile computing devices 35 and 36 and a host computer 34. The mobile computing devices and 36 can roam anywhere within the range of the access point 33 and still communicate with the nost computer 34 via the access point 33.

Two peripheral LANs 40 and 41 allow for wireless i0 communication between each mobile computinq device 35 and 36 and its respective peripheral devices 43, 44 anJ when the mobile computing device is not communicatincz on the premises LAN 37. Specifically, the peripheral LAN 0SS consists of the mobile computing device 35 and the peripheral device 43, while the peripheral LAI; 41 consists of the mobile computing device 36 and the two peripheral devices 44 and Fig. IC illustrates another embodiment accordina to the present invention of a larger hierarchal 20 communication system 50. The host computer 55 is connected to access points 56, 57, 58 and 59. The host 0computer 55 and the access points 56, 57, 58 and 59 5: provide the infrastructure for the premises LAN. The access points need not be hard-wired together. For example, as illustrated in Fig. iC, the access points 56, 57 and 58 access each other and the host computer 55 via a hard-wired link, while the access point 59 accomplishes such access via a wireless link with the access point 58.

-24- The access points 56, 58 and 59 can support multiple mobile computing devices. For example, the access point 56 uses a frequency-hopping communication protocol for maintaining communication with mobile computing devices 61 and 62. Moreover, each of the mobile computing devices may roam out of range of the access point with which they have been communicating and into the range of an access point with which they will at least temporarily communicate. Together, the host computer 55 and the 10 access points 56, 57, 58 and 59 and mobile computing devices 61, 62, 64, 65 and 66 constitte a premises LAN.

More particularly, each access point operates with a different set of communication parameters. For example, each access point may use a different frequency hopping sequence. Additionally, different access points may not employ a common master clock and will not be synchronized so as to have the frequency hopping 0* sequences start at the same time.

Mobile computing devices 61, 62, 64, 65 and 6r are capable of roaming into the vicinity of any of the access 6 points 56, 58 and 59 and connecting thereto. For o 0 example, mobile computing device 62 may roam into the 0* coverage area of access point 58, disconnecting from access point 56 and connecting to access point 58, without losing connectivity with the premises LAN.

Each mobile computing device 1l, 62, 64, 65 and 66 also participates with associated peripherals in a peripheral LAN. Each peripheral LAN is made up of the master device and its slave device. Similarly,- as illustrated, the access point 57 is shown as a direct participant in not only the premises LAN but also ir tne peripheral LAI. The access point 57 may either have limited or full participation in the premises LAN. For example, the access point 57 may be configured as a mobile computing device with the full RF capabi it of transmission in both the premises and peripheral LANs.

Instead, however, participation in the premises LAN 7mav 10 be limited to communicating through the hard-wired iink, effectively dedicating the access point 57 -o the task or servicing peripherals.

*Although the use of a plurality of built-in radio transceivers could be used so as to permit simultaneous participation by a single device, factors of cost, size, power and weight make it desirable to only bu-lJ-in a single radio transceiver capable of muitiole participation. Furthermore, even where a plurality of radir transceivers are built-in, simultaneous 20 participation may not be possible drpending upon the potential transmission interference between transceivers.

S* In fact, full simultaneous participation may not be

S.

*desirable at least from a processing standooint when one transceiver, servicing one LAN, always or usually t.,kes precedence over the other. Justification for such precedence generally exists in a premises LAN over a peripheral LAN.

-26- For example, communication flow in most premises LANs must be fast, efficient and rather robust wner, considering the multitude of participants tnat operate thereon. In the peripheral LAN, however, response times and other transmission related delays are generally more acceptable even adding extra seconds to a peripheral printer's print time will usually not bother the 'iser.

Thus, in such communication environments, it may be desirable to design the transmitters a..d associated 10 prctocols so that the premises LAN takes precedence over the peripheral LAN. This may yield a communicaticn system where fully simultaneous participation in both the premises and peripheral LANs does not exist.

In communication environments wherein fully simultaneous participation does ,ot ex st or is not desired, transmitter circuitry might be shared for ~participation in both the premises dnd peripheral LANs.

Similarly, in such environments, the communication protocol for the peripheral LAN can oe tightly coupled 'U with the protocol for the premises LAN, integLated protocols, so as to accommodate multiple participation.

Moreover, one protocol might be designed to take precedence over the other. For example, the premises LAN protocol might be designed so as to minimize participation or response time in the peripheral LAN. As described in more detail below, such transceiver and protocol analysis also takes place when considering -27additional multiple participation in the vehicular LAN and WAN environments.

Fig. 2 illustrates an embodiment of a communication protocol for the premises LAN which uses a basic Access Interval 200 structure according to the present invention. Generally, an Access Interval is the basic communication unit, a fixed block of time, that allocates bandwidth to synchronization, media access, polled communications, contention based communications, and scheduled services. The Access Interval in Fig. 2 includes a SYNC header 201 generated by a Control Point device of a NET. The term NET describes a group of users of a given hopping sequence or a hopping sequence itself. The Control Point device is aenerally the access point 15 referenced above with regard to Fig.

i. The SYNC header 201 is used by constituents of the NE to attain and maintain hooping synchronization. A reservation phase 203 follows permitting a reservation poll, which provides the NET constituents an opportunity 20 to gain access to media. A sessions frame 205 is next 00** allocated for communication protocol. A frame 207 follows for optional time division multiple access 0 ("TDMA") slots in order to accommodate scheduled services. Scheduled services, for example, real time voice or slow scan video, are such that they require a dedicated time slot to provide acceptable quality of service. The function of frames 201, 203, 205 and 207 will be discussed in greater detail below.

-28- As was shown in Fig. 2, Fig. 21 illustrates a sequence in an access interval 210C with the Time Division Multiple Access slots 2113 positioned at the end of the access interval 2100. In present example, if this were also a HELLO interval, the HELLO would immediately fo.low the SYNC 1201. Location of the Time Division Multiple Access slots at such a position provides certain advantages including, for example, 1) tne SYNC 2101, HELLO (not shown), Reservation Poll 2103, may all be combined into a single transmission (concatenated frames); 2) hopping information may be moved to or included in the Reservation Poll 2103 allowinq for a shorter preamble in the SYNC 2101; and 3) the HELLO messages will occur early in the Access Interval 2100 15 providing for shorter receiver on times for sleeping terminals.

The Time Division Multiple Access slots may also be located at dif'erent points within the access interval.

Positioning the Time Division Multiple Access slots allow 20 for various systemic advantages. Referring now to Fig.

22, an access interval 2200 is illustrated showing the Time Division Multiple Access slots 2203 immediately 0O following the SYNC 2201. Location of the Time Division Multiple Access slots 2203 at this position provides certain advantages including, for example, 1) better timing accuracy is achieved when the Time Division Multiple Access slots 2203 immediately follow the SYNC 2201; 2) Session Overruns do not interfere with the Time -29- Division Multiple Access slots 2203; 3) devices which do not use the Time Division Multiple Access slots 2203 do not necessarily need to be informed of the Time Division Multiple Access slot allocation; and 4) HELLO message may follow Time Division Multiple Access slots 2203, Reservation Slots 2207 or Reservation Resolution ;oll 2209.

Referring now to Fig. 23, an access interval 2300 is illustrated showing the Time Division Multiple Access 1 0 slots 2305 immediately following the SYNC 2301 and the S: Reservation Poll 2303. In the present example, if this were a HELLO interval, a HELLO messace would immediately follow the Reservation Resolution Poll 23-1.

Location of the Time Division Multiple Access slots 2305 at the position shown in Fig. 23 provides certain advantages including, for example, 1) the Time Division Multiple Access slot timing is keyed to SYNC 2301 for better accuracy; 2) the number of Time Division Multiple Access slots 2305 may be indicated in SYNC 2301 or the 20 Reservation Poll 2303, providing greater flexibility; 3) Session frame overruns do not interfere with Time Division Multiple Access slots 2305; 4) only one maintenance transmission is required per Access Interval 2300; and 5) hopping information may be moved to or included in the Reservation Poll 2303, permitting a shorter preamble in SYNC 2301.

In the acc-ess interval 2300 configuration shown in Fig. 23, it is possible that the Time Division Multiple Access slots 2305 and the response slots 2307 could be the same. The Reservation Poll 2303 would allocate the correct number of slots and indicate which are reserved for Time Division Multiple Access. For example, to use Idle Sense Multiple Access 1 slot) with 1 inbound and 1 outbound Time Division Multiple Access slots, three slots would be allocated with the first two slots reserved.

The apprcpriate Time Division Multiple Access slot duration is 80 bits at a hop rate of 200 hops per second 1 0 which is just about the expected duration of a Request for Poll. At slower hop rates, multiple slots could be allocated to Time Division Multiple Access allowing the Time Division Multiple Access slot duration to be constant regardless of hop rate.

Referring now to Fig. 24, another access interval 2400 is illustrated showing the Time Division Multiple Access slots 2403 immediately following the SYNC 2401.

In this example the Poll Message Queue 2405 immediately follows the Time Division Multiple Access slots 2403.

20 The configuration shown in Fig. 24 provides for certain S" advantages including, for example, 1) the Time Division Multiple Access slot timing is keyed to SYNC 2401 for better accuracy; and 2) Session frame overruns do not interfere with Time Division Multiple Access slots 2403.

The configurations shown in Fig. 21 and in Fig. 23 are preferred because they allow the Reservation Poll messages to be transmitted.immediately following the SYNC -31and because of the power management and interference reduction advantages.

In one embodiment of the Access Interval structure, all message transmissions use standard high-level data link control ("HDLC") data framing. Each message is delimited by High-Level Data Link Control Flags, consisting of the binary string 01111110, at the beginning of the message. A preamble, consisting of a known data pattern, precedes the initial FLAG. This 1 0 preamble is used to attain clock and bit synchronization prior to start of data. Receiver antenna selection is also made during the preamble for antenna diversity. A CRC for error detection immediately precedes the ending FLAG. Data is NRZ-I (differentially) encoded to improve data clock recovery. High-Level Data Link Control NRZ-I data is run-length-limited to six consecutive bits of the same state. Alternatively, a shift register scrambler could be applied instead of differential encoding to obtain sufficient transitions for clock recovery. Data 20 frames may be concatenated, with two or more frames sent during the same transmission, with a single FLAG separating them. An example of this is SYNC, followed by a HELLO or Reservation Poll (SYNC, HELLO and Reservation Poll are discussed more fully below).

While mucn of the following discussion centers on the use of frequency hopping in the premises LAN, the Access Interval structure of the present invention is also suitable for single channel and direct sequence -32spread spectrum systems. The consistent timing of channel access, and the relative freedom from collisions due to channel contention, provide desirable benefits in systems that support portable, battery powered devices regardless of modulation type or channelization.

Functions that are unique to frequency hopping may be omitted if other channelization approaches are used.

Figs. 3a and 3b illustrate the frequency of operation periodically changing corresponding to Access 0 Interval boundaries in a frequency hopping system.

Frequency hopping systems use a hopping sequence, which is a repeating list of frequencies of length selected in a pseudo random order and is known to all devices within a coverage area. Fig. 3a illustrates a frequency hopping system having one Access Interval 301 per frequency hop (the hop occurring every 10 milliseconds) and a length of 79. Fig. 3b illustrates a frequency hopping system having one Access Interval 303 per frequency hop (the hop occurring every 20 milliseconds) 20 and a length of 79. The 20ms time frame is preferred for a protocol stack that uses a maximum network layer frame of up to 1536 bytes payload while maintaining two real time voice communications channels. Access interval duration may be optimized for other conditions. Access Interval length is communicated to the NET during the SYNC portion of the Access Interval. This allows Access Interval duration, and other NET parameters to be -33adjusted without reprogramming every device -'Ithin the

NET.

The Access Interval is a building block. The length of the Access Interval can be optimized based on network layer packet size, expected mix of Bandwidth on Demand ("BWOD") and Scheduled Access traffic, expected velocities of devices within the NET, acceptable duration of channel outages, latency or delay for scheduled services, etc. The preferred Access Interval duration of 10 20ms (and maximum packet length of 256 Bytes at 1 MBIT/sec) represents a value chosen for systems with device velocities up to 15 MPH, and a mix between Bandvidth On Demand and scheduled service traffic.

9 Within a frequency hopping network, one or more Access Intervals may be used during each dwell in a frequency hopping system. A dwell is the length of time each frequency in the hopping sequence is occupied by *4 the system. For example, Figs. 4a and 4b show illustrations of cases where more than one 20 ms Access Interval 401 is used per hop. This may be appropriate for some instances where it is undesirable to hop at higher rates because of relatively long frequency switching times of the radio hardware, where import, export, or regulatory restrictions disallow hopping at a faster rate, or in some applications where it is desirable to maintain operation on each channel for a longer period. An example of the latter is the case -34where larger files or data records are transferred routinely.

In a frequency hopping operation, the Access Interval 200 of Fig. 2 begins with a SYNC header 201. As mentioned above, the SYNC is generated by the Control Point (CP) device of the NET. The SYNC is used by constituents of the NET to attain and maintain hopping synchronization. Included in the SYNC are: 1. Address of the Control Point device.

10 2. Identification of the Hopping Sequence, and e* index of the current frequency within the hop table.

29 Identification of the hop rate, number ot Access Intervals per hop, and Access Intervals 4 15 before next hop.

e** A timing character for synchronization of device local clocks to the NET clock contained within the Control Point device.

Status field indicating reduced SYNC transmissions due to low NET activity (Priority SYNC Indicator).

6. Status field indicating if the Access Interval will contain a broadcast message to all devices within the NET.

25 7. Status field indicating premises or spontaneous LAN operation.

8. The SYNC field information is optionally encrypted using a block encryption algorithm, with a key provided by the network user. A 30 random character is added to each SYNC message to provide scrambling.

However, there are two circumstances during which a SYNC message is not transmitted: 1) co-channel interference; and 2) low NET utilization. With regard to co-channel interference, before issuing a SYNC message, the Control Point device performs channel monitoring for a brief interval. If the Received Signal Strength Indicator (RSSI) level indicates an ON channel signal greater than the system defer threshold, then the Access Interval is skipped. Alternatively, a strong ON channel signal may dictate a reduction in Control Point device power to limit the interference distance of the net for the duration of the Access Interval. A system defer threshold 30 dB above the receiver sensitivity is a preferred choice. Communication within the NET is 10 deferred for the duration of the Access Interval if SYNC Sis not transmitted due to co-channel interference.

In times of low system utilization, SYNC and Reservation Poll messages are reduced to every third V. Access Interval. The SYNC message includes a status ar field indicating this mode of operation. This allows devices to access the NET, even during Access Intervals where SYNC is skipped, by using an Implicit Idle Sense algorithm. If the hopping sequence is 79 frequencies in length as shown in Figs. 3a and 3b, use of every third 20 Access Interval guarantees that a SYNC message will be a transmitted on each frequency within the hcpping sequence once each three cycles of the sequence, regardless of whether 1, 2 or 4 Access Intervals occur each hop dwell.

This addresses US and European regulatory requirements for uniform channel occupancy, and improves the prospects for synchronization of new units coming into the NET during periods when the NET is otherwise inactive. SYNC messages that are on multiples of 3 Access intervals are -36labeled as priority SYNC messages. "Sleeping" terminals use priority SYNCs to manage their internal sleep algorithms. Sleeping terminals and Implicit Idle Sense are discussed in more detail below.

It should be noted that SYNC messages are preceded by dead time, which must be allocated to account for timing uncertainty tween NET clocks and local clocks within NET constitue- In frequency hopping systems, the dead time must also include frequency switching time for the RF modem.

The Reservation Po.l frame 203 immediately follows the SYNC header 201. The two messages are concatenated High-Level Data Link Control frames separated by one or more Flags. The reservation poll provides

NET

constituents an opportunity to gain access to the media.

It includes: 1. A field specifying one or more access slots.

2. A field specifying a probability factor between 0 and 1.

3. A list of addresses for which the access points has pending messages in queue.

4. Allocation of Time Division Multiple Access slots for scheduled services by address.

2 5. Control Point device Transmitted Power level 25 for SYNC and Reservation Polls.

S* The number of access slots, n, and the access probability factor, p, are used by the Control Point device to manage contention on the channel. They may each be increased or decreased from Access Interval to -37- Access Interval to optimize access opportunity versus overhead.

If the NET is lightly loaded, the pending message list is short, and the NET is not subject to significant interference from other nearby NETs, the control point device will generally specify a single slot 501 as shown in Fig. 5a, with a p factor In this case, tht reservation phase is Idle Sense Multiple Access ("ISMA").

Devices with transmission requirements that successfully 10 detect the Reservation Poll will transmit a Request for Poll with probability p and defer transmission with probability 1-p. Fig. b shows a device response (address 65 503 following the reservation poll.

o* In cases when the transmission density is higher, n multiple reservation slots will be specified, generally with a probability factor p of 1. In this case a device will randomly choose one of n slots -or transmission of their Request for Poll. The slotted reservation approach is particularly appropriate in instances where many NETs 20 are operating in near proximity, since it diminishes reliance on listen before talk (explained more fully below). The number of slots n is determined by a slot allocation algorithm that allocates additional slots as system loading increases. Fig. 6a shows multiple slots 601.

In cases where NET loading is extreme, the Control Point may indicate a number of slots, not-more than 6, and a probability less than 1. This will cause some -38number of devices to defer responding with a Request for Poll in any of the slots. This prevents the control point device from introducing the overhead of a large number of slots in response to heavy demand for communications, by dictating that some units back off until demand diminishes.

A pending nessage list is included in the Reservation Poll. The pending message list 7:;ncudes the addresses of devices for which the Control Point device 10 has messages in queue. Devices receiving their address may contend for the channel by responding with a Request For Poll (RFP) in the slot response phase. Fig. 6b shows several devices 603, 605 and 607 contending for channel access. Messages that the Control Point device receives through the wired infrastructure that are destined for Type 1 devices, and inactive Type 3 devices whose awake window has expired, are immediately buffered, and the device addresses are added to the pending message list.

When a message is received through the infrastructure for 20 a Type 2 device, or an active Type 3 device, their address is prioritized at the top of the polling queue.

(Device Types and polling queue are described below.) The pending message list is aged over a period of several seconds. If pending messages are not accessed within this period, they are dropped.

Devices with transmission requirements respond in slots with a Request for Poll. This message type includes the addresses of the Control Point device and -39requesting device, the type and lenath of the message it has to transmit, and a field that identifies the type of device. Devices that detect their address in the pending message list also contend for access in this manner.

As mentioned above, devices may be Type 1, Type 2-, or Type 3. Type 1 devices are those which require critical battery management. These may be in a power saving, non-operational mode much of the time, only occasionally "waking" to receive sufficient numbers of 10 SYNC and Reservation Poll messages to maintain connectivity to the NET. Type 2 devices are those that are typically powered up and monitoring the NET at all times. Type 3 units are devices that will remain awake for a window period following their last transmission in anticipation of a response. Other device types employing different power management schemes may be added.

Slot responses are subject to collision in both the single and multiple slot cases. Collisions iay' occur when two or more devices attempt to send Request for 20 Polls in the same slot. However, if the signal strength of one device is significantly stronger than the others, it is likely to capture the slot, and be serviced as if it were the only responding unit. Fig. 6b shows two devices 605, address 111, and 607, address 02, that may be subject to collision or capture.

The Control Point device may or may not be able to detect collisions by detecting evidence of recovered clock or data in a slot, or by detecting an increase in RF energy in the receiver (using the Received Signal Strength Indicator, corresponding to the slot interval. Collision detection is used in the slot allocation algorithm for determining addition or deletion of slots in upcoming Reservation Polls.

As an optional feature to improve collision detection in the multiple slot case, devices that respond in later slots may transmit the addresses of devices they detect in earlier slots as part of their Request for Poll. Request for Polls which result in collisions at */oo0 the Control Point device often are captured at other remote devices, since the spatial relationship between devices that created the collision at the base does not exist for other device locations within the NET. The 15 duration of the response slots must be increased slightly to provide this capability.

If the Control Point device receives one or more valid Request for Polls following a Reservation Poll, it issues a Reservation Resolution Poll and places 20 the addresses of the identified devices in a polling queue. The Reservation Resolution message also serves as a poll of the first unit in the queue. Addresses from previous Access Intervals and addresses of intended recipients of outbound messages are also in the queue.

If the Polling Queue is empty, then no valid Request for Polls were received or collision detected and no Reservation Resolution poll is issued. If within this scenario a collision is detected, a CLEAR message -41indicating an Explicit Idle Sense (explained more full%, below) is transmitted containing a reduc- ed probability factor to allow colliding units to immediately reartempt NET access.

outbound me--sages obtained through the networkz infrastructure may result in recipient addresses being prioritized in the queue, that is, if the recipients are active devices Type 2 devices or Type 3 devices whose awake windo%, has riot expired. This eliminates the need for channel contention for many outbound messages, 9*improving efficiency. Messages for Type 1 devices are buffered, and the recipient address is placed in the pending message list for the next Access Interv~l.

Generally the queue is polled on a first in first out (FIFO) tasis. The polling order is: a. Addresses of active ur.its with outbound messages.

b. Addresses from previous Access Intervals C. Addresses from the current Access Interval *00.20 Since propagation character istics vary with time and operating frequency, it is counterproductive to attempt ~retries if Poll responses are not received. If a response to a Poll is not received, the next address in the queue is polled after a short response time-out period. Addresses of unsuccessful Polls remain in the queue for Polling during the next Access Interval.

Addresses are aged, so that after several unsuccessful Polls they are dropped from the queue. Addresses linked -42to outbound messages are added to the pending message list. Devices with inbound requirements must re-enter the queue through the next reservation phase.

Data is transferred in fragments. A maximum frag-ment payload of 256 bytes is used in the preferred implementation. If transfer of network packets larger tha n of 256 bytes is required, two or more fragments are transferred. Frdgments may be any length up to the maxirnum, eliminating the inefficiency that results when messages that are not integer multiples of the fragment length are transmitted in systems that employ fixed e~g sizes.

sequence for transferring data from a remote device to the control noint device is illustrated in Fig.

15 7a. It is assumed that address 65 is the first address in the polling queue. The Reservation Resolution poll 701 from the control point device includes the device address and the message length that device 65 provided in its initial Request for Poll. A first fragment 703 transmitted back from device 65 is a full length *fragment. its header includes a fragment identifier and a field providing indicationr of the total length of the message. Length information is included in most message C. types during the sessions period to provide reservation information to devices that may wish to attempt to access the NET following an Explicit Idle sense (explained more fully below).

-43- Following successful receipt of the first fragment, the Control Point device sends a second poll 705, which both acknowledges the first fragment, and initiates transmission of the second. The length parameter is decremented to reflect that the time required for completion of the message transfer is reduced. A second fragment 707 is transmitted in response, and also contains a decremented length field. Following receipt of the second fragnent 707, the Control Point device sends a third poll 709. This pattern is continued until a final fragment 711 containing an End of Data (EOD) indication is received. In Fig. 7, the final fragment is shorter than a maximum length fragment. The Control Point device sends a final Acknowledge (ACK), and the 15 device sends a final CLEAR 713 to indicate conclusion of the transmission. The CLEAR message contains a probability factor p for Explicit Idle Sense (explained more fully below). The value of p is determined by the Control Point device in the ACK and echoed by the device

S

termination communication. A p of zero indicates that the control point device will be initiating other communications immediately following receipt of the CLEAR message. A probability other than 0 indicates an Explicit Idle Sense.

If for some reason a fragment is not successfully received, the next poll from the Control Point device would indicate a REJECT, and request re-transmission of the same fragment. The length field would remain fixed -44at the previous value, prolonging reservation ofT the channel for the duration of the message- After a fragment is transmitted more than once without successful reception, the Control Point device may suspend attempts to communicate with the device based upon a retry limit, and begin polling of the next address in the queue.

A flow chart depicting how inboupd messages are received during an access interval is shown in Figs. 19A and 19B. A flow chart depicting how outbound messages are transmitted during an ac-:ess interval shown in Figs. 20A and .Outbound messages are transmitted in a similar fashion as inbound messages, with the Control Point and device roles largely reversed as illustrated in Fig. 7b.

41** 15 When the Control Point reaches an address In the queue for which it has an outbound message, the Control Point transmits a Request for Poll 721 identifying the address ot t-ie device and the length of the message. The response back from the device w.ould be a poll with an embedded length field. The same POLL/ FRAGME!NT/ACK/CLEAR structure and retry mechanisms as described above with regard to inbound messages in reference to Fig. 7a are .maintained. The CLEAR from the device indicates a C,,.probability p of zero. If the polling queue is empty, the Control Point may send a final or terminating CLEAR 723 containing a probability for Explicit Idle Sense.

All terminating ACK or. CLEAR messages contain fields to aid in synchronization of new units to the NET. The content of these fields is identical to that in the SYNC message, except that the timing character is deleted.

Synchronization is discussed more fully below.

Broadcast Messages intended for groups of addresses, or all addresses within a NET may be transmitted during the sessions period. Broadcast messages are not individually acknowledged. These messages nay be communicated at intervals over the course of several Access Intervals to provide reliable communication.

Messages such as SYNC and Reservation Polls are specialized broadcast messages, with dedicated bandwidth in the Access Interval structure.

Security of payload data is left to the higher

SO

protocol layers. Application programs resident in 15 portable/mobile devices may employ encryption or other means of providing protection against undesired use of transmitted data.

Portable/mobile devices may employ transmitter power control during the sessions period to reduce potential interference with other NETs that may occasionally be on the same or adjacent channels. These devices will use Received Signal Strength Indicator readings from outbound •messages to determine if transmitter power may be reduced for their inbound transmission. Because of the need to maintain channel reservations and Listen Before Talk capabilities, the Control Point device does not use transmitter power control. Since Control Point devices are generally part of an installed system infrastructure, -46they are likely to be physically separated from devices operating in other NETs. They are therefore less likely to cause interference to devices in other NETs than portable devices, which may operate in proximity to devices in other NETs.

Often, control point devices will empty the polling queue before the conclusion of the access interval. Two mechanisms within the Access Control Protocol, Explicit and Implicit Idle Sense, are provided to improve bandwidth utilization. These supplemental access mechanisms often provide means for devices that failed to gain reservations during the reservation phase to gain access to the NET within the Access Interval. To assume an Explicit or implicit Idle Sense, a device must have 15 detected a valid SYNc and Reservation Poll in the current Access Interval.

The incorporation of a probability factor p e 0 in the final (terminating) ACK or CLEAR from the control point device provides the function of an Explicit Idle Sense (mentioned above). Devices with transmission requirements solicit Request for Polls using the same rules normally used for a single slot Reservation Poll.

Successfully identified addresses are placed in the polling queue, and are polled immediately or in the next Access Interval depending on the time remaining in the current Access Interval. The p factor for Explicit Idle Sense is subject to the same optimization algorithm as the Reservation Poll probability.

-47- Communication of channel reservation~s, inf the form of the length fields in Polls and Message Fragmnents is useful to units seeking to access the NET through Explicit Idle Sense. Reservations allow devices to predictably power down during the period that an3ther device has reserved the NET to conserve battery power, without loosing the ability to gain access to the NET.

Implicit Idle Sense provides an addit ional means of channel access. An Implicit Idle Sense is assumed whernever a device detects a quiet interval period greater than or equal to the duration of a Poll plus the maximum a.....fragment length after a channel reservation has ex<pired.

Detctonbased upon simple physical metrics, such as a U change in Received signal Strength Indicator or lack of 0e15 receiver clock recovery during the quiet interval, are *4Spieferred methods of ascertaining channel activity.

Algozithms based upon these types or indicators are generally less likely to provide a false indication of an 0inactive channel than t:Aose that require successful .4 4 20 decoding of transmissions to determine channel activity.

False invocation of an Implicit Idle Sense is the only mechanism by which data transmissions are subject to within the NET. Thus, the Implicit Algorithm must be conservative.

Quiet interval sensing m~ay begin at the following times within the Access Irterval: a. Any time after the last reservation slot following a Reservation Poll;- -48b. Any time after a terminating ACK or CLEAR indicating an Explicit Idle Sense; c. Following an unsuccessful response to a single Slot Reservation Poll; or d. Any time prior to reserved Time Division Multiple Access time slots at the end of the Access Interval.

It is preferable that devices detecting a quiet interval use a p persistent algorithm for channel access to avoid collisions. The probability factor for Implicit Idle Sense Access will generally be less than or equal to the factor in Explicit Idle Sense.

00.000 A device must receive the SYNC and Reservation Polls o*S* at the beginning of an Access Interval to use Implicit *also. 15 Idle Sense. The Reservation Poll provides indication of 4 *0 00,* guaranteed bandwidth allocation to scheduled services at S the end of the Access Interval, which may shorten the period available for Bandwidth On Demand communications.

Devices requiring schedule.- services must contend S• 20 for the channel in the same fashion as those requirinci Bandwidth On Demand access. When polled, these r initiating devices will initiate a connection request that indicates the number of inbound and outbound Time Division Multiple Access slots required for o 25 communication, and the address of the target device with *S which communication is desired. The network infrastructure will then attempt to establish the connection to the target device. Once the connection is established, the Control Point device will signal the allocation of slots to the initiating device. Time -49- Division Multiple Access slots are relinquished by transmitting a disconnect message to the control point device in the Time Division Multiple Access slot until the disconnect is confirmed in the next Reservation Poll.

The transmission requirements of speech and slow scan video (scheduled services) are similar. In one embodiment, Time Division Multiple Access slots are allocated as multiples of 160 bits payload at I MBIT/sec, plus overhead for a total of 300ps. For 10 ms access intervals, acceptable voice communication can be obtained S. by allocating 1 Time Division Multiple Access slot each for inbound and outbound communication per access interval. For 20 ms access intervals, two slots each way 0* are required. A system employing 10 ms access intervals at 100 hops per second may improve tr-ns ssion quality by using two or three slote each Access Interval and sending information redundantly over two or three access intervals using interleaved block codes. Scheduled transmissions are generally not subject to processing or 20 validation by the control point device, and are passed through from source to destination. Use of interleaved error correction coding or other measures to improve reliability are transparent to the NET.

The selection of certain system parameters are important when considering scheduled services. As an example, since speech is quantized over the duration of the access interval and transmitted as a burst, the length of the access interval translates directly into a transport delay perceptible to the recipient of that speech. In real time voice communications, delays longer than 20 ms are perceptible, and delays longer than 30 -s may be unacceptable. This is particularly the case where the premises LAN is interconnected with the public switched telephone network which introduces its own delays. Two way services such as voice communications are the most sensitive to transport delay because delay impacts the interaction of the communicating parties. One way services are less sensitive to transport delay. One way services are good candidates for interleaving or other forms of redundant transmission.

o* Similarly, the selection of hop rate is important, 15 as hoo rate determines the duration of outages that may occur. If one or more frequencies in the hop sequence are subject to interference, for instance, scheduled transmissions during those hops will be disrupted. in a system that hops slowly, detrimental outages of hundreds 20 of milliseconds will occur resulting in poor transmission quality. Occasional losses of smaller durations, e.g., ms or 20 ms, are generally less perceptible, 00 indicating that faster hop rates are desirable if the NET is to offer real time voice transport.

Scheduled service intervals may also be used for data transport on a scheduled or priority basis.

Telemetry, data logging, print spooling, modem replacement, or other functions are possible. For these -51activities, a few Time Division Multiple Access slots scheduled for examole every fourth, eighth, or sixteenth Al are necessary.

Because of multipath and dispersion issues with 2.4 GHz transmission at relatively high data rates, the ability of the NET to adaptively switch between two or more data rates is desirable.

In one embodiment, implementation ot data rate switching may be accomplished by selecting a standard 1P rate of communications, 250 KBPS and high rat- cf 0 000000 o communications of 1 Mbit/sec. Messages that contain 00 0 0 system status information, including SYNC, Reservation 0 Polls Reservation Resolution Polls (Request for Polls), 0 9 Polls, ACKs and CLEARS are transmitted a: the :tandard 000.

09* 0 15 rate. These messages are generally shor-, and the time required for transmission is largely determined by hardware overhead, transmitter receiver switching 00* time. The incremental overhead introduced by 0* transmitting these messages at the lower rate is 00 0 20 therefore small in comparison to the total length of an access interval. The reliability of reception of these 0 messages will increase, which will eliminate unnecessary 0 retries in some instances where fragments are received 0 successfully, but acknowledgements or polls are missed.

A test pattern at the higher data rate is inserted in each. Poll (not in Reservation Polls, however). The Poll recipient evaluates signal quality based on the high data rate test pattern, Received Signal Strength -52- Indicator, and other parameters tc determine whether to transmit a frag~ment at the high rate or the low rate.

.Frag-men t lengths are selected such that high and low rate maximum fragment lengths are the same duration. In other words, a fragment at the low rate conveys approximately 1/4 the payload of a fragment for the case where the data rate is four ti-me greater. This method is generally suitable for transaction oriented communications, which frequently require short message transmissions.

Alterna~tively, the length field in P'olls and messages can be used to allow different fragment lengths for the two goo *.data rates while :-till providing channel reservation information to other devices in the NET. This method SS**also provides for forward migration. As modulation and goo** *15 demodulation methods improve, newer products can be added to old networks by upgrading Control Points devices.

Both new and old devices share the ability to communicate at a common low data rate.

S *An alternate embodiment use,. signaling messages such as SYNC, Reservation Polls, Request for Polls, etc., at **Ethe higher rate with fallback operation to the £:tandard rate for the communications sessions only. SY!NC and *Reservation Polls at the high rate constit ute a high data 0* *rate test message. The Request for Poll rpsponse to the Reservation Poll at the high rate may include a field indicating that sessions communications should take place at the fallback, standard rate. Signal quality measures such as signal strength and clock jitter are appropriate.

-53- Data rate selection information is included with the device address in the polling queue. When the device is polled, it will be polled at the rate indicated in the Request for Poll. Channel reservation information in the Reservation Resolution Poll will indicate the reservation duration based upon the data rate indicated.

In this alternate embodiment, the fact that SYNC and Reservation Polls must be detectable at the hiar. data rate prioritizes access to the NET for those devices that have acceptable connectivity during the current access interval. This general approach has desirable characteristics in a frequency hopping system, as the Spropagation characteristics between devices may chanae significantly as the NET changes from frequency to 1 frequency within the hopping sequence, or over several Access Intervals during the dwell time on a single frequency. Reduction in data rate in this system is primarily intended to remedy the data smearing (intersymbol interference) effects of dispersion due to excess 20 delay, rather than temporary poor signal to noise ratio due to frequency selective fading. Devices that receive high data rate transmissions with acceptable signal C strength but high jitter are likely to be experiencing the effect of dispersion.

The concept of allowing Polls and message fragments to occur at either a high or low data rate could create difficulties for other NET constituents that need to be able to monitor the channel for reservation information.

-54- Two embodiments for solving this problem are the use of auto-discriminating receivers or the use of fixed data rate headers for system communications.

Auto discrimination requires the receiver to process messages sent at either data rate, without necessarily having prior knowledge of the rate.

Given a high rate of 1 MBIT/SEC, and a low Rate of 250 KBPS, one being a binary multiple of the other, it is possible to devise preambles that can be received at either rare. Consider that 01 and 110 sent at the low *o rate correspond to 00001111 and l111111l0000 at the high rate. These preambles are transmitted continuously before the transmission of the High-Level Data Link Control FLAG character at the correct data rate 15 indicating the start of a message. In this exanple, a preamble of 20 bits of 01 at the low rate indicates operation at the high rate. A preamble of 30 bits of 110 indicates operation at the low rate. A receiver tuned to either rate is capable of receiving both types of 20 preambles and initiating the proper decoding mechanisms for the intended rate of transmission.

This general technique, with appropriate selection of preamble content, is applicable to binary modulation schemes, for example, a frequency modulated system where a common frequency deviation value is used for both data rates. It is also applicable to systems where switching occurs between binary and multilevel modulation, such as disclosed in pending U. S. Application Sezr. No.

07/9l0,?65, filed July 6, 1992.

Referring now to Fig. 25, a preamble 2501, a SYNC 2503 and a Reservation Poll 2505 is illustrated. The preamble 2501 starts at the beginning of the Interval 2500 and is applied to an RF modem while it is switching f requencies. Since the s witching time is a worst case, this causes the preamble 2501 to be present and detectable prior to the allocated i504sec period in some instances. It would be equally appropriate to begin O&OGO:preamble transmission 50 or 100 4isec into the switching perioa if that would be more convenient. The c'iming has been selected to allow 100 jisec.

*0 Referring to Fig. 26, a sample SYNC message 2600 is 15 shown. Referring to Fig. 27, a sample Reservation Pohi 2700 is shown. In these c'xamples, the hopping synchronization information has beer, positioned in the Reservation Poll 2700.

with auodiciinto, it is possible to change data rates on a per-poll basis, thereby adjusting for ~channel temnporal dynamics. Since all devices in the NET have auto discrimination capabilities, and channel S reservaticn information Is incl~udedt in message headers as lerngth field, the bandwidth reservation features of the NET are preserved. The maximum fragment duration may be maintained at a fixed value, meaning that low data rate fragments convey less data than their high rate -56counterparts, or may be scaled in the ratio of the data rates to allow consistent fragment data payloads.

An alternative to auto-discrimination is the use of headers to communicate system information. This embodiment is less preferred, but may be appropriate if econcmics, size, or power constraints dictate a simpler design than that required for auto-discrimination. In this embodiment, any transmission at the lower data rate is preceded by a header at the high data rate that conveys NET management information, channel reservation status. Devices other than those directly involved in polling or fragment transmission need only monitor at the high rate for channel reservation information. The header at the high rate and the 0 S 15 following transmission at the low rate are concatenated SHigh-Level Data Link Control frames, with an apprerpiate preamole for low rate clock recovery synchronization inbetween.

For the communicating devices, the header can serve 9309 9* 20 the additional purpose of acting as a test patterr at the 9. o high rate. For example, if a device is polled at the low rate, but successtully decodes the high rate header with 9 adequate signal quality, it may indicate back to the 5 0 polling unit to poll again at the high rate.

0* 0 In a premises LAN as discussed in reference to Fig.

1, many NETs may be distributed geographically to provide enhanced coverage or additional system capacity. The wired portion of the network infrastructure, such as -57- Ethernet or Token Ring, pz-ovides a meens for coordination of NETs to achieve optimum system performance. An equally important role of the wired infrastructure is to allow resource sharing. Portable devices with limited memory capacities, processing power, and relatively small batteries may access large data bases on, or remotely initiate processing capabDilities of, larger AC powered computer systems. Portable/mobile devices may also share communication with other like devices which are serviced by other NE-Is well beyond the radio coverage range of their own NET.

4 The basic method for communication of status 4 information regarding the premises LAN is the HELLO message. HEL LO messages are sent routinely, but relatively infrequently, for example, every 90 Access **SIntervals. The HELL-0 transmission interval is tied to the Priority SiYNC interval, so that the HELLO interval corresponds to Access Intervals where SYNC is transmitted if the network is lightly utilized.

0*20 In an alternate embodiment, HELLOs could be inserted 04 as a broadcast nmessag: at the beginning of the Sessions period. Fig. 8 illustrates a preferred Access Interval embodiment where a HELLO message 801 is inserted between O ~9a SYNC 803 and a Reservation Poll 805. The SYNC frame at the beginning of the Access Interval indir7ates that the Access Interval will contain a HELLO, allowing power managed devices to remain awake to receive the HELLO.

-58- HELLO messages may also contain information regarding pending changes in the local NET. If the local NET is changing Access Interval durations or hop sequences, for instance, changes may be communicated in several consecutive HELLOs so that the information is reliably communicated to all NET constituents, permitting all devices to make the change in coordinated fashion.

Further discussion of HELLO message content is provided below.

1 0 For purposes of channel management in the Access t Interval structure, the maximum transmission duration by a device should be limited to the time that the device 4 moving at a maximum expected velocity can traverse 1/4 wavelength of the maximum carrier frequency. The 15 duration may be further reduced to compensate for link bit error rate characteristics or expected duration or frequency of interfarence bursts. A maximum transmission duration of 2.5 ms is suitable for 1 MBIT/SEC transmission, with a device velocity of 15 mph, i- a multiple NET environment.

S'a Use of spatial or polarization antenna selection diversity is also desirable in indoor propagation environments. First, the receiving unit makes an antenna diversity decision during the preamble portion of each transmission. The antenna used for reception for each device address is then recorded in memory so that the correct antenna will be used for response messages to each address. While diversity selection is only valid -59for a short time, it is not necessary to age this information, because antenna selection is equi-probable even after diversity information is no longer valid.

The Access Interval structure of the present invention also inherently provides routine channel sounding for each hop. This is important in a frequency hopping system, as channel conditions will vary considerably from frequency to frequency within the hopping sequence. NET constituents must, in most cases, be able to receive SYNC and Reservation Poll transmissions from the Control Point device to attempt inbound access in an Access Interval. This provides a positive indication that the device is not experiencing a channel outage, allowing power saving and eliminating 15 possible channel contention. Channel sounding does not need to be emplbyed during periods where the NET is not busy since contention is unlikely in this situation.

Channel sounding for Outbound messages is accomplished through a Request ror Poll/Poll cycle where handshaking messages with short time out periods must be successfully communicated before longer message 6* transmissions may be attempted.

As discussed above with regard to Fig. 1, a premises LAN consists of several access points 15 located throughout an environment requiring wireless communications, a building or other facility, or a campus comprised of several buildings. The access points are placed to provide coverage of intended usage areas for the roaming portable or mobile computing devices Coverage areas must overlap to eliminate dead spots between coverage areas.

The access points 15 may be interconnected via industry standard wired LANs, such as IEEE 802.3 Ethernet, or IEEE 802.5 Token Ring. Access points may be added to an existing LAN without the need to install additional LAN cable. Alternatively, it may be desirable to install access points on dedicated LAN segments to maximize performance of both the radio network and other S* collocated computer devices.

Access points within the premises LAN provide Control Point functions for individual NETs. NETs employ different hopping sequences to minimize potential *o 15 interference between NETs. Regulatory restrictions generally preclude synchronization of multiple NETs to a single mas-er clock, requiring that individual NETs operate independently from one another. The lack of the ability to coordinate timing or frequency usage between 20 NET introduces the potential for collisions between independent NETs with overlapping coverage areas.

Figs. 9a and b illustrate conceptually how multiple o NETs may be employed in an idealized "cellular" type installation. Each hexagon 901 and 903 in Fig. 9a represents the primary coverage area of a given NET.

Coverage areas are modeled as circles 905 based upon some reliability criterion, for example a 5% mean fragment retry rate (on average 95% of fragments are successfully -61communicated on the first attempt). Typical coverage areas are determined by physical attributes of the area in which the NET operates. As is illustrated in Fig. b for the hexagon (NET) 903 of Fig. 9a, an actual coverage area 907 meeting the reliability criterion is likely to be irregular. This may require access points to be offset significantly from the hexagonal grid.

Fig. 10 illustrates a coverage contour overlap for the multiple NETs in the premises LAN of Fig. 1. Darken 10 shaded areas 1001 indicate areas where access point S. coverage overlaps. Because the coverage distance of a radio system on an instantaneous basis greatly exceeds the coverage that can be provided on average to sustain a given quality of service, the overlap at any instant may be significantly greater than the coverage contours indicate.

Fig. 11 illustrates hopping sequence reuse in a multiple NET configuration. Hopping sequence re-use may be necessary if there are physical constraints on the number of hopping sequences that can be supported. For example, devices may have limited memory available for S* hopping sequence storage. Use of a smaller set of sequences also simplifies the task of determining sets of sequences that have acceptable cross correlation properties. In Fig. 12, 7 hopping sequences 1 through 7 are u-,ed throughout the coverage area. Other NETs may reuse the same hopping sequence at some distance removed.

While 7 NETs are illustrated, larger numbers, such as 9 -62or 15 may provide a better compromise between minimizing the number of hopping sequences used, and reuse distance .between NETs using the same sequence. Reuse requires coordination of hopping sequence assignment either the system installer can coordinate the installation, or the system may include automated management features to assign hopping sequences to individual NETs.

Since NETs are not synchronized, different NETs that use the same hopping sequence are likely to interfere during periods where oscillator drift causes them to be temporarily synchronized. At other times, they may only interfere due to imperfect channelization. For example, for a worst case 100 ppm frequency error between two NETs o. using the same 79 frequency sequence at one Access Interval per hop and 50 hops per second, NETs will partially or fully overlap for a duration of IC minutes every 4.3 hours. Typically the frequency error will be to 50% of the worst case, leadinq to longer overlap periods occurring less frequently.

20 NETs using the same hopping sequence must be physically isolated from one another to reduce interference to an acceptable level. Extensive hopping 0 sequence reuse generally requires site engineering and optimization of access point placement. Using more hopping sequences reduces the need for critical system engineering during installation. Fifteen hopping sequences is a preferred number for hopping sequence -63reuse, allowing simplified installation and minimal coordination.

NETs that use different hopping sequences will also temporarily synchronize in timing relationships that cause mutual co-channel interference on common channel frequencies. Since the number of channels that must be used in a sequence is a significant fraction of the total number of channels available, all sequences will share some number of frequencies in common. When sequences are 10 time aligned so that a common frequency is used simultaneously, interference can occur. Opticization of sets of sequences for low cross correlation is necessary

S.

to prevent various time alignments of sequences from having more than one or two frequencies in cormon.

Optimization of hopping sequences for multiple NETs must also include analysis of imperfect channelization.

The performance characteristics of the RF modems may not, S* *for economic or power consumption reasons, provide sufficient transmitter spectral containment, receiver S* 20 dynamic range, or receiver selectivity to guarantee that devices operating on different frequencies in proximity S* to one another will not interfere. In selecting hopping sequences for desirable cross correlation properties, aajacent and alternate adjacent channel interference must be considered. Protocol retry mechanisms for fragments lost to adjacent channel interference or limited dynamic range may be randomized to prevent continued disruption of communications in the affected NET.

-64- Often in campus environments where systems must provide coverage in several buildings, the cost of wiring LAN cable between access points is prohibitive. To establish connectivity between access points in an premises LAN, it may be necessary to provide wireless links between groups of access points connected to separate LAN segments. Fig. 12 illustrates a wireless link 1201 connecting groups of access points 1203 and 1205. The access points 1203 and 1205 are connected on separate LAN segments 1207 and 1209.

In one embodiment, the access points 1203 and 1205 may be configured in a wireless point to point mode, wherein one access point serves as a control point device while the others operate in a slaved mode dedicated to point to point data transfer. Slave access points are configured to operate as portable/mobile devices, and forward communications to master bases by sending Request for Polls during reservation opportunities or Implicit Idle Sense periods. Because of the potential high 2C traffic of point to point links, separate NETs may be allocated for this purpose, with a master communicating with one or more slave units. Master units may also communicate with other portable/mobile devices. The COST weighing (discussed below) in a slave's HELLO transmission is preferably set to a high value, to force portable/mobile devices which can connect to another NET to do so.

In another embodiment, it may also be desirable to support wireless access points. Wireless access points serve. as control points, but are not connected to the infrastructure through a LAN cable. As is illustrated in Fig. 13, a wireless access point 1301 participates in the premises LAN through a wireless link 1303 to an access point 1305 that is connected to a LAN 1307.

Wireless access points operate as slave devices to master access points which are connected to the wired infrastructure. The wired and wireless access points share the same hopping sequence, and are synchrcnized as S" a common NET. Because they are not connected to the Infrastructure, wireless access points must De used as store and forward devices. Each transmission to a wireless base must be retransmitted to the intended destination device, doubling the number of transmissions occurring in the NET. Wireless access points are preferably used for supplemerting coverage area of the premises LAN. For example, a wireless access point night 20 provide spot coverage of isolated "dead spots" where data traffic is limited or where providing a wired LAN connection is difficult. Wireless access points may also 0 serve as emergency spares to provide coverage in the event of a failure of a primary access point. In this role, the wireless access point may be either permanently installed in selected locations, or stored in a maintenance area and quickly positioned and connected to AC or battery power to provide communications while -66repairs are made to the primary wired access point.

Moreover, permanently installed wireless access points *might also be used for redundancy, to monitor an associated access point and to take over when a breakdown is detected.

The preferred wireless access point embodiment uses interi.ved access intervals. The parent wired access point and secondary wireless access point coordinate Access Intervals, the wired access point deferring every third or sixth access interval to the wireless base.

Since the wired access point transmits priority SYNC messages every third Access Interval, the wireless access point may routinely be allocated one of the two

S.

intervening Access Intervals for priority SYNC communications with devices that are attached to it.

Communication between the wired and wireless access points may occur during Access Intervals initiated by Seither access point. Wireless access points may also communicate with devices during an Access Interval using S* 20 Implicit or Explicit Idle Sense.

This embodiment provides predictable access for devices attached to the wireless NET, and allows the same 0 power management algorithms to be usei regardless of whether the access point is wired or wireless. The wireless access point may transmit its own priority SYNC and HELLO messages. Also, devices seeking communications with the wireless access point will automatically be synchronized with the wired base as well, allowing -67immediate improved access to the network if their mobility has put them within range of the wired base.

Because of the constraint of sharing bandwidth with a wired access point, connectivity of wireless access points is normally limited to one per wired access point.

However, in cases where system loading is predictably and consistently light, multiple wireless access points could share a single wired base, each transmittin in turn in the Access Intervals between the Wired Base Priority SYNC Access Intervals.

Wireless access points are capable of supporting scheduled traffic. However, since each transmission to a wireless access point must be forwarded, scheduled *4 transmissions through wireless access points use twice 15 the bandwidth as those through wired access points. In other words; twice the number of Time Division Multple Access slots must be allocated. To avoid introducing excessive delay, communications must be forwarded during the same Access Interval that they are received, or shorter Access Intervals must be used. Scheduled traffic slot Pssignments must be common to all wireless oases operating within a single NET.

Wireless access points require reliable communication with their wired counterparts. This dictates smaller coverage contours for wireless access points. If a wired access point provides 80,000 square feet of coverage area, a wireless base can be predicted to provide only an additional forty percent coverage -68improvement, due to overlap with the wired access point.

Frequently, access points are mounted at ceiling level, providing a relatively clearer transmission path between access points than exists between bases and portable/mobile devices locatea in more obstructed areas near the floor. With careful site engineerinq and installation, a wireless access point can provide somewhat better than the forty. percent predicted improvement, but still less than the coverage of an 10 additional wired base.

As discussed above, HELLO messages are used to i* communicate NET and premises LAN status messaces. They

S*

facilitate load leveling and roaming within the premises LAN and allow sequence maintenance to improve security and performance within the NET. HELLO messages occur periodically in Access Intervals that contain priority SYNC messages. HELLOs are sent periodically relative to *9 we:. the sequence length, for instance, every 90 Access Incervals. HELLOs, like SYNC information, are optionally .9 20 encrypted to provide greater security.

Each HELLO message includes a field for COST. COST is a measure of the access point to handle additional traffic. A device determining which of tv.o or more access points having adequate signal strength to register which will select the base with the lowest COST factor.

The base computes COST on the basis of how many devices are attached to the NET, the degree of bandwidth utilization, whether the base is wired or wireless, the -69number of frequencies experiencing consistent interference within th- sequence, and the quality of zhe coi.nection the base has within the premises LAN.

Fig. 14 illustrates the concept of access points communicating neighboring access point informat ion through HELLO messages to facilitate roaming of portable/mobile devices. In a premises LAN, access points 1401, 1403 and 1405 communicate SYNC information amongst themaselves via wired backbone (LAN) 1407. In addition, a wireless access point 1409 (discussed above) S" similarly communicates with the access points 1401, dO q Sand 1405 via a wireless link 1411. A portable/mobile device 1413 is initially registered with access poilnt 1401, which acts as a control point for the 9C09 portable/mobile device 1413. HELLO messages transmitted by access point ]401 to portable/mobile device 1,13 cot.tain fields for neighboring access points 1403, 1405 9 4 and 1409. These fields may indicate, for example, 4* 4addresses of tho neighboring bases, their COST, the C. 20 hopping sequences, hopping sequence indices, number of 4 OS Access Intervals per hop, and NET clock. The 00 04 pcrtable/mobile device 1413 detrects the HELLOs

U

transmitted from access point 1401 and uses the 09 information for coarse synchronization with the other access points 1403, 1405 and 1409. This permits the portable/mobile device to roam between access point coverage areas between different NETs) without going through a full acquisition phase. Roaming of portable/mobile devices is discussed in more detail below.

Simply put, communication of neighbors' infor ition permits each access point to advise its associated portable/mobile devices those having common communication parameters) on how to capture HELLO messages from neighboring access points having different communication parameters. Such communication parameters may include, for example, hopping sequences, spreading 10 codes, or channel frecuencies.

S* For example, neighbors' information transmission is 0* appropriate in any case where the system uses more than a single channel. For instance, in a direct sequence architecture, a single spreading code is often used.

Capacity can be added to such a network by emplr ying different spreading codes at eiach access point. The neighbors' information included in the HELLO message from a given access point would include the spreading sequences of access points providing coverage in adjacent 20 coverage areas. Likewise, in a multiple frequency channelized system, HELLO messages would include the *b channel frequencies of adjacent access points.

In addition to facilitating roaming, communication of neighbors' information may also facilitate the initial selection of an access point by a portable/mobile device attaching to the premises LAN for the first time.

Access point HELLO messages may also facilitate adaptive access point transmitter power control. For -71example, each access point HELLO transmissioll could specify the transmitter power level being used by the access point. If a given attached portable/mobile device noites that the current access point transmitter power level is unnecessarily high (creating the possibility of interference with other access points) the portable/mobile unit could sen'd a message to the access point indicating as such, and the access point could adjust the transmitter power level accordingly.

HELLO messages also enable communication of information indicating to all devices that 2erta~n changes in the NET are required. For example, the NET1 *fees:may switch hopping sequences periodically to improve security, or to avoid interference sources that consistently interfere with one or two frequencies with~n a given sequence. Interference may result from outside sources, or from other NETs. Changes tcu the NET are *communicated over the course of several HELLO messages (with a countdown) before the change occurs, so that all devices are likely to be aware of changes and synchronize at the instant of change.

5000In addition, if encryption *s used, the encryption :.sskey may be periodically changed in HELLOs. Like hopping 00 0sequence changes, KEY changes are sent over several HELLOs, and are encrypted using the existing key until the change goes into effect.

As mentioned above, roaming portable and mobile computing devices operating_ in the premises LAN will -72routinely move between access point coverage areas. At the maximum device velocity and expected coverage area per access point, a mobile device may be expected to cross a NET coverage contour in several seconds. Because of the use of multiple, non-synchronized frequency hopping NETs, it is more difficult to provide for simple hand-off between access points than it would be in a system that used cellular techniques with a single frequency per cell. The premises LAN makes special provisions for roaming by transmitting coarse frequency hopping synchronization information in HELLO messages.

S* The premises LAN uses a spanning tree algorithm to maintain current information regarding the general location of mobile devices within the network. When a *o 15 device changes registration from one NET Control Point to another, routing information is updated, throuchout tne infrastructure. Wired access points may broadcast spanning tree updates to attached wireless access points.

0* In tne premises LAN, roaming portable and mobile "0 20 devices initially select and register with an access point Control Point on the basis of link quality, i.e.

signal quality, signal strength and COST information transmitted within HELLO messages. A device will remain attached to a particular access point until the link 0 quality degrades below an acceptable level, then it will attempt to determine if an alternative NET is available.

Different device operating scenarios dictate different roamirn strategies, discussed below.

-73- An idle device monitors SYNC and HELLO messages from the Control Point device to maintain NET connectivity.

Type 2 devices do not employ power management, and always maintain their receivers in an active state. They monitor all SYNC messages. Type 1 and Type 3 devices typically employ power management, operating in standby or sleep modes of operation for many Access Intervals before activating their receivers for monitoring SYNC and HELLO messages. Control Points are guaranteed to send Priority SYNC frames every third Access Interval. HELLOs occur every 30th Priority SYNC frame. Power managed 0 devices employ sleep algorithms synchronized to wake for the minimum period necessary to guarantee receipt of priority SYNC, HELLO, and Pending Message transmissions 15 before resuming

SLEEP.

Type 2 devices are typically operated from high capacity vehicular power systems, which eliminates the need for power management. These devices may travel at velocities near the maximum system design specification, dictating more frequent roaming. Type 2 devices will S..

initiate a search for an alternative NET if SYNC messages 0 are consistently received at signal strengths below a Roaming Threshold or if reception errors are consistently S: detected. Because of the effects of frequency selective fading, signal strength information is averaged over the course of several hops within the hopping sequence.

If roaming is indicated, the device initiates a Roaming Algorithm, using Neighbors' information from the -74- 0 most recent HELLO to attempt synchronization with another candidate NET. if SYNC is not detected within 6 hops, another candidate from the Neighbors list will be selected, and the process repeated. Once SYNC is attained on an alternative NET, the device will monitor signal strength and data errors for several hops to determine link quality. If link quality is acceptable, the device will continue monitoring until a HELLO is received. If COST is acceptable, it will then register with the new NET. The Control Point device will update the spanning tree over the wired backbone (or by RF if a wireless base) If link quality or COST is unacceptable, a 0.0 a another candidate from the Neighbo:rs list is selected and the process repeated. This continues until an acceptable 15 connection is established. If a connection 'anno- be established, the device must return to the or- iFT or employ the initial acquisition algorithm.

Type 2 devices also have the option of :ionitoring 0* other NETs before degradation of their NET connection.

They may do so by monitoring their own NET for the SYNC and pending message list transmissions, then scanning o0 other candidate NETs during the Sessions period of their 0* NET. Other type devices may do so less frequently.

0 00 Type 1 and Type 3 devices may sleep extensively when idle, preferably activating every nine Access Intervals to resynchronize and check penling messages. Successful reception of at least one SYN' during three monitoring periods is necessary to maintai fine synchronization to the NET clock. Failure to receive two of three SYNC frames, or receipt of two or three SYNC messages with -poor signal strength are possible indications of the need to further test link quality by remaining active for several consecutive SYNC transmissions. If signal strength or data errors over several hops indicates that link quality is poor, or if a received HELLO message indicates high COST, the roaming algorithm is initiated, and alternative NETs are evaluated, as in the case of Type 2 devices.

o aSome battery powered devices may sleep for periods

S

of time more than nine Access Intervals. For example, devices with extremely limited battery capacity may sleep e- betweeii HELLOs, or several HELLO periods, after which they must remain active for several consecutive Access Intervals to regain fine synchronization and assess whether to initiate roaming.

A Type 1, Type 2, or Type 3 device that has inbound message requirements immediately activates its receiver and waits for a SYNC and subsequent Reservation Opportunities. A device that does not detect SYNC messages over the course of six Access Intervals immediately initiates the Roaming Algorithm.

6 a Outbound messages for devices that have changed coverage areas, but which have not yet registered with a new Control Point device, are problematic. For example, in the premiset LAN, messages will be forwarded to the access point that the device had previously been attached -76to. The access point may attempt to poll the device during one or more Access Intervals, then transmit the unit address in the pending message list periodically for severa. seconds before disregarding it. Once the unit attaches to a base, the message must be transferred from the previous access point for delivery to the unit. All of thes- activities require transmission bandwidth on either the backbone or RF media, waste processing resources within the premise LAN, and result in delayed delivery.

As this premises LAN embodiment is designed, the *0 network has no means of distinguishing messages it cannot deliver due to roaming from messages that should be retried due to signal propagation characteristics, 15 interference, or sleeping devices. For this reason, the roaming algorithm may be designed to allow devices to quickly detect that they have lost connectivity within their current NET, and re-attach to a more favorably located access point.

Some improvement in delivering pending messages to 4 roaming terminals can be obtained by routinely propagating pending message lists over the wired backbone. When a device attaches to an access point, that base is able to immediately ascertain that the 0* device has a pending message, and initiate forwarding of the message for delivery to the device.

In the preferred frequency hopping embodiment of the present invention, the hopping sequence consists of 3m -77- I frequencies, where m is an integer. 79 frequencies are preferred. This embodiment will support hopping rates of 100, 50 hops per second at 1 Access Interval per dwell, hops per seccnd at 2 frames per dwell, and 12.5 hops per second at 4 frames per dwell. Other rates can be supported for other Access Interval Durations. For example, if the Access Interval is optimized to 25 ms, hop rates of 80, 40, 20, and 10 hops per second would be supported.

All devices within the NET may have one or more hopping tables that contain potential hopping sequences that may be used. Up to 64 sequences may be stored in each device. Each sequence has an identifier, and each frequency in each sequence has an index. The sequence 15 identifier and index aze communicated in the SYNC transmission.

All SYNC transmissions may be block encrypted to prevent unauthorized devices from readily acquiring Oe* hopping synchronization information. To facilitate S 20 encryption, the encryption key may initially be factory set to a universal value in all devices. Users would then have the optiun of changing this key, by providing a new key to each device in the system. This may be 0 accomplished through keyboard entry or other secure means. Keys may also be changed through the NET.

To facilitate hopping management, a hopping control portion of a protocol controller will download a hopping table to a radio modem, and will signal the radio modem -78when to hop. This approach consolidates timing functions in the protocol controller, while not requiring the controller to be concerned with conveying frequency selection data to the modem each hop.

The NET may switch hopping sequences periodically to improve security, or to avoid interference sources that consistently interfere with one or two frequencies within a given sequence. As mentioned above, changes to the NET are communicated over the course of several HELLO messages before the change occurs so that all devices are likely to be aware of changes.

Initial synchronization requires devices to ascertain the hof'ing sequence, the hop rate, anJ the 0.

specific frequency from the hopping sequence currently in 1 use. Synchronization informetion is contained in two types of routine messages. The SYNC field at the beginning of an Access Interval contains synchronization information including the hopping sequence, the index of the current frequency within the sequence, the number of Access Intervals per hop, and the length of the Access SInterval. It contains a timing character that Se communicates the NET master clock to all listening 0* devices. Termination messages in the Sessions period, ACK and CLEAR, contain the same information, but do not contain the timing character.

The simplest method for attaining synchronization is to Camp select a quiet frequency that is likely to be within a sequence in use and listen for valid -79synchronization information. If a SYNC message is detected, the listening device immediately has both coarse and fine synchronization, and can begin the registration process.

If SYNC is net detected, but a termination message is, then the device has acquired coarse synchronization.

The particulars of the hopping sequence are known, but the boundaries of the dwells are not. To acquire fine synchronization, it begins hopping at the indicated hopping rate, listening for SYNC. If SYNC is not detected after a reasonable number of hops, preferably 12 or 15, the device reverts to camping.

The worst case scenario for synchronization is to 6 synchronize to a single NET that is idle. Given a 79 15 frequency hopping sequence, one Access Interval per hop, and SYNC transmissions every third Access Interval if the NET is idle, it may take nine cycle times to guarantee that a SYNC transmission will be detected with 99.5% probability. At 50 hops per second, synchronization 0.

could require as long as 14 seconds. At !00 hops per S* second, 7 seconds is required.

0 At 2 Access Intervals per hop, a SYNC transmission is guaranteed to occur every frequency over 2 cycles of the hopping sequence. Six cycles are required for 99.5% probability of acquisition, corresponding to 19 seconds at 25 hops per second.

At 4 Access Intervals per hop, at least one SYNC is guaranteed to occur each hop. Three cycles of the hopping sequence are required for 99.5% acquisition probability. At 12.5 hops per second, this also requires 19 seconds.

This illustrates the advantage of scalability.

A

device that uses an acquisition algorithm suitable for 2 or 4 Access Intervals per hop will also acquire a NET that hops at 1 Access Interval per hop. The algorithm may be as follows: 1. The device scans candidate frequencies until iL finds one with no Received Signal Strengtn Indicator indication.

2. The device remains on the frequency for 6.32 seconds 2 Access Interval/hop 25 Hops/second x 2, or 4 Access Interval/hop 12.5 hops/second x 1, or until it detects a SYNC message or a valid termination message.

3. If SYNC is detected, the device synchronizes its internal clock to the SYNC, and begins hopping with the IiET for tne next 11 hops. It may attempt registration after detecting valid SYNC and any Reservation Opportunity. If synchronization is not verified by detection of SYNC within the 11 hops, the acquisition algorithm is reinitialized.

4. If a message termination (either an ACK or CLEAR) is detected, the device immediately hops to the next frequency in the sequence and waits for the SYNC. It is coarsely synchronized to the NET but has a timing offset from the NET clock.

When the next SYNC is received, the device 0o synchronizes its clock to the NET clock and initiates 0 registration. If SYNC is not received within a dwell time, the device hops to the next frequency in sequence.

This continues until SYNC is attained, or until 15 hops have passed without receiving SYNC, after which the acquisition sequence is restarted.

-81- If coarse acquisition is not cbtained within 6.j seconds, the device selects another frequency and repeats the process beginning with step 2.

Camping provides a worst case acquisition performance that is perceptibly slow to the human user of a portable device. The preferred approach has the receiver scan all potential frequencies in ascending order, at 125 psec increments. When the highest frequency is reached, the search begins again at the lowest frequency. The 125is sampling rate is much faster than th- 250 psec channel switching time specification cf the RF modem. This is possible because the overall .2 switching time specification applies to worst cdse 15 frequency switching intervals, from the highest to the lowest operating frequency. By switching a single channel at a time, switching may be maintained over frequency intervals very near a synthesizer phase detectors' phase lock range, allowing nearly 20 instantaneous frequency switching. The change from highest to lowest frequency at the end of the scan requires the standard 250 Asec.

The 125 jsec monitoring interval allows 85;s to ascertain if receive clock has been detected prior to

S

25 switching to the next frequency. The monitoring interval

S

should be selected to be non-periodic with respect to the access interval. For example, the 125 usec interval allows the entire hopping sequence to be scanned 2(n 1) times in a 20 ms access interval.

-82- If clock is recovered at any frequency, the receiver remains on frequency for a Reservation Opportunity and initiates channel access through the procedure described above. The scanning approach is less deterministic in terms of acquisition probability than camping, but the search time required for 99.5% acquisiticn probability is about 80 Access Intervals, or three times faster than that for camping.

A hybrid approach that scans only three or four consecutive frequencies incorporates the deterministic aspects of camping with some of the improved performance of the scanning algorithm. For scanning over a small number of frequencies an up/down scan is preferred, i.e., 1,2,3,2,1,2, since all frequency changes can be 15 accomplished at the faster switching rate. T';e end frequencies are visited less often than those in the center. The number of frequencies used, 3 or 4, is *0 P selected so that all can be scanned during the preamble 00 0.e •duration of a minimum length transmission.

All devices are required to hae unique 48 bit 0global addresses. Loc 1 16 bit addresses will be assigned for reduced overhead in communications. Local S 0 addresses will not be assigned to devices whose global 0* addresses are not on an authentication list maintained in each access point and routinely updated over the infrastructure.

Once a device has attained synchronizaticn, it must register with the control point to be connected with the -83- NET. It initiates this by sending a Request for Poll indicating a registration request, and ircluding its global address. The control point will register the device, and provide a short. Network Address as an outbound message. The Control point will generate the short address if it is a single NET, or exchange the global address for a short Network Address with a Network Address Server if the NET is part of a larger infrastructured network of a premises LAN.

Once a device is synchronized to a NET, it must periodically update its local clock to the NET clock b, communicated in the SYNC message. The SYNC message contains a character designated as the SYNC character that transfers the NET clock synchronization. This may 15 De the beginning or ending FLAG in the SYNC message, or a specific character within the message.

The maximum expected frequency error between NET and device local clocks ic 100 parts per million. To c* maintain a 50 ps maximum clock error, the local device 20 clock must be re-synchronized at 500 ms intervals. At o ms per access interval, a non-sleeping device has up to 26 SYNC opportunities within that period in which to resynchronize and maintain required accuracy.

As mentioned above, it is desirable that battery powered devices have the capability to sleep, or power off, for extended periods of time to conserve power. The term sleeping terminal in this instance may refer to a device that powers down its radio communication hardware -84to save power while maintainina other functions in an operational state, or a device that power manages those functions as well. In the power managed state, the device must maintain its hop clock so that full acquisition is not required every time power management is invoked.

Devices that must sleep to manage their power consumption use Priority SYNC Messages to maintain synchronization. Priority SYNC Messages occur every three Access Intervals. In times of low NET activity, non-priority SYNC messages are omitted. By coordinating power management with Priority SYNC Messages, power managed devices can be guaranteed to wake up for Access Intervals where SYNCs will be present, even if the NET 15 activity is low during the sleep period.

A sleeping device with no transmission require.nents may sleep for eight 20 ms access intervals, and wake only for the SYNC and Reservation Poll at the beginning of the ninth Access Interval to monitor pending messages before -a returning to the sleep state, for a duty cycle of less e than This provides three opportunities to *wa* synchronize to the NET clock within a 540 ms window. A flow chart depicting the a device sleeping fcr several r* access intervals is shown in Fig. 17.

Devices may also sleep for longer periods of time, at the risk of losing fine synchronization. They may compensate by advancing their local clocks to account for the maximum timing uncertainty. For example. a terminal could sleep for 5 seconds witho'ut re-synchronizing by waking up 500 microseconds before it expects an Access Interval to begin, and successfully receive

SYNC

messages. This technique is valid for extended periods of time, up to the point where the maximum timing error apprcaches 50% of an Access Interval. A flow chart depicting the a device sleeping for several seconds is shown in Fig. 18.

A power managed device that requires communication during a sleep period may immediately wake and attempt access to the NET at the next available Reservation S" opportunity.

A device requiring communications may be able to register with one of several NETs operating in its 15 vicinity, with transmissions occurring on many S* frequencies simultaneously. A good strategy is to synchronize to a NET that provides an acceptable communication link, then monitor HELLO messages to determine other candidate NETs before attaching to a particular NET by registering with the control point 0" device.

As described above, a spontaneous wireless local area network or spontaneous LAN is one that is S OeS established for a limited time for a specific purpose, and which does not use the premises LAN to facilitate communications between devices or provide access to outside resources. Use of spontaneous LAN allows portable devices to share information, files, data, etc., -86in environments where communication via the premises LAN, is not economically justifiable or physically possible.

A spontaneous LAN capability also allows portable/mobile devices to have an equally portable network. Peripheral and vehicular LANs are examples of such spontaneous LANs.

Requirements for spontaneous LAN differ from an infrastrlictured premises LAN in several E;iqn if _4cant areas. The number of devices in a spontaneous LAN is likely to be smaller than the number that a single NET in a premises LAN must be capable of supporting. In e:se*: addition, coverage areas for spontancous LANs are S.0 typically smaller than coverage areas for an access point participating in the premises LAN. In a spontanejus LAkN, S~e *ecommunication often takes place over relatively short 00 15 distances, where devices are within line of siqht -f eacr.

other.

In an premises LAN, the majority of commun ications Se relklyt nvolve accessing communication network so 0e resources. For example, portable devices with limitei 0e0 20 processing capabilities, memory, and power supplies are :eee~able to access large databases or powerful computin9g engines connected to the AC power grid. Access points within the premises LAN are well suited to the role of Control Points for managing synchronization and medi,2 access within each NET.

In a spontaneous LAN, however, communications are limited to exchanges with spontaneous NET constituents.

Additionally, NET constituents may potentially leave at -87any time, making it difficult to assign control point responsibilities to a single device. A shared mechanism for synchronization and media access is preferable in most cases.

In a spontaneous LAN, battery power limitations may preclude assignment of a single device as a control point. The routine transmission of SYNC and access control messages places a significant power drain on a portable, battery powered device. Also, "he control point architecture dictates that transmissions intended for devices other than the control point be stored and forwarded to the destination device, further increasing battery drain, and reducing system throughput.

Moreover, the use of scheduled transmission in a O* premises LAN is likely to differ from use in a spontaneous LAN. For example, unlike the premises LAN, in the spontaneous LAN, applications such as massaging and two way voice communications may only occasionally oe used, whereas video transmission and telemetry exchange 20 may be prevalent.

To promote compatibility and integration with the premises LAN, operational differences required by multiple participating devices should be minimized. For example, selecting relatively close frequency bands for each LAN aids in the design of a multiple LAN transceiver, reducing circuitry, cost, power, weight and size while increasing reliability. Similarly, selecting communication protocols so that the spontaneous LAN -88protocol constitutes a subset or superset of premises LAN may enable a given device to more effectively communication in both LANs, while minimizing ooth the overall protocol complexity and potentially limited memory and processing power.

Use of frequency hopping is desirable in premises LAN because of its ability to mitigate the effects of interference and frequency selective fading. In the case of the latter, frequency hopping allows systems to be installed with less fade margin than single frequency *0 ystems with otherwise identical radio modem characteristics, providing improved coverage.

The potentially smaller coverage area requirement of *0 spontaneous LANs, however, allows single frequency operation to be considered for some applications, e.g., such as a peripheral LAN. Regulatory structures are in place in some countries to allow single frequency operation in the same bands as frequency hopping systems, providing that single frequency devices operate at 20 reduced power levels. The lower transmit power of single 5* frequency operation dna elimination of periodic channel switching are desirable methods of reducing battery drain. The choice of single frequency or frequency hopped operation is dictated by the coverage requirements of the network, and may be left as an option to device users.

As noted earlier, the basic Access Interval structure is suited to single frequency operation as well -89as to frequency hopping. SYNC messages in a single frequency system substitute a single frequency indication in the hopping sequence identifier field.

A spontaneous LAN comes into existence when two or more devices establish communications, and ceases when its population falls to less than two. Before a spontaneous LAN can be established, at least two devices must agree upon a set of operating parameters for the network. Such agreement may be pre-programmed else exchanged and acknowledged prior to establishing the spontaneous LAN. Once the spontaneous LAN is established, other devices coming into the network must be able to obtain the operating parameters and acquire access.

More specifically, to establish a spontaneous LAN, a computing device must first identify at least one other network device with which spontaneous LAN comjniunlcation S* is desired. To identify another network device, the computing device may play an active or parsive role. In 20 an active role, the computing device periodically oroadcasts a request to form spontaneous LAN with either ,0 a specific network device or, more likely, with a specific type of network device. If a network device fitting the description of the request happens to be in range or happens into range and is available, it responds to the periodic requests to bind with the computing device, establishing the spontaneous LAN. Alternately, the network device may. take a passive role in establishing the spontaneous LAN. In a passive role, the computing device merely listens for a request to form a spontaneous LAN transmitted by the appropriate network device. Once such a network device comes into range, the computing device responds to bind with the network device, establishing the spontaneous

LAN.

The choice of whether a device should take a passive or active role is a matter of design choice. For example, in one embodiment where peripheral devices have access to AC power, the roaming computer terminals take a passie role, while the peripheral devices take a more S.active role. Similarly, in another embodiment where a vehicle terminal has access to a relatively larger battery source, an active role is taken when attempting to form a spontaneous LAN, a vehicular LN, with a hand-held computing device.

Binding, a process carried out pursuant to a binding protocol stored in each network device, may be a very simple process such as might exist when creating a 20 spontaneous LANs that operates on a single frequency channel. Under such a scenario, a simple acknowledge handshake between the computing terminal and the other network device may be sufficient to establish a spontaneous LAN pursuant to commonly stored, preprogrammed operating parameters. However, more complex binding schemes may also be implemented so as to support correspondingly more complex spontaneous LAids as proves -91necessary. An example of a more ccmplex bind2na scheme is described below.

It is desirable in some large spontaneous LANs for one device to be designated as a fully functional control point, providing identical NET operation to a single NET in the premises LAN. Providing that all devices share ahopping table and encryption key, the designated device would initiate control point activities, and other devices would synchronize to the designated unit. A device with greater battery capacity, or one that can be temporarily connected to AC power is best suited to the 0dedicated control point function. This architecture is applicable to Client-Server applications (where the server assumes the control point function) or to other applications where a single device is tjhc2 predominaint source or destination cf communicat ions. A portable device used as a dedicated control point is ruquired to COhave additional programming and memory capacity, to manage reservation based media access, pending message list:, .:20 and scheduled service slot allocations.

.6.9.:In embodiments where communication requirements of Voo 00a spontaneous LAN are largely peer to peer, there may be :no overwhelming candidate for a dedicated Control Point.

Thus, in such cases, the Control Point function is either distributed among some or all the devices within the spontaneous LAN. In such scenarios, the interleaved Access interval approach used for wireless access points is employed. Initially, control point responsibilities -92are determined during the binding process. Users may designate or redesignate a Control Point device when 'several candidates are available.

Fc- spontaneous LANs, access intervals may be simplified to reduce power consumption, program storage and processing power requirements for portable devices used as control points. Control Point devices transmit SYNC, pending message lists, and Time Division Multiple Access slot reservations normally, but only use the i single slot reservation Poll (Idle Sense Multiple Access). The reservation poll contains a field indicating reduced control point functionality. This places other devices in a point-to-point communication mode, using the Implicit Idle Sense Algorithm. The probability factor p communicated in the reservation poll is used for the Implicit Idle Sense algorithm. Control point devices may use the deferred SYNC mechanism for light system loading, transmitting Priority SYNC every third Access Interval to further decrease their transmission requirements. Control point devices must monitor the reservation slot for messages addressed to them, but may sleep afterwards.

5 Request for Polls initiated under Implicit Idle Sense use point-to-point addressing, indicating the address of the destination device directly, rather than the control point device. This eliminates the need for the Control Point device to store and forward transmissions within the spontaneous LAN. The device -93detecting its address in a Request for Poll begins a session, after employing the Implicit Idle Sense algorithm, by Polling the source address identified in the Request for Poll. The terminating ACK and CLEAR messages contain an Explicit Idle Sense probability factor equal to that in the original reservation poll.

To allow for power managed devices, the Control Point device maintains a pending message list. Devices that have been unable to establish communication with a 10 sleeping device initiate a session with the Control Point device to register tie pending message. Upon becoming active, the sleeping device will initiate a Poll to the .device originating the pending message. The Control Point device will eliminate the pending message indication by aging, or by receipt of communication from the destination device clearing the pending nessage Control point devices are not required to store pending messages, only addresses.

As mentioned above, HELLO messages are broadcast to 9 20 indicate changes in NET parameters. HELLO messages may be omitted to simplify the Control Point function in Sspontaneous LANs.

Devices are assigned local addresses upon registration with the Control Point device. Devices may communicate an alias that identifies the device user to other users to the Control Point device where it is stcred in an address table. The address table may be obtained by other network constituents by querying the -94- Control Point device. A peripheral LAN is a type of spontaneous LAN which serves as a short range interconnect between a portable or mobile computing device (MCD) and peripheral devices.

Designers of portable products are constantly challenged with reducing size, weint,, and power consumption of these devices, while at the same time incrpasing their functionality and improvina user ergonomics. Functions that may be used infrequently, or S* 10 '.nich are too large to fit within the constraints of good 00 6 ergonomic design may be provided in peripheral devices, including printers, measurement and data acquisition 3 units, optical scanners, etc. When cabled or otherwise physicall> connected to a portable product, these peripherals often encumber the user, preventing freedom of movement or mobility. This becomes more problematic 0** when use of more than one peripheral is required.

A seconu consideration :or portable product design is commrunication docking. A communication dock is a 20 device that holsters or houses a portable unit, and provides for communication interconnection for such tasks as program downloading, data uploading, or communication with large printers, such as those used for printing full sized invoices in vehicular applications. Communication docking of a portable unit may also involve power supply sharing and/or charging.

The requirement for communication docking capability forces newer portable product designs to be mechanically compatible with older docking scnemes, or may require that new docks, or adapters, be developed for each new generation of portable device. Product specific docking approaches eliminate compatibility between devices manufactured by different suppliers. This has hindered development of uniform standards for Electronic Data Interchange between portabic devices and fixed computing systems.

Physical connection between a portable device with 0. a peripheral or corrmunication dock also hinders user efficiency. Peripheral devices are- generally attached S C with cable. If a peripheral is small enough to be carried or worn on a belt, the mobility of the user may be maintained. If a user must carry a hand-held portable device that is connected to a belt mounted peripheral the assembly cannot be set down while a task that r-quires movement to a location several feet away is undertaken

O

unless the portable device and peripheral e disconnected. Likewise, connection to peripherals too C 20 large to be portable requires the user to freq-gently l connect and disconnect the device and the peripneral.

Use of wireless peripheral LAN interconnection greatly simplifies the task of portable devices communicating with peripherals. _n doing so, wireless connectivity allows improved ergonomics in portatle pro.uct design, flexibility in interconnection to one cr more peripherals, freedom of movement over a radius of operaticn, forward and backward compatibility between portable units and peripherals, and potential communications among products manufactured by different vendors.

Constituents within a peripheral LAN generally number six or fewer devices. One roaming computing device and one or two peripherals comprise a typical configuration. Operating range is typically less than fifty feet.

Because the computing devices generally control the 10 operation of karipheral devices, in a peripheral LAN a 9 master/slave type protocol is appropriate. Moreover, roaming computing devices serving as master are well S:suited to the role of Control Points for managing synchronization and media access within each peripheral LAN. All peripheral communications are slaved to the master.

In a peripheral LAN, roaming mcbile or portable computing devices and wireless peripherals may all operate from bettery power. Operating cycles between 20 charging dictate use of power management techniques.

Although all participants in a periphera' LAN might also be configured to directly participate in the premises LAN, the trade-offs in cost, power usage and added complexity often times weighs against such configuration. Even so, participants within a peripheral LAN can be expected to function in a hierarchical manner, through a multiple participating device, with the premises LAN. Thus, the usa of a much simpler, lower- -97power transceiver and associated protocol may be used in the peripheral LAN.

As previously described, a roaming computing device serving as a master device may itself be simultaneously attempting to participate in other networks such as the premises or vehicular LANs. Considerable benefits arise if the radio and processing hardware that supports operation within the wireless network can also support such operation. For example, a device that is capable of frequency hopping is inherently suited to single frequency operation. If it can adjust transmitter power level and data rate to be compatibie .ith the requirements of the peripherals LAN, it can function in both systems. The major benefits of common transceiver hardware across LANs include smaller product size, improved ergonomics, and lower cost.

Specifically, in one embodiment, radio communication *0 on the premises LAN, as described herein, takes place e using radio transceivers capable of performing frequency- 20 hopping. To communicate on a peripheral LAN, such -transceivers could also utilize frequency-hopping at a lower power. However, such transceivers are relatively expensive in comparison to a lower power, narrow-band, single frequency transceivers. Because of the cost differential, it proves desirable to use the single frequency transceivers for all peripheral devices which will not participate in the premises LAN. Therefore, the more expensive, frequency-hopping transceivers which are -98fitted into roaming computing devices are further designed to stop hopping and lock into the frequency of the single frequency transceiver, allowing the establishment of peripheral LANs.

Instead of frequency hopping, the perioheral LAN may also use narrow-band, sincle frequency communication, further simplifying the radio transceiver design for commonality. In another embodiment of the peripheral LAN transceivers, operation using one of a plurality of 10 single frequency channels is provided. Thus, to overcome interference on one channel, the transceiver might select from the remaining of the plurality an alternate, single operating frequency with lesser channel interference. To accommodate the plurality of single frequency channels, 0e the peripheral LAN transceivers mAy either communicate an upcoming frequency change so that corresponding peripheral LAN participants can also change frequency, or .9 the transceivers may be configured to use frequency synthesis techniques to determine which of the plurality 20 a current transmission happens to be.

The Access Interval structure is also an appropriate choice for peripheral LAN operations. In one embodiment, to provide for simplicity and tighter integration, the Access Interval for the peripheral LAN is a subset of the Access Interval used in the premises LAN. HELLO messages, Implicit Idle Sense, Data Rate Switching, and scheduled services are not implemented. Peripheral devices normally sleep, activate their receivers for SYNC -99transmissions from the participating master device, and resume sleeping if no pending messages are indicated and they have no inbound transmission requirements. Access Intervals occur at regular intervals, allowing for Dower management. Access Intervals may be skipped if the master has other priority tasks to complete.

To initialize the peripheral LAN, a device desiring initialization, a master device, selects a single operating frequency by scanning the available frequencies for one with no activity. A typical master device might be a roaming computing device desiring access to a local peripheral. Default values for other parameters, including Access Interval duration, are contained within each participant's memory. Such parameters may be pre- 15 adjusted in each participant to yield specific performance characteristics in the peripheral

LAN.

Once a master device identifies a single frequency, 9 slaves, which are generally peripherals, are brought into 0* the peripheral LAN through a process called binding.

20 Binding is initiated by the master device by invoking a binding program contained therein. Slaves, such as peripherals, are generally programmed to enter a receptive state when idle. Thus, in one embodiment, the Smaster device accomplishes binding by transmitting Access Intervals of known duration sequentially on a series of four frequencies spread throughout the available frequency range. The specific frequencies and Access Interval durations used are stored as parameters in all -100potential participating devices. A 250 KBPS transfer rate is appropriate in some embodiments of the peripheral LAN, reflecting a balance between performance and complexity in peripheral devices.

A slave, a peripheral, responds to the binding attempts by the master device on a given frequency until the slave successfully receives and establishes communication with the master device. If they do not establish communication after four Access Intervals, the slave switches to the next frequency for four Access eg" Interval periods. Once communication is established, the woo slave registers with the master and obtains the master device's selected operating frequency and related Ccommunication parameters. When all slave devices have b-een bound, the master terminates the binding program and C. C normal operation at the selected single frequency may begin.

Referring to Fig. 15, in a hierarchical network, peripheral LAN masters use a secondary access interval be S 20 1501 that is synchronized to the Access Interval of a parent (premises) LAN control point. Peripheral

LA?,N

0*ee~eAccess Intervals occur less frequently than premises

LAN

Access Irtervals, every other or every third Priority SYNC Access Interval.

During the premises LAN Access Interval, the peripheral LAN master device monitors the premises

LAN

control point for SYNC 1503 reservation poll 1505 and exchanges inbound and outbound message according to the 101normal rules of the access protocol. The master switches to the peripheral LAN frequency, and transmits its own SYNC frame 1507 during the session period 1509 of its parent control point allowing communication with its peripherals. The peripheral LAN Access Interval is generally shorter than the premises LAN Access Interval, so that it does not extend heyond the premises LAN Access Interval boundary. At the end of the peripheral LAN Access Interval 1501, the master switches to the premises LAN frequency for the next SYNC 1503.

The secondary SYNC 1507 may only be transmitted if the peripheral LAN master is not busy communicating through the premises LAN. If a communicition session is I occurring, the master must defer SYNC, preventing 15 communication with its peripherals during that Access Interval. The marter must also defer SYNC if the current frequency in the LAN is prone to interference from the peripheral LAN frequency, they are the same frequency or adjacent frequencies. If two consecutive SYNCs Pre deferred, peripherals will activate their S" receivers continuously for a period of time, allowing the master to transmit during any Access Interval. This approach is also applicable when the master roams between frequency hopping NETs. Since NETs are not synchronized to one another, the devices in the peripheral LAN adjust Access Interval boundaries each time the master roams.

If peripherals do not detect SYNC within a time-out -102period, they may duty cycle their reception to conserve battery power.

Referring to Fig. 16, a Roaming Algorithm Flow Diagram illustrates how a roaming computing device will select a suitable access point. Roaming computing devices operating in the infrastructured network environment formed by the access points will routinely move between access point coverage areas. The roaming computing devices are able to disconnect fron their current access point communication link and reconnect a *.e communication link to a different access point, as necessitated by device roaming.

S* Access points transmit HELLO messages to devices in *o their coverage area. These HELLO messages communicate to 15 roaming computing devices the cost of connection through the access point, addresses cf neighboring access points, and the cost of connection through these neighboring a access points. This information allows roaming computing **4 devices to determine the lowest cost connection available 20 and to connect to the access point with the lowest cost.

In addition, access point HELLO message may include communication parameters of neighboring access points, A* such as frequency hopping sequences and indices, spread 4 spectrum spreading codes, or FM carrier channel frequencies. This information allows roaming computing devices to roam and change access point connections without going through a full acquisition phase of the new access point's parameters.

-103- Roaming computing devices initially select and register with an access point control point on the basis of link quality: signal strength and cost information transmitted within HELLO messages. A device will remain attached to a particular access point until the link quality degrades below an acceptable level; then it will attempt to determine if an alternative access point connection is available. The device initiates a roaming algorithm, using neighbors information from the most S 10 recent HELLO message to attempt connection with another candidate access point. If connection fails, another

**S

candidate from the neighbors list will be selected, and the process repeated. Once connection is made with an alternative access point, the device will monitor signal 0 strength and data errors to determine link quality. If link quality is acceptable, the device will continue monitoring until a HELLO message is received. If the cost is acceptable, it will register with the new access point, and the access point will update the spanning tree S 20 over the infrastructure. If link quality or cost is unacceptable, another candidate from the neighbors list is selected and the process repeated. This continues 0• until an acceptable connection is established. If one cannot be established, the device must return to the original access point connection or employ the initial acquisition algorithm.

Fig. 28a illustrates an embodiment of the hierarchical communication system according to the -104present invention communication is maintained in a warehouse environment. Specifically, a worker utilizes -a roaming computing device, a computer cerminal 3007, and a code reader 3009 to collect data such as identifying numbers or codes warehoused goods, such as the box 3010. As the numbers and codes are collected, they are forwarded through the network to a host computer 3011 for storage and cross-referencing. :n addition, the host computer 3011 may, for example, forward cross-referenced information relating to the collected numbers or codes back through the network for display on the terminal 3007 or for printing on a printer 3013. The host computer 3011 can be configured as a file server to perform such functions. Similarly, the collected information may be 15 printed from the computer terminal 3007 directly on the printer 3013. Other exemplary communication pathways supported include message exchanges between the computer terminal 3007 and other computer terminals (not shown) or the host computer 3011.

20 The host computer 3011 provides the termiial 3007 a with remote database storage, access and processing.

However, the terminal 3007 also provides for local processing within its architecture to minimize the need to access the remote host computer 3011. For example, the terminal 3007 may store a local database for local processing. Similarly, the terminal 3007 may run a variety of application programs which never, occasionally or often need access to the remote host computer 3011.

-105- Many of the devices found in the illustrative network are battery powered and therefore must conservatively utilize their radio transceivers. For example, the hand-held computer terminal 3007 receives its power from either an enclosed battery or a forklift battery (not shown) via a communication dock within the forklift 3014. Similarly, the code reader 3009 operates on portable battery power as may the printer 3013. The arrangement of the communication network, communication protocols used, and data rate and power level adjustments help to optimize battery conservation without substantially degrading network performance.

In the illustrated embodiment shown in Fig. 28a, the hierarchical communication system of the present 15 invention consists of a premises LAN covering a building r or group of buildings. The premises LAN in the illustrated embodiment includes a hard-wired backbone LAN 3019 and access points 3015 and 3017. A nost computer 3011 and any other non-mobile network device located in the vicinity of the backbone LAN 3019 can be directly attached to the backbone LAN 3019. However, mobile devices and remotely located devices must maintain connectivity to the backbone LAN 3019 through either a single access point such as the access point 3015, or through a multi-hop network of access points such as is illustrated by the access points 3015 and 3017. The access points 3015 and 3017 contain a relatively higher power transmitter, and provide coverage over the entire -106warehouse floor. Although a single access point may be sufficient, if the warehouse is too large or contains interfering physical barriers, the multi-hop plurality of access points 3017 may be desirable. Otherwise, the backbone LAN 3019 must be extended to connect all of the access points 3017 directly to provide sufficient radio coverage. Through the premises LAN, relatively stable, longer range wireless and hard-wired communication is maintained.

Because roaming computing devices, such as the handheld computer terminal 30C7, cannot be directly hardwired to the backbone LAN 3019, they are fitted with RF transceivers. To guarantee that such a network device can directly communicate on the premises LAN with at 15 least one of the access points 3015 and 3017, the ritted transceiver is selected to yield approximately the same transmission power as do the access points 3015 and 3017.

However, not all roaming network devices require a direct RF link to the access points 3015 and 3017, and some may 20 not require any link at all. Instead, with such devices, communication exchange is generally localized to a small area and, as such, only requires the use of relatively lower power, short range transceivers. The devices which *b participate in such localized, shorter range b oe communication form spontaneous LANs.

For example, the desire by a roaming terminal to access peripheral devices such as the printer 3013 and modem 3023, results in the roaming terminal establishing -107a peripheral LAN with the peripheral devices. Similarly, a peripheral LAN might be established when needed to maintain local communication between 3 code scanner 3009 and the terminal 3007. In an exemplary embodiment, the printer 3013 are located in a warehouse dock with the sole assignment of printing out forms based on the code information gathered from boxes delivered to the dock.

In particular, as soon as the code reader cathers information, it relays the information along a peripheral LAN to the terminal 3007. Upon receipt, the terminal 3007 comunicates via the premises LAN to the host o: computer 3011 to gather related information regarding a given box. Upon receipt of the related iniormation, the 09 terminal 3007 determines that printing is desired with 15 the printer 3013 located at the dock. When the forklift 3014 enters the vicinity of the dock, the terminal 300 7 establishes a peripheral LAN with the printer 3013 which begins printing the collected code information.

e.

To carry out the previous communication exchange, the printer 3013 and code reader 3009 are fitted with a lower power peripheral LAN transceivers for short range communication. The computer terminal 3007 transceiver is not only capable of peripheral LAN communication, but also with the capability of maintaining premises LAN communication. In an alternate exchange however, the code reader 3009 might be configured to participate on both LANs, so that the code reader 3009 participates in the premises LAN to request associated code information -108from the host computer 3011. In such a configuration, either the code reader 3009 or terminal 3007 could act as the control point of the peripheral LAN. Alternately, both coi..d share the task.

With capability to participate in the peripheral LAN only, the code reader 3009, or any other peripheral LAN participant, might still gain access to the premises LAN idirectly through the terminal 3007 acting as a relaying device. For example, to reach the host computer 3011, the code reader 3009 first transmits to the computer terminal 3007 via the peripheral LAN. Upon -eceipt, the o: computer terminal 3007 relays the t.-ansnmission to one of the access points 3015 and 3017 for forwardinq to the

S

1• "host 3011. Communication from the host 3011 to the code 15 reader 3009 is accomplished via the same pathway.

It is also possible for any two devices with no access to the premises LAN to communicate to each other.

For example, the modem 3023 could receive data and directly transmit it for printing to the printer 3013 via 20 a ptripheral LAN established between tle two. Similarly, the code reader 3009 might choose to directly communicate ~ode signals through a peripheral LAN to other network devices via the modem 3023.

S

In an alternate configuration, a peripheral LAN access point 3021 is provided which may be directly connected to the backbone LAN 3019 (as shown), acting as a direct access point to the backbone LAN 3019, or indirectly connected via the access points 3015 and 3017.

-109- The peripheral LAN access point 3n21 is positioned in the vicinity of other peripheral LAN devices and thereafter becomes a control point participant. Thus, peripheral LAN communication flowing to or from the premises LAN avoids high power radio transmissions altogether.

However, it can be appreciated that a stationary peripheral LAN access point may not alwa/s be an option when all of the p2ripherai LAN participants are mobile.

In such cases, a high power transmission to reach the premises LAN mzy be required.

Fig. 28b ilustrates other features ot the present 0 invention in the use of spontaneous LANs in association with a vehicle which illustrate the capabiiity of t automatically establishing a premises and a peripheral *f 1' LAN 1nen moving in and out of range to perform services and report on services rendered. In particular, like the forklift 3014 of Fig. 28a, a delivery truck 3033 provides Sa focal point for a spontaneous LAN utilization. Within the truck 3033, 1 storage terminal 3031 is docked so as to draw power from the truck 3033's battery supply.

Similarly, a computer terminal 3007 may either be docked o or ported. Because of greater battery access, the storage terminal 3031 need only be configured for multiple participation in the premises, peripheral and vehicular LANs and in a radio WAN, such as RAM Mobile Data, CDPD, MTEL, ARDIS, satellite communication, etc.

The storage terminal 3031, although also capable of -110premises and peripheral LAN participation, need only be configured for vehicular LAN participatic' 1 .i Prior to making a delivery, the truck enters a docking area for loading. As goods are loaded into the truick, the information regarding the goods is down-loaded into the storage terminal 3031 via the terminal 3007 or code reader 3009 (Fig. 28a) via the premises or peripheral LAN communications. This loading might also be accomplished automatically as the fork:lift 3014 comes IL0 into range of the delivery truck 3023, establishes or joins the peripheral LAN, and transmits the previously *@S~e.collected data as described above in relation to Fig 28a.

5Alternately, loading might also be accomplished via the premises

LAN.

*S515 As information regarding a good is .received and S. *stored, the storage terminal 3031 might_ also request further information regarding a-.y or all of the goods via the peripheral LAN's link to the host computer 3011 .through the premises LAN. More likely howe%,:r, the 5*20 storage terminal 3031 if appropriately configured would participate on the premises LAY4 to communicate directly :*5swith the host computer 3011 to retrieve such information.

0*The peripheral LAN access point 3021 if located on the dock could provide a direct low power peripheral

LAN

S.25 connection to the backbone LAN 3019 and to the host computer 301'-. Specifically, in one embodiment, the access point 3021 is located on the dock and comprises a low power ("short hop") radio cperating in a frequenmcy -111hopping mode over a 902-928Mj-z frequency band. However, the access point 3021 can instead be configured to communicate using, for example, infrared, UHF, 2.4GHz or 902MiHZ spread spectrum direct sequence frequencies.

Once fully loaded and prior to leaving the dock, the storage device 3031 may generate a printout of the information relating to the loaded goods via a peripheral LAN establisried with the printer 3013 on the dock. In addition, the information may be transmitted via the peripheral LAN modem 3023 to a given destination site.

As illustrated in Fig. 28c, once the storage terminal 3031 and hand-held terminal 3007 moves out of 999 9range of the premises and peripheral LANs, i.e. the 9 truck 3033 drives away from the dock, the vehicular LAN .915 can only gain access to the premises LAN via the more costly radio WAN communication. Thus, although the storage terminal 3031 might only be configured with relaying control point functionality, to minimize radio WAN communication, the storage terminal 3031 can be 2 0 configured to store relatively large amounts of B.*:information and to provide processing power. Thus, the 9:...:terminal 3007 can access such information and processing power without having to access device s on the premises 9 LAN via the radio WA-N.

Upon reaching the destination, the storage terminal 3031 may participate in any in range peripheral and premises LAN at the delivery site dock. Specifically, as specific goods are unloaded1, they are scanned for -112delivery verification, preventing delivery of unwanted goods. The driver is also informed it goods that should have been delivered are still in the truck. As this process takes place, a report can also be generated via a peripheral or premises LAN printrr at the destination dock for receipt signature. Similarly, the peripheral LAN modern on the destination dock can relay the delivery information back to the host computer 3011 for billing information or gather additional informat Lfl needed, avoiding use of the radio WA.N.

Ifthe truck 2-033 is used for service purposes, the truck 3033 leaves the dock in the morning with the *addresses and directions of the service 2stinations, technical manuals, and service notes which have been *.15 sekectively downloaded from the host computer 3011 via either the premises or peripheral L.AN to the storage terminal 3031 which may be configured with a hard drive and substantial processing power. Upon pulling out of *range, the storage terminal 3031 and the computer :20 terminal 3007 automatically form an independent detached :vehicular LAN. Alternately, the terminals 3007 and 3031 may have previously formed the vehicular LAN before leaving dock. In one embodiment, the 'vehicular

LAN

operates using frequency hopping protocol much the same as that of the premises LAN, with the storage terminal 3031 acting much like the premises LAN access points.

Thus, the radio transceiver circuitry for the premises LAN participation may also be used for the vehicular LAN -113and, as detailed above, a peripheral LAN. Similarly, if the radio WAN chosen has similar characteristics, it may to be incorporated into a single radio transceiver.

At each service address, the driver collects information using the terminal 3007 either as the data is collected, if within vehicular LAN transmission range of the storage terminal 3031, or as soon as the terminal 3007 comes within range. Any stored information within storage terminal 3031 may be requested via the vehicular LAN by the hand-held termina) 3007. Information not stored within the vehicular LAN may be communicated via a radio WAN as described above.

Referring again to Fig. 28b, upon returning to the dock, the storage terminal 3031, also referred to herein 15 as a vehicle terminal, joins in or establishes a peripheral LAN with the peripheral LAN devices on the dock, if necessary. Communication is also established via the premises LAN. Thereafter, the storage terminal 3031 automatically transfers the service information to 20 the host computer 3011 which uses the information for billing and in formulating service destinations for automatic downloading the next day.

Fig. 29a is a diagrammatic illustration of another embodiment using a peripheral LAN to supporting roaming 25 data collection by an operator according to the present invention. As an operator 3061 roams the warehouse floor he carries with him a peripheral LAN comprising the terminal 3007, code reader 3009 and a portable printer -114- 3058. The operator collects information regarding goods, such as the box 3010, with the code reader 3009 and the terminal 3007. If the power resources are equal, the terminal 3007 may be configured and designated to also participate in the premises

LAN.

Corresponding information to the code data must be retrieved from the host computer 3011. The collected code information and retrieved corresponding information can be displayed on the terminal 3007. After viewing for verification, the information can be printed on the printer 3058. Because of this data flow requirement, the computer terminal 3007 is selected as the peripheral

LAN

S*device which must also carry the responsibility o~f 9 communicating with the premises

LAN.

If durinq- collection, th~e operator decides to power 4*6:0,down the computer terminal 3007 because it is not needed, the peripheral LAN becomes detached from the premises LAN. Although it might be possible for the detached peripheral LAN to function, all communication with the host computer 3011 through the premises LAN is placed in a queue awaiting reattachment. As soon as the detached .peripheral LAN comes within range of an attar-hed LAN device, device attached to the ~.premises LAN, the queued communications are relayed to the host. It should be clear from this description that the peripheral LAN may roam in relation to a device attached to the premises IAN ("premises LAN device").

Similarly, the premises LAN device may roam in relaticn -115to the peripheral LAN. The roaming constitutes a relative positioning. Moreover, whenever a peripheral LAN and a master device move out of range of each other, the peripheral LAN may either poll for or scan for another master device for attachment. The master device may constitute a premises LAN device, yet need not be.

To avoid detachment when the terminal 3007 is powered down, the code reader 3009 may be designated as a backup to the terminal 3007 for performing the higher power communication with the premises LAN. As described ii. more detail below in reference to Fig. 33c regarding the idle sense protocol, whenever the code reader 3009 determines that the terminal 3007 has stopped providing access to the premises LAN, the code reader 3009 will

S.

15 take over the role if it is next in line to perform the 5 backup service. Thereafter, when the computer terminal 3007 is powered up, it monitors the peripheral

LAN

channel, requests and regains from the code reader 3009 the role of providing an interface with the premises LAN.

This, however, does not restrict the code reader 3009 from accessing the premises LAN although the reader 3009 e*555 may choose to use the computer terminal 3007 for power oee° °conservation reasons.

S

In addition, if the computer terminal 3007 reaches a predetermined low battery threshold level, the terminal 3007 will attempt to pass the burden of providing premises LAN access to other peripheral LAN backup devices. If no backup device exists in the current -116peripheral LAN, the computer terminal 3007 may refuse all high power transmissions to the premises LAN.

Alternatively, the computer terminal 3007 may either refuse predetermined select types of requests, or prompt the operator before performing any transmission to the premises LAN. However, the computer terminal 3007 may still listen to the communications from the premises LAN and inform peripheral LAN members of waiting messages.

Fig. 29b is a diagrammatic illustration of another embodiment of a peripheral LAN which supports roaming data collection by an operator according to the present invention. An operator is equipped with a peripheral LAN 3065 comprising a housing 3067, which incorporates a printer 3069 and a dock 3071, a roaming ccmputing

S

terminal 3073, and a code reader 3075. The operator may roam a warehouse floor or a shipping dock and collect and retrieve data using the peripheral LAN 3065 as discussed above with respect to Fig. 29a. In this embodiment, the 00 operator may elect to leave the housing 3067, and hence

O

20 the printer, in one area of the warehouse, or on the truck, and carry only the code reader 3075 and terminal 3073. In addition, the operator may also elect to dock the terminal 3073 in the dock 3071 and carry only the 00 of code reader 3075. In any event, the terminal is capable of communicating data to the printer 3069 via RF signals or via the dock 3071.

The housing 3067 may optionally include a cigarette lighter power input cable 3077 to power the printer 3069, -117and recharge the battery of the terminal 3073 via the dock 3071. The housing 3067 may also optionally include a wide area network radio to permit communication with a remote warehouse or station 3079. In addition, the housing 3067 may also be configured to include the functionality of the storage terminal 3031 discussed above with respect to Figs. 28b and 28c.

The peripheral LAN embodiments of Figs. 29a and 29b may, of course, function when detached from the premises LAN. This feature is particularly desirable in situations where attachment to the premises LAN may be more costly, such as, for example, during the remote pick-up or delivery of goods by a driver. In the

S

situation where a driver is picking up goods, the driver 15 may, for example, use the code reader and terminal to collect and/or enter information regarding the goods, such as their origin, destination, weight, etc. The terminal may then encode the information, and transmit it ,to the printer so that thu driver can label each box *0 appropriately with a bar or other type of code for later identification and routing of the goods.

*ewe** 8 Once information regarding a particular pick-up has been stored, either in the terminal or storage terminal, 0 the driver may download the stored data using the WAN .9 radio to the premises LAN host computer at the remote warehouse cr station 3079 so that the information may be used to pre-schedule further routing of the goods before the driver even arrives. Because WAN communication is -118costly, however, the information may instead be automatically transferred wirelessly to the premises LAN host once the driver comes into range of the premises LAN, as discussed above with respect to Fig. 28b.

Alternatively, or as a check to verify information previously transmitted to the premises LAN wirelessly, the information may be downloaded from the terminal to the premises LAN host via a docking system 3081 located at the warehouse or station 3079. The docking system 3081 may also be used to recharge the terminals 3073.

Once the host computer has the information regarding o. the goods picked up by the driver(s), the host can download the data via RF or the docking system 3081 to any number of terminals 3073 used by warehouse personnel 15 who unload the trucks. While unloading, these personnel 4* can, for example, use a terminal 3073 and a code reader 3075 to build containers for further distribution of the goods to various destinations. Specifically, as a container is unloaded, the label previously placed on the container by the driver is scanned by the code reader 3075, and destination information is displayed on the S* terminal 3073. -The box may then be taken to and loaded into the container headed for the same destination. Each container may also have a label which can be scanned to verify the destination of that particular container.

Fig. 30 is a block diagram illustrating the functionality of RF transceivers built in accordance with the present invention. Although preferably plugging into -119- PCMCIA -lots of the computer terminals and peripherals, the transceiver 3110 may also be built-in or externally attached via available serial, parallel or ethernet connectors for example. Although the transceivers used by potential peripheral LAN master devices may vary from those used by peripheral LAN slave devices (as detailed ow), they all contain the illustrated functional bi cks.

In particular, the transceiver 3110 contains a radio unit 3112 which attaches to an attached antenna 3113.

The radio unit 3112 used in peripheral LAN slave devices need only provide reliable low power transmissions, and are designed to conserve cost, weiaht and size.

t I. Potential peripheral LAN master devices not only require 15 the ability to communicate with peripheral LAN slave 4. e devices, but also require higher power radios to alLo communicate with the premises LAN. Thus, potential S w. peripheral LAN master devices and othe. non-peripheral LAN slave devices might contain two radio units 3112 or

S*

two transceivers 3110 one serving the premises LAN and 0j the other serving the peripheral LAN else only contain a single radio unit to service both networks.

In embodiments where cost and additional weight is not an issue, a dual radio unit configuration for potential peripheral LAN master devices may provide several advantages. For example, simultaneous transceiver operation is possible by choosing a different operating band for each radio. In such embodiments, a -120- 2.4GHz radio is included for premises LAN communication while a 27 MHz radio supports the peripheral LAN.

Peripheral LAN slave devices receive only the 27 MHz radio, while the non-potential peripheral LAN participants from the premises LAN are fitted with only the 2.4GHz radios. Potential peripheral LAN master devices receive both radios. The low power 27 MHz peripheral LAN radio is capable of reliably transferring information at a range of approximately 40 to 100 feet asynchronously at 19.2 KBPS. An additional benefit of using the 27 MHz frequency is that it is an unlicensed a frequency band. The 2.4 GHz radio provides sufficient power (up to 1 Watt) to communicate with other premises a* LAN devices. Another benefit of choosing 2.4 GHz or 27 **64 15 MHz bands is that neither require FCC licensing. Many different frequency choices could also be made such as the 900 MHz band, UHF, etc. Alternatively, infrared k* communication may be used in situations where line of *4 sight may be achieved between devices on the network.

20 In embodiments where cost and additional weight are at issue, a single radio unit configuration is used for potential peripheral LAN master devices. Specifically, in such embodiments, a dual mode 2.4 GHz radio supports e* both the peripheral LAN and premises LANs. In a peripheral LAN mode, the 2.4 GHz radio operates at a single frequency, low power level (sub-milliwatt) to support peripheral LAN communication at relatively close distances 20-30 feet). In a high power (up to 1 Watt) or -121main mode, the 2.4 GHz radio provides for frequencyhopping communication over relatively long distance communication connectivity with the premises LAN.

Although all network devices might be fitted with such a dual mode radio, only peripheral LAN master devices use both modes. Peripheral LAN slave devices would only use the low power mode while all other premises LAN devices would use only the high power mode. Because of this, to save cost, peripheral LAN slave devices are fitted with a single mode radio operating in the peripheral LAN mode.

Non-peripheral LAN participants are also fitted with a single mode (main mode) radio unit for cost savings.

see Connected between the radio unit 3112 and an

I

*o interface 3110, a microprocessor 3120 controls the 15 information flow between throug.i the transceiver 3110.

04 6 Specifically, the interface 3115 connects the transceiver 3110 to a selected computer terminal, a peripheral device or other network device. Many different interfaces 3115

S**

*ee* are used and the choice will depend upon the connection 0e j port of the device to which the transceiver 3110 will be attached. Virtually any type of interface 3110 could be adapted for use with the transceiver 3110 of the present invention. Common industry interface standards include RS-232, RS-422, RS-485, 10BASE2 Ethernet, 10BASE5 Ethernet, 10BASE-T Ethernet, fiber optics, IBM 4/16 Token Ring, V.11, V.24, V.35, Apple Localtalk and telephone interfaces. In addition, via the interface 3115, the microprocessor 3120 maintains a radio independent, -122interface protocol with tile attached network device, isolating the attached device from the variations in radios being used.

The microprocessor 3120 also controls the radio unit 3112 to accommodate communication with the either the premises LAN, the peripheral LAN, or both (for dual mode radios). Moreover, the same radio might also be used for vehicular LAN and radio WAN communication as described above. For example, a radio located in a vehicle or in a hand held terminal can be configured to commnunicate not only within a local network, but might also be capable of receiving paging messages.

More specifically, in a main mode transceiver, the microprocessor 3120 utilizes a premises LAN protocol to 15 communicate with the premises LAN. Similarly, in a peripheral LAN mode transceiver, the microprocessor 3120 operates pursuant to a peripheral LAN protocol to communicate in the peripheral LAN. In the dual mode transceiver, the microprocessor 3120 manages the use of 20 and potential conflicts between both the premises and peripheral LAN protocols. Detail regarding the premises Sand peripheral LAN protocols can be found in reference to geeoo.

Figs. 33-36 below.

In addition, as directed by the corresponding 0 25 communication protocol, the microprocessor 3120 controls the power consumption of the radio 3112, itself and the interface 3115 for power conservation. This is accomplished in two ways. First, the peripheral LAN and S-123premises protocols are designed to provide tor a low power mode or sleep mode during periods when no communication involving the subject transmitter is desired as described below in relation to Figs. 33-34.

Second, both protocols are designed to adapt in both data rate and transmission power based on power supply battery) parameters and range information as described in reference to Figs. 35-36.

In order to insure that the proper device is receiving the information transmitted, each device is assigned a unique address. Specifically, the transceiver 3110 can either have a unique address of its own or can S use the unique address of the device to which it is attached. The unique address of the transceiver can 15 either be one selected by the operator or system designer or one which is permanently assigned at the factory such as an IEEE address. The address 3121 of the particular transceiver 3110 is stored with the microprocessor 3120.

In the illustrated embodiments of Figs. 28-29b, the 20 peripheral LAN master device is shown as being either a peripheral LAN access point or a mobile or portable computer terminal. From a data flow viewpoint, in considering the fastest access through the network, such choices for the peripheral LAN master devices appear S. 25 optimal. However, any peripheral LAN device might be assigned the role of the master, even those that do not seem to provide an optimal data flow pathway but may provide for optical battery usage. For example, in the -124personal peripheral LAN of Fig. 29a, because of the support from the belt 3059, the printer might contain the greatest battery capacity of the personal peripheral LAN devices. As such, the printer might be designated the peripheral LAN master device and be fitted with either a dual mode radio or two radios as master devices require.

The printer, or other peripheral LAN slave devices, might also be fitted with such required radios to serve only as a peripheral LAN master backup. If the battery power on the actual peripheral LAN master, the hand-held terminal 300? (Fig. 29a, drops below a preset threshold, t the backup master takes over.

Fig. 31 is a drawing which illustrates an embodiment of the personal peripheral LAN shown in Fig. 29a which 15 designates a printer as the peripheral LAN master device.

Specifically, in a personal peripheral LAN 3165, a computer terminal 3i70 is strapped to the forearn -f the operator. A code reader 3171 straps to the back of the hand of the user and is triggered by pressing a button 20 3173 with the thumb. Because of their relatively low battery energy, the computer terminal 3170 and code reader 3171 are designated peripheral LAN slave devices S* e and each contain a peripheral LAN transceiver having a S* broadcast range of two meters or less. Because of its 25 greater battery energy, the printer 3172 contains a dual mode radio and is designated the peripheral LAN master device.

-125- Fig. 32 is a block diagrain illustrating a channel access algorithm used by peripheral LA.N slave devices according to the pres--t invention. At a block 318=,t when a slave device has a message to send, it waits for an idle sense message to be received from the peripheral LAN master device at a block 3183. When an idle sense message is received, the slave device executes a back-off prctocol at a block 3187 in an attempt to, avoid collisions with other slave devices waiting to transmit.

Basically, instead of permitting every slave device from repeatedly transm-tting immediately after an islie sense Soo&*:message is received, each waiting slave is required to *0first wait for a pseudo-random tine period before attempting a transmission. The pseudo-random back-off time period is generated and the waiting takes place at 000ea block 3187. At a block 3189, the channel is sensed to determine whether it is clear for transmission. If not, a branch is made back to the block 3183 to attempt a @0 transmission upon receipt of the next idle sense message.

*0020 If the channel is still clear, at a block 3191, a 00 0 relatively small "request to send" type packet is transmitted indicating the desire to send a message. if @000no responsive "certo sed tp message isreceived 00 00from the master device, the slave device assumes that a @02 collision occurred at a block 3193 and branches back to the block 3183 to try again. If the "clear to send" message is received, the slave device transmits the message at a block 3195.

-126- Several alternate channel access strategies have been developed for carrier sense multiple access (CSMA) systems and include 1-persistent, non-persistent and ppersistent. Such strategies or variations thereof could easily be adapted to work with the present invention.

Fig. 33a is a timing diagram of the protocol used according to one embodiment the present invention illustrating a typicr' communication exchange betwe-n a peripheral LAN master device having virtually unlimited power resources and a peripheral LAN slave device. Time line 3201 represents communication activity by the peripheral LAN master device while time line 3203 g represents the corresponding activity by the peripheral LAN slave device. The master periodically transmits an S* 15 idle sense message 3205 indicating that it is available *9 for communication or that it has data for transmission to a slave device. Because the master has virtually unlimited power resources, it "stays awake" for the entire time period 3207 between the idle sense messages 20 3205. In other words, the master does not enter a power conserving mode during the time periods 3207.

The slave device uses a binding protocol (discussed 9*9**g below with regard to Fig. 33c) to synchronize to the master device so that the slave may enter a power 25 conserving mode and still monitor the idle sense messages of the master to determine if the master requires servicing. For example, referring to Fig. 33a, the slave device monitors an idle sense message of the master -127during a time period 3209, determines that no servicing is required, and enters a power conserving mode during the time period 3211. The slave then activates during a time period 3213 to monitor the next idle sense message of the master. Again, the slave determines that no servicing is required and enters a power conserving mode during a time period 3215. When the slave activates again during a time period 3217 to monitor the next idle sense message, it determines from a "request to sen," type message from the master that the master has data for transmission to the slave. The slave responds by sending a "clear to send" type message during the time period 3 217 and stays activated in order to receive transmission of the data. The master is thus able to transmit the 15 data to the slave during a time period 3219. Once the data is received by the slave at the end of the time period 3221, the slave again enters a power conserving mode during a time period 3223 and activates again during the time period 3225 to monitor the next idle sense message.

Alternatively, the slave may have data for transfer to the master. If so, the slave indicates as such to the master by transmitting a message during the time period 3217 and then executes a backoff algorithm to determine 25 how long it must wait before transmitting the data. The slave determines from the backoff algorithm that it must wait the time period 3227 before transmitting the data during the time period 3221. The slave devices use the -128backoff algorithm in an attempt to avoid the collision of data with that from other slave devices which are also trying to communicate with the master. The backoff algorithm is discussed more fully above in reference to Fig. 32.

The idle sense messages of the master may also aid in scheduling communication between two slave devices.

For example, if a first slave device has data for transfer to a second slave device, the first slave sends a message to the master during the time period 3209 requesting communication with the second slave. The master then broadcasts the request during the next idle sense message. Because the second slave is monitoring S the idle sense message, the second slave receives the 15 request and stays activated at the end of the idle sense message in order to receive the communication. Likewise, because the first slave is also monitoring the idle sense message, it too receives the request and stays activated during the time period 3215 to send the communication.

20 Fig. 33b is a timing diagram of the protocol used 9 according to one embodiment of the present invention illustrating a typical communication exchange between a peripheral LAN.master having limited power resources and a peripheral LAN slave device. This exchange is similar 25 to that illustrated in Fig. 33a except that, because it has limited power resources, the master enters a power conserving mode. Before transmitting an idle sense message, the master listens to determine if the channel -129is idle. If the channel is idle, the master transmits an idle sense message 3205 and then waits a time period 3231 to determine if any devices desire communication. If no communication is desired, the master enters a power conserving mode during a time period 3233 before activating again to listen to the channel. If the channel is not idle, the master does not send the idle sense message and enters a power saving mode for a time period 3235 before activating again to listen to the channel.

Communication between the master and slave devices is the same as that discussed above in reference to Fic 33a except that, after sending or receiving data during the time period 3219, the master device enters a power 0* .9 conserving mode during the time period 3237.

Fig. 33c is also a timing diagram of one embodiment of the protocol used accoroing to the present invention which illustrates a scenario wherein the peripheral LAN' master device fails to service peripheral LAN slave devices. The master device periodically sends an idle 20 sense message 3205, waits a time period 3231, and enters a power conserving mode during a time period 3233 as discussed above in reference to Fig 33b. Similarly, the slave device monitors the idle sense messages during time periods 3209 and 3213 and enters a power conserving mode during time periods 3211 and 3215. For some reason, however, the master stops transmitting idle sense messages. Such a situation may occur, for example, if the master device is portable and is carried outside the -130range of the slave's radio. During a time period 3241, the slave unsuccessfully attempts to monitor an idle sense message. The slave then goes to sleep for a time period 3243 and activates to attempt to monitor a next idle sense message during a time period 3245, but is again unsuccessful.

The slave device thereafter initiates a binding protocol to attempt to regain synchronization with the master. While two time periods 3241 and 3245 are shown, the slave may initiate such a protocol after any number of unsuccessful attempts to locate an idle sense message.

With this protocol, the slave stays active for a time period 3247, which is equal to the time period from one idle sense message to the next, in an attempt to locate 15 a next idle sense message. If the slave is again unsuccessful, it may stay active until it locates an idle S* sense message from the master, or, ii power consumption is a concern, the slave may enter a power conserving mode at the end of the time period 3247 and activate at a 20 later time to monitor for an idle sense message.

In the event the master device remains outside the range of the slave devices in the peripheral LAN for a period long enough such that communication is hindered, one of the slave devices may take over the functionality S* 25 of the master device. Such a situation is useful when the slave devices need to communicate with each other in the absence of the master. Preferably, such a backup device has the ability to communicate with devices on the -131is idle. If the channel is idle, the master transmits an idle sense message 3205 and then waits a time period 3231 to determine if any devices desire communication. If no communication is desired, the master enters a power conserving mode during a time period 3233 before activating again to listen to the channel. If the channel is not idle, the master does not send the idle sense message and enters a power saving mode for a time period 3235 before activating again to listen to the channel.

0 Communication between the master and slave devices is the same as that discussed above in reference to Fic 33a except that, after sending or receiving data during the time period 3219, the master device enters a power conserving mode during the time period 3237.

5 Fig. 33c is also a timing diagram of one embodiment of the protocol used accoroing to the present invention a which illustrates a scenario wherein the peripheral LAN master device fails to service peripheral LAN slave devices. The master device periodically sends an idle !0 sense message 3205, waits a time period 3231, and enters f a power conserving mode during a time period 3233 as .discussed above in reference to Fig 33b. Similarly, the slave device monitors the idle sense messages during time p periods 3209 and 3213 and enters a power conserving mode 25 during time periods 3211 and 3215. For some reason, *4 however, the master stops transmitting idle sense messages. Such a situation may occur, for example, if the master device is portable and is carried outside the -130range of the slave's radio. During a time period 3241, the slave unsuccessfully attempts to monitor an idle sense message. The slave then goes to sleep for a time period 3243 and activates to attempt to monitor a next idle sense message during a time period 3245, but is again unsuccessful.

The slave device thereafter initiates a binding protocol to attempt to regain synchronization with the master. While two time periods 3241 and 3245 are shown, the slave may initiate such a protocol after any number of unsuccessful attempts to locate an idle sense message.

With this protocol, the slave stays active for a time period 3247, which is equal to the time period from one idle sense message to the next, in an attempt to locate 15 a next idle sense message. If the slave is again 0 unsuccessful, it may stay active until it locates an idle 0 sense message from the master, or, ii power consumption is a concern, the slave may enter a power conserving mode at the end of the time period 3247 and activate at a 00 020 later time to monitor for an idle sense message.

In the event the master device remains outside the range of the slave devices in the peripheral LAN for a period long enough such that communication is hindered,

S

one of the slave devices may take over the functionality

S*

25 of the master device. Such a situation is useful when Q* the slave devices need to communicate with each other in the absence of the master. Preferably, such a backup device has the ability to communicate with devices on the -131premises LAN. If the oricr'nal master returns, it listens to the channel to deter-mine idle sense messages from the backup, indicates to the backup that it has returned and then begins idle sense transmissions when it reestablishes dominance over the peripheral

LAN.

Fig. 34 is a timing diagram illustrating one embodiment of the peripheral LAN master device's servicing of both the high powered premises LAN and the low powered peripheral LAN subnetwork, with a single or plural. radio transceivers, in accordance with present invention. Block 3251 represents typical comnrunication activity of the master device. Line 3253 illustrates the master's communication with an access point on the premises LAN while line 3255 illustrates the master's commiunication with a slave device on the peripheral

LA-N.

6oeLines 3257 and 3259 illustrate corresponding communication by the a:ccess point and slave dFivice, respectively.

The access point periodically broadcasts

HELLO

a .6 20 messages 3261 indicating that it is available for a.0 communication. The master device monitors the HELLO messages during a time period 3263, and, upon determining that the base does not need servicing, enters 0 a power conserving mode during a time period 3265. The -25 master then activates for a time period to monitor the 06next HELLO message from the base. If the master has data to send to the base, it transmits the data during a time period 3271. Likewise, if the base has data to send to -132the master, the base transmits the data during a time period 3269. Once the data is received or sent by the master, it may again enter a power conserving mode.

While HELLO message protocol is discussed, a number of communication protocols may be used for communication between the base and the master device. As may be appreciated, the peripheral LAN master device acts as a slave to access points in the premises LAN.

Generally, the communication exchange between the master and the slave is similar to that described above in reference to Fig. 33b. Block 3273, however, illustrates a situation where the master encounters a communication conflict, it has data to send to or *0 receive from the slave on the peripheral LAN at the same o 15 time it will monitor the premises LAN for HELLO messages e 0 t 4* from the base. If the master has two radio transcei"ers, 00 e the master can service both networks. If, however, the master only has one radio transceiver, the master chooses to service one network based on network priority e** 20 considerations. For example, in block 3273. it may be desirable to service the slave because of the presence of data rather than monitor the premises LAN for HELLO messages from the base. On the other hand, in block 3275, it may be more desirable to monitor the premises 0 25 LAN for HELLO messages rather than transmit an idle sense message on the peripheral LAN.

Figs. 35 and 36 are block diagrams illustrating additional power saving features according to the present -133invention, wherein ranging and battery parameters are used to optimally select the appropriate data rate and power level for subsequent transmissions. Specifically, even though network devices sucb as the computer terminal 3007 in Figs. 28-29b have the capability of performing high power transmissions, because of battery power concerns, such devices are configured to utilize minimum transmission energy. Adjustments are made based on ranging information and on battery parameters.

Similarly, within the peripheral LAN, even though lower power transceivers are used, battery conservation issues also justify the use of such data rate and power adjustments. This process is described in more detail below in reference to Figs. 35 and 36.

15

M

ore specifically, Fig. 35 is a block diagram which illustrates a protocol 3301 used by a destination S. peripheral LAN device and a corresponding protocol 3303 used by a source peripheral LAN device to adjust the data rate and possibly the power level for future transmission 20 between the two devices. At a block 3311, upon receiving SSo* a transmission from a source device, the destination S* device identifies a range value at a block 3313. In a low cost embodiment, the range value is identified by considering the received signal strength indications 0 25 (RSSI) of the incoming transmission. Although

RSSI

circuitry might be placed in all peripheral LAN radios, the added expense may require that only peripheral

LAN

master devices receive the circuitry. This would mean -134that only peripheral LAN master devices would perform the function of the destination device. Other ranging techniques or signal quality assessments can also be used, such as measuring jitter in received signals, by adding additional functionality to the radios. Finally, after identifying the rangp value at the block 3313, the destination device subsequently transmits the range value to the slave device from which the transmission was received, at a block 3314.

Upon receipt of the range value from the destination device at a block 3321, the source peripheral LAN device evaluates its battery parameters to identify a subsequent data rate for transmission at a block 3323. If range value indicates that the destination periphe-al LAN *15 device is very near, the source peripheral LAN device selects a faster data rate. When the range value 0* 0 indicates a distant master, the source device selects a slower rate. In this way, even without adjusting the power level, the total energy dissipated can be controlled to utilize only that necessary to carry oiut the transmission. However, if constraints are placed on .the maximum or minimum data rates, the transmission power may also need to be modified. For example, to further minimize the complexity associated with a fully random 25 range of data rate values, a standard range and set of

S.

several data rates may be used. Under such a scenario, a transmission power adjustment might also need to supplement the data rate adjustment. Similarly, any -135adjustment of power must take into consideration maximum and minimum operable levels. Data rate adjustment may supplement such limitations. Any attempted modification of the power and data rate might take into consideration any available battery parameters such as those that might indicate a normal or current battery capacity, the drain on the battery under normal conditions and during transmission, or the fact that the battery is currently being charged. The latter parameter proves to be very significant in that when the battery is being charged, the peripheral LAN slave device has access to a much greater power source for transmission, which may justify the highest power transmission and possibly the slowest data rate under certain circumstances.

15 Finally, at a block 3325, an indication of the identified data rate is transmitted back to the destination device so that future transmissions may take place at the newly selected rate. The indication of data rate may be explicit in that a message is transmitted 20 designating the specific rate. Alternately, the data rate may be transferred implicitly in that the new rate is chose and used by the source, requiring the destination to adapt to the change. This might also bL done using a predefined header for synchronization.

25 In addition, at the block 3325, in another embodiment, along with the indication of the identified data rate, priority indications are also be communicated.

Whenever battery power is detected as being low, a radio -136transmits a higher priority indication, and each receiver thereafter treats the radio as having a higher protocol priority than other such radios that exhibit normal power supply energy. Thus, the remaining battery life is optimized. For example, in a non-polling network, the low power device might be directly polled periodically so to allow scheduled wake-ups and contention free access to a receiver. Similarly, in an alternate embodiment, priority indications not need to be sent. Instead, the low battery power device itself exercises protocol priority. For example, for channel access after detecting that the channel is clear at the end of an ongoing transmission, devices with normal energy levels are required to undergo a pseudo-random back-off before 15 attempting a transmission (to avoid collision). The low power device may either minimize the back-off period or ignore the back-off period completely. Thus, the low power device gains channel access easier than other normal power level devices. Other protocol priority 20 schemes may also be assigned by the receivers to the low power device (via the indication), else may be taken directly by the low power device.

Fig. 36 illustrates an alternate embodiment for carrying out the data rate and possibly power level adjustment. At a block 3351 upon binding and possibly periodically, the source peripheral LAN device sends an indication of its current battery parameters to the destination peripheral LAN device. This indication may -137be each of the parameters or may be an averaged indication of all of the parameters together. At a block 3355, upon receipt, the destination peripheral LAN device 355 stores the battery parameters (or indication).

Finally, at a block 3358, upon receiving a transmission from the source device, based on range determinations and the stored battery parameters, the destination terminal identifies the subsequent data rate (and possibly power level). Thereafter: the new data rate and power level are communicated to the source device either explicitly or implicitly for future transmissions.

Fig. 37 illustrates an exemplary block diagram of a radio unit 3501 capable of concurrent participation on multiple LAN's. To transmit, a control processor 3503 15 sends a digital data stream to a modulation encoding circuit 3505. The modulation encoding circuit 3505 encodes the data stream in preparation for modulation by frequency translation circuit 3507. The carrier .o frequency used to translate the data stream is provided 20 by a frequency generator circuit 3509. Thereafter, the mcdulated data stream is amplified by a transmitter amplifier circuit 3511 and then radiated via the one of a plurality of antennas 3513 that has been selected via an antenna switching circuit 2515. Together, the 25 modulation encoding circuitry 3505, translator 3507, amplifier 3511 and associated support circuitry constitute the transmitter circuitry.

Similarly, to receive data, the RF signal received by the selected one of the plurality of antennas 3513 is communicated to a receiver RF processing circuit 3517.

After performing a rather coarse frequency selection, the receiver RF processing circuit 3517 amplifies the RF signal received. The amplified received signal undergoes a frequency shift to an IF range via a frequency translation circuit 3519. The frequency translation circuit 3519 provides the center frequency for the frequency shift. Thereafter, a receiver signal processing circuit receives the IF signal, performs more exact channel filtering and demodulation, and forwards the received data to the control processor 2503, ending the process. Together, the receiver signal 1l processing 352-, translator 3517, receiver RF processing 3517 and associated support circuitry constitute the receiver circuitry.

The control processor 3503 operates pursuant to a set of software routines stored in memory 3522 which may *6 0 also store incoming and outgoing data. Specifically, the memory 3522 contains routines which define a series of protocols for concurrent communication on a nlurality of LANs. As part of such operation, the control processor 3503 provides for power savings via a power source 25 control circuit 3523, whenever the participating protocols permit, the control processor 3503 causes selective power down of the radio transceiver circuitry via a control bus 3525. Also via the bus 3525, the -139control processor sets the frequency of the frequency generator 3509 so as to select the appropriate band and channel of operation required by a correspondingly selected protocol. Similarly, the control processor 3503 selects the appropriate antenna (via tlhz antenna switching circuitry 3515) and channel filtering in preparation for operation on a selected

LAN.

Responding to the software routines stored in the memory 3522, the control processor 3503 selects the appropriate LANs to establish participation, detaches from those of the selected LANs in whicn participation is no longer needed, identifies from the selected LANs a current eae* e priority LAN in which to actively participate, maintains g~ 0a time-shared servicing of the participating LANs.

C15 Further detail regarding this process follows below.

*g*In one embodimen t, the control processor 3503 we 0 constitutes a typical microprocessor on an independent integrated circuiit. In another embodiment, the control processor 3503 comprises a combination of distributed C*20 processing circuitry which could be included in a single C. integrated circuit as is a typical micropro'cessor.

Ce 0 similarly, the memory 3522 could be any type of memory e~eee:unit(s) or device(s) capable of software storage.

The radio circuitry illuFs-.rated is designed %.vith the o frequercy nimble frequency generator 3509 so as to be C Ce capable of operation on a plurality of LANs/WANs.

Because each of the plurality may be allocated different frequency bands, more than one antenna may be desirable -140- (although a single antenna could be used, antenna bandwidth limitations might result in an unacceptable transmission-reception inefficiency). Thus, to select the appropriate configuration, the control processor 3503 first identifies the LAN/WAN on which to participate and selects the corresponding radio configuration parameters from the memory 3521. Thereafter, using the configuration parameters and pursuant to control routines stored in the memory 3522, the control processor 3503 sets the frequency of the generator 3509, selects the appropriate antenna via the antenna switching circuit 3515, and configures the receiver RF and signal .processing circuits 3517 and 3521 for the desired

LAN/WAN.

15 More particularly, the antenna switching circuit 3515 comprises a plurality of digitally controlled switches, each of which is associated with one of the plurality of antennas 3513 so as to permit selective connection by the control processor 3503 of any available 20 antenna to the transceiver circuitry.

Fig. 38 illustrates an exemplary functional layout Sof the frequency generator 3509 of Fig. 37 according to one embodiment of the present invention. Basically, the frequency generator 3509 responds to the control S* 25 processor 3503 by producing the translation frequency necessary for a selected LAN/WAN. The illustrated frequency generator comprises a voltage controlled oscillator (VCO) 3601. As is commonly known. for a VCO, -141the center frequency Fv tracks the input voltage.

However, because typical VCO's are subject to drift, the VCO is stabilized by connecting it in a phase locked loop to a narrowband reference, such as a crystal reference oscillator 3603. The oscillator 3603 outputs a signal -f a fixed or reference frequency F, to a divide-by-R circuit 3605, which divides as its name implies the reference frequency Fm by the known number R. A phase detector 3609 receives the divided-by-R output of the circuit 3609 and the feedback from the output of the VCO 3601 via a divide-by-N circuit 3607. Upon receipt, the phase detector 3609 compares the phase of the outputs from the circuits 3605 and 3607. Based on the comparison, a phase error signal is generated and-applied e 15 to a low-pass loop filter 3611. The output of the filter *e 3611 is applied to the input of the VCO 3601 causing the @o center frequency or the VCO 3601 to lock-in. Therefore, if the output of the VCO 3601 begins to drift out of phase of the reference frequency, the phase detector 3609 20 responds with a corrective output so as to adjust the center frequency of the VCO 3601 back in phase.

With the illustrated configuration, the center frequency of the VCO 3601 is a function of the reference frequency as follows: Fvc (FEF N) R S" Thus, to vary the center frequency of the VCO 3601 to correspond to a band of a selected LAN/WAN in which active participation is desired, the control processor -142- 3503 (Fig. 37) need only vary the variables and "N" and perhaps the frequency of the reference oscillator.

Because the output F. of the reference oscillator 3603 is quite stable, the phase lock loop as shown also keeps the output frequency Fvco of the VCO 3601 stable.

More specifically, although any other scheme might be implemented, the value R in the divide-by-R circuit 3605 is chosen so as to generate an output equal to the channel spacing of a desired LAN/WAN, while the value N is selected as a multiplying factor for stepping up the center frequency of the VCO 3601 to the actual frequency of a given channel. Moreover, the frequency of the reference oscillator is chosen so as to be divisible by values of R to yield the channel spacing frequencies of 9.

15 all potential LANs and WANs. For example, to participate *0 0 *on both MTEL Corporation's Two Way Paging WAN (operating at 900MHz with 25KHz and 50KHz channel spacings) and ARDIS Corporation's 800MHz specialized mobile radio (SMR) WAN (operating at 25KHz channel spacings centered at 20 multiples of 12.5KHz), a single reference frequency may chosen to be a whole multiple of 12.5KHz. Alternately, multiple reference frequencies may be chosen. Moreover, the value 'J is chosen to effectively multiply the output of the divide-by-R circuit 3605 to the base frequency of 25 a given channel in the selected WAN.

9.

For frequency hopping protocols, the value of R is chosen so as to yield the spacing between frequency hops.

Thus, as N is incremented, each hopping frequency can be -143selected. Randomizing the sequence of such values of N provides a hopping sequence for use by an access point as described above. Pluralities of hopping sequences (values of N) may be stored in the memory 3522 (Fig. 37) for operation on the premises LAN, for example.

In addition to the single port phase locked loop configuration for the frequency generator 3509, other configurations might also be implemented. Exemplary circuitry for such configurations can be found in U.S.

Patent Application Ser. No. 08/205,639 (Attorney Docket Nos. DN37139XXA; 10458US03), filed March 4, 1994 by Mahany et al., entitled "Method of and Apparatus For Controlling Modulation of Digital Signals in Frequencv- Modulated Transmissions". This application is 15 incor.orated herein in its entirety,.

Fig. 39 illustrates further detail of The receiver RF processing circuit 3517 of Fig. 37 according to one embodiment of the present invention. Specifically, a preselector 3651 receives an incoming RF data signal from 20 a selected one of the plurality of antennas 3513 (Fig.

37) via an input line 3653. The preselector 3651 provides a bank of passive filters 3657, such as ceramic

•*S*CO

or dielectric resonator filters, each of which provides a coarse filtering for one of the LAN/WAN frequencies to 25 which it is tuned. One of the outputs from the bank of passive filters 3657 is selected by the control processor 3503 via a switching circuit 3655 so as to monitor the desired one of the available LANs/WANs. Thereafter, the -144selected LAN/WAN RF signal is amplified by an RF amplifier 3659 before translation by the frequency translation circuit 3519 (Fig. 37).

Fig. 40 illustrates further detail of the receiver signal processing circuit 3521 of Fig. 37 according to one embodiment of the present invention. In particular, digitally controlled switching circuits 3701 and 3703 respond to the control processor 3503 by selecting an appropriate pathway for the translated IF data signal through one of a bank of IF filters 3705. Each IF filter is an analog crystal filter, although other types of filters such as a saw filter might be used. The IF 0 S0'. filters 3705 provide rather precise tuning to select the specific channel of a given LAN/WAN.

15 After passing through the switching circuit 3703, the filtered IF data signal is then amplified by an IF amplifier 3707. The amplified IF signal is then communicated to a demodulator 3709 for demodulation. The control processc retrieves the incoming demodulated data 20 signal for processing and potential storage in the memory 3522 (Fig. 37).

Fig. 41 illustrates further detail of the receiver signal processing circuit 3521 of Fig. 37 according to another embodiment of the present invention.

25 Specifically, the IF signal resulting from the transla:ion by the frequency translator circuitry 3519, enters the receiver signal processing circuit via an input 3751. Thereafter, the IF signal passes through an -145anti-aliasing filter 3753, and is amplified by a linear amplifier 3755. An IF oscillator 3757 supplies a reference signal f, for translation of the incoming IF signal at frequency translation circuits 3759 and 3761.

A phase shift circuit 3763 provides for a 90 degree shift of fp, if f, is considered a SINE wave, then the output of the circuit 3763 is the COSINE of Both the SINE and COSINE frequency translation pathways provide for channel selection of the incoming data signal. Thereafter the data signals are passed through corresponding low pass filters 3765 and 3767 in preparation for sampling by analog to digital

(A/D)

converters 3769 and 3771. Each A/D converter forwards the sampled data to a digital signal processor 3773 which 15 provides for further filtering and demodulation. The digital signal processor 3773 thereafter forwards the incoming data signal to the control processor 3503 (Fig.

37) via an output line 3775. Moreover, although the digital signal processor 3773 and the control processor *5 20 3507 are discrete components in the illustrated example, they may also be combined into a single integrated circuit.

Fig. 42 illustrates further detail of some of the S storage requirements of the memory 3522 of Fig. 37 25 according to one embodiment of the present invention. To control the radio, the control processor 3503 (Fig. 37) accesses the information in the memory 3522 needed for radio setup and operation on a plurality of LANs/WANs.

-146- Among other information, the memory 3522 stores: i) a plurality of software protocols, one for each LAN/WAN to be supported, which define how the radio is to participate on the corresponding LAN; and 2) an overriding control set of routines which govern the selection, use and interaction of the plurality of protocols for participation on desired LANs/WANs.

Specifically, in the memory unit 3522, among other information and routines, software routines relating to the media access control (MAC) sublayer of the communication protocol layers can be found. In general, a MAC sublayer provides detail regarding how communication generally flows tnrough a corresponding LAN or WAN. Specifically, the MAC sublayer handles functions 15 such as media access control, acknowledge, error Sdetection and retransmission. The MAC layer is fairly independent of the specific radio circuitry and channel characteristics of the LAN or WAN.

As illustrated, premises LAN, peripheral LAN, *0 20 vehicular LAN and WAN MAC routines 3811, 3813, 3815 and •3817 provide definition as to how the control processor 3503 (Fig. 37) should operate while actively participating on each LAN or WAN. Although only the several sets of MAC routines are shown, many other sets

S

25 might also be stored or down-loaded into the memory 3522.

Mcreover, the sets of MAC routines 3811-17 might also share a set of common routines 3819. In fact, the sets -147of MAC routines 3811-17 might be considered a subset of an overall MAC which shares the common MAC routines 3819.

Below the MAC layer in the communication hierarchy, hardware and channel reia~led software routines and parameters are necessary for radiz, control. For example, such routines govern the specific switching for channel filtering and antenna selection required by a given LAN or WAN. Similarly, these routines govern the control processor 3503's selection of parameters such as tar R and N for the frequency generator 3509 (Fig. 38) or the selective power-down (via the power source control circuitry 3503 Fig. 37) of portions or all of the radio circuitry whenever possible to conserve battery power. As illustrated, such routines and parameters are referred to as physical (PHY) layer control software S b:~,3821. Each of the sets of MAC routines 3811-17 and 3819 provide specific interaction with the PHY layer control software 3821.

A set of MAC select/service routines 3623 govern the management of the overall operation of the radio in the network. For example, if participation on the premises LAN is desired, the MAC select/service routines 3823 diret te cntrl prcesor 503(Fig. 37) to the common and premises MAC routines 3819 and 3811 respectively.

we *25 Thereafter, if concurrent participation with a peripheral LAN is desired, the select/service routines 3823 direct the control processor 3503 to enter a sleep mode (if available). The control processor 3503 refers to theF -148premises LAN MAC routines 3811, and follows the protocol necessary to establish sleep mode on the premises LAN.

Thereafter, the select/service routines 3823 directs the control processor 3503 to the peripheral LAN MAC routines 3813 to establish and begin servicing the peripheral LAN.

Whenever the peripheral LAN is no longer needed, the select/service routines 3823 direct a detachment from the peripheral LAN (if required) as specified in the peripheral LAN MAC routines 3813. Similarly, if during the servicing of the peripheral LAN a overriding need to service the premises LAN arises, the processor 3503 is directed to enter a sleep mode via the peripheral LAN MAC Sroutines 3813, and to return to servicing the premises

LAN.

15 Although not shown, additional protocol layers as well as incoming and outgoing data are also stored with the memory 3522, which, as previously articulated, may be a distributed plurality of sto:age devices.

Fig. 43 illustrates a software flow chart describing 20 the operation of the control processor 3503 (Fig. 37) in controlling the radio unit to participate on multiple LANs according to one embodiment of the present invention. Specifically, at a block 3901, the control processc- first determines whether the radio unit needs 0 25 to participate on an additional LAN (or WAN). If such additional participation is needed, at a block 3903, the radio unit may register sleep mode operation with other participating LANs if the protocols of those LANs so -149require and the radio unit has not already done so.

Next, at a block 3905, the control processor causes the radio unit to poll or scan to locate the desired additional LAN. If the additional LAN is located at a block 3907, participation of the radio unit on the additional LAN is established at a block 3909.

Tf additional participation is not needed at block 3901, or if the additional LAN has r.ot been located at block 3907, or once participation of the radio unit on the additional LAN has been established at block 3909, the control processor next determines at a block 3911 whether any of the participating LANs require servicing.

If any given participating LAN; requires servicing, at *a block 3913, the radio unit may be required by the 15 protocol of the given LAN to reestablish an active 00 participation status on that LAN, indicate to the given LAN that tne radio unit has ended the sleep mode.

Next, at a block 3915, the radio unit services the given LAN as needed or until the servicing of another LAN takes 00 priority over that of the given LAN. At a block 3917, the radio unit may then be required to register sleep mode operation with the given LAN if the LAN's protocol so requires.

.6 At that point, or if no participating LAN needs 25 servicing at block 3911, the control processor determines at a block 3919 whether the radio needs to detach from any given participating LAN. If so, the radio unit may implicitly detach at a block 3923 if the protocol of the -150- LAN from which the radio wishes to detach requires no action by the radio unit. However, at a block 3921, the radio unit may be required to establish active participation on the LAN in order to explicitly detach at block 3923. For example, such a situation may arise wheh aportable terminal desires to operate on a shorter range vehicular LAN and detaches from a premises LAN. The portable terminal may be required by the protocol of the premises LAN to establish active communication on the 10 premises LAN to permit the radio unit to inform th e premises LAN that it is detaching and can only be accessed through the vehicular LAN.

Once the radio unit is detached at block 3923, or if the radio unit does not need to detach from any participating LANs at block 3919, the control processor returns to block 3901 to again determine whether the radio unit needs to participate on an additional LAN, and repeats the process.

Fig. 44 is an alternate embodiment of the software 20 flow chart wherein the control processor participates on a master LAN and, when needed, on a slave LAN.

Specifically, at a block 3951, the contrul processor causes the radio unit to poll or scan in order to locate the master LAN. If the master LAN has not beer. located at a block 3953, polling or scanning for the master LAN continues. Once the master LAN is located, participation with the master is established at a block 3955. At a block 3957, the radio unit participates with the master -151- LAN until the need for the radio unit to participate on the slave LAN tak'ps precedence. When thatz condition occurs, the control processor determines at a block 3959 whether participation of the radio unit on the slave network is. established. if not, such participation is established at a block 3961. Next, at a block 3963, the radio unit services the slave LAN as needed or until the servicing of the master LAN takes priority. If the control processor determines at a block 3965 that 010 servicing of the slave LA-N has been completed, the radio 0unit detaches from the slave LAN at a block 396", and returns to block 3957 to continue participati-n on the master LAN.

O**e :However, if the control processor determines at block 3965 that servicing has not been, or may not be, completed, the radio unit does not detach from the slave LAN. In that case, befo)re returning to bloc;', 3957 to service the master LAN, the radio unit may be required by the protocol of the slave LAN to registet sleep mcde operation with the slave LAN at a block 3969.

0 In another embodiment, shown in Fig. 45, the overall 0 0 communication system of the present invention has been *0 adapted to service the envir-3nment found, for example, in a retail store. As illustrated, the premises of the retail stcre are configured with a communication network to provide for inventory control. Specifically, the communication network includes a backbone LAN 4501, a inventory computer 4511, and a plurality of cash -152registers located throughout the store, such as cash registers 4503 and 4505. As illustrated, the backbone LAN 4501 is a single wired link, such as Ethernet.

However, it may be comprised of multiple sections of wired links with or without wireless Link interconnects.

For example, ir another c-mbodiment, each cash reg ister 4503 and 4505 is communicatively interconnected with the inventory computer via an infrared link.

The inventory computer 4511, which can ranae from a S 10 personal to main frame computer, provides cen~tral control over the retail inventory by monitoring the inventory status. Thus, tine inventory computer 4511 must monitor both sales and delivery information r-'jarding inventoried goods. To monitor sales informa.tion, the cash registers 4503 and 4505 include code scanners, such as tethered code scanners 4507 r d 4509, which read codes on product labels or tags as goods are purchased. After receiving the code information read froL. the scanners '1507 and 4509, the cash registers 4503 and 4505 communicat2 sales information to the inv. ntory computer '4511 via the S backbone LAN 4501. To monitor delivery information, when h *te truck 4513 makes a delivery, the information regarding the goods delivered is communicated to the inventory computer 4511 via the access point 45)7. As illustrated, the access point 4517 acts as a direct access point to the backbone LAN 4501, ever., though a series of wireless hops might actually be required- -153- Upon receiving the sales information from the cash registers 4503 and 4505, the inventory computer 4511 automatically debits the inventory count of the goods sold. Similarly, upon receiving the delivery information, the inventory computer 4511 automatically credits the inventory count of the goods delivered. With both the sales and delivery information, the inventory computer 4511 accurately monitors the inventory of all goods stocked by the retail store. From the inventcry 1 0 information, the inventory computer 4511 generates purchase orders for subsequent delivery, automating the entire process.

T n particular, the inventory computer 4511 receives sales information from the cash registers 4503 and 4505 as detailed above. Whenever the restocking process is initiated, the inventrry computer 4511 checks the retail inventory for each item sold to determine if restockinq is needed. If restocking proves necessary, the invento:y computer 4511, evaluating recent sales history,

S

20 oetermines the quantity of the goods needed. From this information, an "inventory request" is automatically generated by the inventory computer 4511. Once verified (as modified if needed), the inventory request is automatically forwarded by the inventory computer 4511 to the warehouse ,519. This forwarding occurs via 2ither a telephone link using a modem 4571, or a WAN link using the backbone LAN 4501, access point 4517, and an antenna tower 4523.

154- At the remote warehouse 4519, the delivery truck 4513 is loaded pursuant to the inventory request received from the inventory computer 4511. After loading, the truck 4513 travels to the premises of the retail store.

When within range of the access point 4517, the radio terminal 4515 in the truck 4513 automatically gains access to the retail premises LAN via the access point 4517 (as detailed above), and communicates an anticipated delivery list (a "preliminary invoice"), responsive to 10 the inventory request, to the inventory computer 4511.

In response, dock workers can be notified to prepare for the arrival of the delivery truck 4513. In addition, any c. rerouting information can be communicated to the terminal 4515 in the delivery truck 4513. If a complete rerouting is indicated, the truck 4513 may be redirected without ever having reached the dock, While unloading the delivery truck 4513, codes are read from all goods as they are unloaded usirg portable code readers, which may be built into or otherwise communicatively attached to the radio terminal 4515. The codes read are compared with and debited against the Cpreliminary invoice as the goods are unloaded. This comparing and debiting occurs either solely within the terminal 4515 or jointly within the terminal 4515 and the inventory computer 4511. If the codes read do not correspond to goods on the inventory request, or if the codes read do correspond but are in excess of .what was required by the inventory request, the goods are -155rejected. Rejection, therefore, occurs prior to the actual unloading of the goods from the delivery truck 4513.

At the dock, the goods received from the delivery truck 4513 undergo a confirmation process by a dock worker who, using a radio terminal 4525 configured with a code reader, reads the codes from the goods on the dock to guarantee that the proper goods, those requested pursuant to the inventory request, were actually 10 unloaded. This extra step of confirmation can be eliminated, however, where the dock worker directly participates in the code reading during the unloading C process in the delivery truck 4513. Similarly, the code e C reading within the delivery truck 4513 could be eliminated in favor of the above described on-dock confirmation process, but, reloading of any wrongly C* unloaded goods would be required.

Upon confirmation of the delivery by the dock C: worker, a verified invoice is automatically generated by 20 the radio terminal 4515 and routed to the inventory computer 4511 for inventory and billing purposes. In C addition, the verified invoice is routed to the warehouse 4519. Such routing may occur as soon as the delivery truck returns to the warehouse 4519. However, to accommodate rerouting in situations where goods have been turned away at the retail store, the radio terminal 4515 communicates the final invoice immediately to the warehouse 4519. The warehouse 4519, upon receiving the -156final invoice, checks the final invoice with the list of goods loaded in the delivery truck 4513, and deterrines whether delivery of the remaining goods is possible. If so, the warehouse 4519 reroutes the truck 4513 to the next delivery site.

The communication of the final invoice and the rerouting information between the warehouse 4519 and the terminal 4515 may utilize a low cost communication pathway through the telephone link in the premises S. 10 network of the retail store. In particular, the pathway for such communication utilizes the access point 4517, backbone LAN 4501, inventory computer 4511 and modem 4521. Alternately, the communication pathway might also utilize the WAN directly from the radio terminal 4515 to the warehouse 4519 via the antenna tower 4523. Moreover, the antenna tower 4523 is merely representative of a 0.09 backbone network for the WAN. Depending on the specific WAN used, the tower 4523 may actually be comprised of a plurality of towers using microwave links to span the 20 distance between the retail premises and the warehouse 4519. Similarly, satellite relaying of the communications might also be used.

Figs. 46a-b illustrate a further embodiment of the communication system of the present invention which illustrate the use of access servers that support local processing and provide both data and program migration.

Specifically, as with the previous figures, Fig. 46a illustrates a wireless and hardwired communication -157network which uses a spanninr tree protocol to provide ubiquitous coverage throughout a premises.

For example, if any network device, an endpoint device such as a wireless, hand-held computer terminal 4601, easires to coLmunicate with another network device, a hard-wired computer 4603, a routing request is constructed which specifically identifies the destination device. After construction, the routing request is transmitted through a spanning 1 0 tree pathway to the destination device.

In particular, the terminal 460: formulates a routing request identifying the computer 4603. The routing request may also contain, for example, a message or data to be delivered or a request for data or program code. The terminal 46C1 tra-7-its the routing request downstream (toward the root of the spanning tree; to an access device 4605. The access device 4605 examines its spanning tree routing table entries, attempting to locate 'S an upstream path to the destination device identified by 20 the request. Because no entry exists, the access device 4605 transmits the routing request downstream to an access device 4607. After finding no routing table entry, the access device 4607 routes the request to a root access device 4609. Finding no routing table entry for the computer 4603, the access device 4609 transmits the routing request onto a wired LAN 4610. Using its routing table which has an ent y for the computer 4603, a root access device 4611 fields the routing request and -158transmits the request upstream to an access device 4613.

Likewise, the access device 4613, having an entry, sends the routing request to an access device 4617. Upon receipt, the access device 4617 forwards the routing request to the computer 4603.

When a network device, an end-point device for example, has a need for remotely stored program code program objects) or data data objects) such as a schematic diagram, delivery address or repair 0 S* 10 manual, the end-point device formulates a code or data request and sends it in a downstream spanning tree pathway. Unlike a routing request, data and code requests do not have a specific destination designated.

Instead, data/code requests (data requests and/or code request) only identify the specific data or code needed.

This is because the requesting device need not know the destination of the data or code needed, promoting dynamic, spanning tree migration as will become Sapparent below.

In addition, where possible, program code will be reduced to an interpretive form. Common libraries of program objects (in an object code form, i.e., executable form) are stored at each network terminal, computer or access server. Upon any request for an application program, for example, first, the sequence of calls to each program object is delivered along with a list of all program objects that are needed to fully execute the application program. Thereafter, if the -159specific underlying code for any of the delivered objects is not found locally, a renewed request for the executable code for those program objects is made. Upon delivery, the application program may be executed.

Moreover, the movement of the program application and other specific program objects are tracked and migrated as described above in relation to generic data.

For example, the terminal 4667 typically operates using an application program directed to an exemplary S* 10 installation and service industry. A driver of a vehicle 4666 enters the premises via a dock. Upon establishing 4* a link with the network, the terminal 4667 reports its status. In response, the terminal 4667 receives from the computer 4652 a command via the premises network to load a docking application.; After determining that it does not have the docking application stored locally, the terminal 4667 transmits a program code request specifying the application. Because of previous activity, for example, the access device 4659 (which receives the transmission) happens to have the program code stored locally. It fields the request, sending the list of program objects along with the "interpretive" program object sequence.. Upon receipt, the terminal 4667 might identify that all program object executable code is stored locally, and, therefore, begins to execute the application program. Otherwise, if certain program object executable is not locally stored, the terminal 4667 transmits a subsequent request. This time, the -160access device 4659 might not currently store the executable program object code. Thus, the access server 4659 routes the request downstream toward a device which does store the code. Once located, the code is delivered upstream to the terminal 4667 for execution.

Requested data or program code may reside in one or more of those of the access devices 4605, 4607, 4609, 4611, 4613, 4615, 4617, 4619 and 4621 which happen to be configured as access servers. Otherwise, the data or S, 10 code may be reside in one or more of the computers 4603, a 4 4621, 4623 or 4625, it they are configured as servers.

For example, assuming that the access device 4619 has been configured as an access server and happens to store data needed by the terminal 4601, the terminal 4601 would begin the process of retrieving the data by formulating a data request. As previously mentioned, the data request does not identify the access device 4619, but only identifies the needed data. After formulation, the terminal 4601 routes the request downstream to the 20 access device 4605. Upon receipt, the access device 4605 4, determines that it does not store the requested data, and fails to identify the requested data in a routing table entry. Thus, the access device 4605 forwards the data request to the access device 4607. As with device 4605, the access device 4607 cannot identify the requested data and routes the request to the access device 4609. Upon receipt, the access device 4609 consults its routing table and identifies an entry for the requested data.

-161- The entry lists the next device in an upstream path to the data, the access device 4619 is listed. Thus, the access device 4609 forwards the data request upstream to the access device 4619. The access device 4619 responds to the data request by: 1) locating the stored data; 2) formulating a routing request (containing the data) destined for the requesting device, the terminal 4601; and 3) sending the routing request downstream to the access device 4609. Using its routing table, the **Also 10 access device 4609 identifies the terminal 4601, anu u sends the routing request (with attached data) upstream J to the access device 4607. Likewise, the access device 4607 sends the routing request upstream to the access device 4605. Finally, the access device 4607 sends the routing request to the destination, the terminal 4601, completing the process. Program code program objects) may be similarly stored, requested and delivered.

1 Similarly, when remote processing is required, a network device formulates a processing request which 20 identifies the specific remote processing needed, yet need not identify a processing destination. After formulation, the processing request is transmitted 0 S downstream toward an access server or computer server capable of performing the requested processing. For example, the access device 4617 fields processing requests from the computer 4603. After determining that it cannot perform the processing, the access device 4617 consults its spanning tree routing table, yet finds no -162upstream entry for any network device capable of performing the processing. Thus, the access device 4617 routes the processing request downstream to the access server 4613. Although the access device 4613 has not been configured for such processing, the access device 4613 does find an entry identifying a first network device, the access device 4615, in an upstream pathway to a location where such processing is handled. The access device 4613 forwards the processing request to the access device 4615 which is configured as an access server to handle the processing. Thereafter, the requested processing is carried out by the access device 4615, with 0 any associated intercommunication with the computer 4603 needed via the same pathway using routing requests.

15 Thus, each spanning tree routing table not only o includes entries for all upstream network devices, each also includes entries for all upstream data, program code and processing resources. Moreover, each such entry only identifies the next network device through which 0* 20 forwarded requests are to made in the pathway to the request destination. Each spanning tree table also contains an entry designating a downstream route for use when no upstream entry can be located.

In the communication network of the present 25 invention, program code, data and local processing capabilities dynamically migrate through the network to optimize network performance. Specifically, each the access devices 4605, 4607, 4609, 4611, 4613, 4615, 4617, -163- 4619 and 4621 are configured as access servers. However, a specific data object in high demand is not initially stored in any of the access devices. Instead, the data object in high demand is originally stored on the computer 4623, configured as a server.

Upon encountering a first data request by the terminal 4601 for the data object in high demand, each of the intermediate access servers, the access devices 4605, 4607 and 4609, fail to identify the Jata object which results ir the sequential forwarding o' the data request to the computer 4623. However, each of the intermediate access servers record entries for the data in their @0 routing tables with a downstream destination.

Thereafter, each time that a network device, such as the *e 15 terminal 4601, requests the data object, the intermediate access servers which receive the request bump up a count stored in the routing table entry.

To make a determination of whether to migrate the

S.

data object or not, upon encountering a data request, each intermediate access server considers the. 1) associated count entry; 2) duration of time over which the count entry has accumulated; 3) cost of retrieving the data from the downstream source; 4) the size of the S S* data object; and 5) its own resource availability 0 remaining storage space).

For example, after receiving a high number of recent requests for the data object and having a relatively high cost in extracting the downstream object, the access -164device 4605 determines that migration of a copy of the data object into its own available storage could improve network performance. Thus, instead of sending the data request downstream to the access device 4607, the access device 4605 substitutes and forwards a migration request instead of the data request.

Upon receiving the migration request, the remaining intermediate access servers, the access devices 4607 and 4609 merely forward the migration request to the computer 4623. In response, the computer 4623 records the migration event, the data object migrated and the migration destination (the access device 4605~ for future updating control.

The computer 4623 also forwards a copy of the data 5 object to the access device 4C09 for relaying to the access device 4605 via the access device 4607. Upon receipt, the access device 4605 stores the data object locally, and forwards a further copy back to the requesting network device, the terminal 4601.

20 Thereafter, instead of relaying each data request for that data object downstream, the access device 4605 responds by sending a copy of the locally stored data object toward the requesting device. In other words, the access device 4505 has effectively intercepted a copy of S. 25 the data for local storage, and, thereafter, forwards a copy of the locally stored copy to service any incoming requests.

-165- In addition, upon forwarding the data object from the source, the computer 4623, to the destination, the term~inali 4601, the data object size and link cost associated with reaching a given intermediate access device is recorded. For example, if wired communication link between the computer 4623 and the access device 4609 is assigned a cost of 11", after fielding the data request, the computer 4623 constructs a data response which not only includes the requested data object, but also includes a link cost entry of 1"1 and an indication of the data object size. In turn, the access device 4609 identifies the cost to the access device 4607, for example a cost of the acces-- device adds the o the pending cost entry in the data *015 response, and forwards the re-ponse to the access device *:*4607 Similarly, the access device 4607 assesses a cost of '1311 for the communication link to the access device 4605, adids the "13"1 to -Lhe pending cost entry of and *0forwards the data response to the acco-Ls device 4605.

After assessing a cost for the link t~o the termirnal 4601, for example a cost of the data response is delivered to the terminal 4601. Thus, the terminal 4601 sees that to access the data again, it will most !likely result in I'll" units of communication cost. More"'ver, for example, the tei.minal 4607 considers the Cost Of n" when determining whether tc migrate the data object or not.

Similarly, when a migration of a data object occurs, all intermediate access devices record the cost of the -166upstream link to the copy of the data object.

Thereafter, upon receiving a data request for the data object, an intermediate access device can compare the cost of the upstream pathway to the copy with the downstream pathway to the original data object to choose the pathway with the lesser cost. A notification of deletion of a copy of a data object destined for a downstream source is also noted by each intermediate access devices, requiring deletions of the entries for the "copied then deleted" data object.

For example, if a locally stored copy of the data fails to be used for a period of time determined by the access device 4605 co be too lonq to justify local storage (in view of the communication link costs back to 15 the original source, the size of the data object, and potentially dwindling local resources), the access device 0@ 4605 deletes the locally stored copy of the data, and routes to the computer 462" an indication that the local copy of the data object has been deleted. Upon receiving S* 20 the indication for relaying, the intermediate access 0* devices 4607 and 4609, in turn, remove from their routing tables the entries to the recently deleted upstream copy.

Upon receipt of the indication, the computer 4623 records the deletion, completing the purging process.

Although data objects were used above to describe the migration process, program code (or program objects) are similarly migrated to and deleted from local-storage.

In addition, to prevent instability, a certain amount of -167hysteresis must be built in to prevent vacillating migration and purging decisions.

In assigning cost units to the various communication links, comparisons between factors such as actual monetary costs, bandwidths, delays, loading and power consumption are taken into consideration. Moreover, such costs are stored as sub-entries in the spanning tree routing tables.

Although only migration of a copy from a source to a single destination was previously described, if a data or program object proves to be in high enough demand, several, or even all, access devices in the network might store a copy. All that is required is that each access Sdevice experience a significant and sustained quantity of 15 requests for a common data object (or program code/object) to justify the storage of a local copy in view of communication link costs ,.nd available local resources.

Processing resources are similarly migrated and 0 purged. To service a processing request, an access m device must be configured not only with sufficient hardware resources but must also store the programming code and associated data necessary to perform the requested processing.

For example, if the terminal 4601 desires to search prior sales information but can neither store the information or the necessary search program routines because of limited local resources, the terminal 4601 -168formulates a processing request which it routes downstream to the access device 4605. In the illustrated embodiment, the access device 4609 is originally configured with the hardware and software necessary to perform the processing request. In particular, the access device 4609 uses bulk storage d.vices to store past sales data, and executes a search program in response to received processing requests.

Although the intermediate access server 4607 is configured with appropriate processing and storage resources, originally, it does not store the search pregram or the past sales data. Thus, while receiving repeated processing requests from the terminal 4601 via the intermediate access device 4605, the access device 0 15 4607 initially logs the request in its routing table and forwards the request downstream to the access device 4609 0* which fields, processes and responds to the requests.

Because the frequency of the requests, costs and available local resources, when not busy, the access device 4607 sends a migration inquiry downstream to the access device 4609. Upon receipt, the access device 4609 responds by sending an indication of the volume of the potential transfer upstream, to the access device 4607.

Based on the indication along with the aforementioned 25 other migration factors, the access device 4607 may or may not pursue the migration.

If migration is chosen, the access device 4607 assembles a migration request identifying the desired -169processing, and routes the request downstream to the access device 4609. In response, the access device 4609, records the migration (for future updating) and begins to transfer a copy of the program (or programming object(s)) and the past sales information to the access device 4607, preferably occurs during periods of low network traffic.

Although intermediate access devices between the source and destination of the processing migration are not shown in the exemplary illustration above, any intermediate access devices that do occur follow the same procedures previously set forth in reference to data object migration, recording and purging routing table entries to upstream and downstream proc ssing devices.

As may be appreciated in vie'. of the foregoing, in S* 15 many instances, migration does not always flow immediately to the access device nearest a requesting network device. Instead, for example, an access device which receives the same data or program code requests from a plurality of different terminals will perform 20 migration before any upstream access device unless upstream link costs are comparatively much higher.

*0 Fig. 46b is a diagram which further illustrates the migration and purging process. In particular, a premises network consists of computers 4651 and 4652 configured 9 25 as servers, a wired LAN 4653, and access devices 4655, 4657, 4659, 4661 and 4663 configured as access servers. A portable computer terminal 4664 participates in the premises network which exhibits migration and -170purging as described above in reference to Fig. 46a. In addition, a vehicular network is shown which consists of a mobile access server 4665 and a portable computer terminal 4667.

As illustrated, each of the access devices 4655 and 4659 are configured for long distance, wireless communication with the access server 4665 via a second higher power radio and associated antenna, e WAN, paging, cellular, etc. The corresponding first radio and associated antenna are used for relatively lower power premises network communication.

Because of the much higher cost associated w.ith the communication link between the access server 466 i and the access device 4659, the access servers 4665 is much more likely to enqage in the dynamic migration of data/code objects or processing resources than the other- access servers located within the premises network. Wit~ii a link cost assessed at "20" for example, the mobile access server 4665 rapidly decides to migrate, while slowly deciding to purge migrated data. The migration/purging :process used is the same as that described above in reference to the premises network of Fig. 46a.

In addition, because of the high link cost, the o mobile access server 4665 is also configured to provide anticipatory migration, and responds to direct migration commands from the terminal 4667 or other co-ntrolling network devices. Specifically, anticipatory migration may occur in two ways. First, if a driver is ore paring -171to leave the premises to service a specific appliance, for example, the schematic diagram of the appliance may be migrated to the mobile access server 4665 in anticipation of future use. This form of anticipatory migration may be directed from a controlling device downstream in the premises network, the computer 4652 which also stores the schematic diaqram, from the terminal 4667 upstream, or from the access server 4665 itself upon analysis of the work order.

A second form of anticipatory migration originates at the access server 4665 (although the resulting migration control could originate either up or *3 *downstream). The access server 4665 an.ticipates future migration needs through the storage and analysis of 15 previous requests for data/code objects or processing resources. For example, if the access server 4665 033.

determines that nearly every time the terminal 4667 request a given program code or program object, the terminal 4667 follows that request a short time 20 thereafter with further requests for specific data objects. In such circumstances, instead of repeatedly initiating, requesting and delivering portions of data over the communication the higher cost link, the requested and anticipated requests are all handled in one 25 communication session, saving money and time.

Similarly, the terminal 4667, through program design or through request monitoring, can also participate in anticipatory migration. For example, the terminal 4667 -172can be programmed to make all upcoming requests at one time, and often in advance of leaving the low power radio range of the premises network. The terminal 4667 can also be specifically programmed to issue direct migration and purging commands to the access server 4665, permitting further ccntrol of the migration process and system resources of the mobile access server 4665.

Moreover, the terminal 4667 may be configured to historically monitor all requests so as to anticipate subsequent requests in the manner described above in reference to the access server 4665.

In addition, the terminils 4664 and 4667 are configured to receive keyed, voice and pen input. Other types of input such as video or thumbprint image canture 15 might also be added. The terminals 4664 and 4667 can also be configured with code reading image capturing devices, or be configured to receive input from external code reading image capturing devices (via tethering or low power wireless links). Each terminal also provides 20 voice and LCD (liquid crystal display) output for the user. Thus, it can be appreciated that there are many types of data to be delivered to and from the terminals 4664 and 4667. The data may take on the forms of keyed or penned in command information, penned images or 25 signatures, captured images of 2-D codes, signatures, etc. and voice signals, for example.

Each type of data handled by the terminals 4664 and 4667 places specific requirements on the communication -173network. For example, when communicating voice signals, a communication channel or link providing real time voice delivery is often required. Dedicated bandwidth mav be reserved for such communications through the spanning tree network illustrated, or can be established via a cellular link with the access device 4655. Cellular radios may be built into the terminals 4664 and 4667 (via PCMCIA slots, for example) or via tethered cellular phones.

If post-processed, signature images require at most a delayed delivery of a plurality of such images over inexpensive and possibly sJower or less convenient 9 communication links. Relatively small packages of one way communication to the terminal 4667 may travel through 15 a lower cost paging net, -rk for delivery. They could also travel through the spanning tree network, cellular networks, or through other higher cost, two way WAN- Because programs cannot always anticipate all of the available communication channels through which the 20 different types of data may flow (availability which not only changes from one network installation to another, S but also changes within a given installation due to •terminal and device configurations and their locations within the network), the routing tables within each 9 25 network device subdivide routing information based on the 9* type of data to be forwarded.

For example, the access device 4665 begins receives a communication from the terminal 4667. The -174communication takes the form of a requested link for voice signal data destined for the computer 4651. In response, the access device 4665 consults its routing table, determines that voice data can take one of two pathways: through either a cellular radio or W.N route to the access device 4655. In response, the access device 4665 delivers the communication route options to the terminal 4667 for user and/or software consideration.

If the request is not aborted and the cellular route is selected, the access device 4665 establishes a cellular link with the access device 4655, and requests a voice link with the computer 4651. In response, the access point 4655 consults its routing table, and, for p.

voice data to the computer 4651, it identifies the need for dedicated wired bandwidth on the wired LAN 4653 directly with the computer 4651. In response, the access device 4655 places the request on the wired LAN 4653. In response, the computer 4651 communicates an acknowledge message which is routed through the access device 4655 to 20 the access device 4665. The access device 4665 delivers the acknowledge message to the terminal 4667. At that point, the terminal 4667 begins sending the voice data to the computer 4651 through the designated route.

If the cellular link to the access device 4655 is in 25 use or the end-to-end link otherwise proves unavailable, the access device 4665 reports the status, again offering the remaining communication route via the WAN. If selected, the access device 4-665 establishes the pathway -175to the access device 4659 via WAN communications. In turn, the pathway is established through the access device 4657, access device 4655 and the computer 4653.

With a returned acknowledge from the computer 4653, the terminal 4667 begins voice communication.

Similarly, communication pathway between any other two network devices, such as from the computer 4651 to the terminal 4664, can be established. For example, if the computer 4651 desires the user of the terminal 46C4 to obtain and compare a penned signature image for ccmparisor with an authenticated signature storea at the computer 4651, the computer 4651 first attempts to 00 communicate the request and image data to the terminal ,4664 via the premises network. If the terminal 4664 *6 15 happens to be out of range of the premises network, che **00 computer 4651 attempts to page the terminal 4664 witn the comparison request. In response, the terminal 4664 considers the data type via its routing table, identifies S0* the route(s) available, and offers the route options to 20 the user and/or program at the terminal 4664. If selected, the terminal 4664 establishes the selected *0 communication link for the delivery of the associated comparison image.

00 *0 S* Moreover, because of the high cost associated with 25 the communication link from the access device 4665 to the premises network, the access device 4665 stores several types of lower priority data until such time or data storage size justifies delivery. Such deliver may not -176occur until the vehicle returns to the premises network, to a dock at the premises.

In addition, requests for communication may also include specific limitations. For example, the need for voice data only in real time can be specified, and will result in no consideration by any intermediate network device of other pseudo-random real time link options.

Lowest cost delayed delivery can also be specified corresponding results. Requests with high priority specified, choose the astest communication link regardless of cost.

Moreover, the terminals 4664 and 4667 can be configured to operate running application software under the DOS, Windows or OS/2 operating system environments.

15 Communication between the terminal 4667 ind the access device 4665 occurs via an infrared link if the termiral 4667 is docked within the vehicle. Routing tables within the access device 4665 and terminal 4667 both contain dual entries for communication exchange 20 pathways. First, the infrared link is attempted, if available. Otherwise a lower power RF zommunication Stransmission is used. Although a wired docking msa.*: S* arrangement might be used instead of infrared, infrared is preferred inside the vehicle for ease of installation 25 and to minimize wire clutter. Such infrared installations also pre'ide support for communicating with printers, scanners and other peripheral devices within the vehicle, the vehicular LAN preferably operates -177via infrared except when communicating with a remotely located terminal 4667 or with other remotrly located network devices.

In another embodiment illustrated by Fig. 46b, service personnel use the vehicle 4666 for isitinq customer sites. At the site, the terminal 4667 is carried within the customer's premises. Ordinarily, communication with the premises network would take place via relatively low power radio transmissions between the terminal 4667 and the access Jevice 4665. However, communication can be achieved via a telephone iack link at the customer site, if: 1) the customer site blocks 0 Va such transmissions; 2) the transmission range is exceeded; or 3) link costs or channel speed so justify.

15 Once plugged into the telephone jack, the ter-nal 467automatically activates inactive routing table entries (by setting a flag therein) corresponding to possible telephone jack links. Theceafter, communication attempts to either the vehicul,. or premises LAN will offer routes 20 via the customer's telephone jack link.

Fig. 47a is a flow diagram which more specifically illustrates the functionality of the access servers of 0600** Figs. 46a-b in handling data, processing and routing requests. At a block 4701, an access server awaits 25 incoming communicationr which take the form of several types of previously mentioned requests sucn as data, object, processing, migration and routing requests. In addition at the bl.ck 4701, the access server awaits the -178need to perform migration evaluation and processing, a time out period to lapse which occurs once e-ery fifteen (15) minutes. This period may be modified (lengthened or shortened) as proves necessary depending on channel loading conditions.

Upon receiving a routing request as indicated at the event block 4703, the access server accesses its routing table, at a block 4705, in an attempt to identif' the destination of the routing request in an upstream path.

If th destination is identified, the access server forwards the routing request to the next network device in the upstream pati toward the destination, at a block I 4707. Otherwise, if the destination is not identified in the routing table at the block 4705, a- the block 4707 the access server transmits the routing request to the next network device in the downstream path. Thereafter, the access server returns to the block 4701 to await another event.

b* h 0 S* At the block 4701, upon receiving and loaging a 20 migration request, the access server vectors from an associated event block 4709 to determine whether it ,1009! stores the requested migration information the requested code or data or the program code and/or data associated with a processing resource migration request) S locally or not at a block 4711. If not, the access point branches to the block 4705 to identify the closest (or any) network device in the spanning tree pathway. For exaple, if the routing table carries .o entries for the -179migration informatica, the access server routes the migration request tn the next downstream network device.

otherwise, if the routing table rarries only an upstream or a downstream .arrry, the ac-cess server routes the request as specified by the routing table. However, if more than one entry exists for the requested migration information, the access server routes the migration request along the lowest cost spanning tree pathway (as indicated in the routing table).

However, if, at the block 4711, the access server determines that it stores the migration information locally, the access server: 1) retrieves the migration Aa information and records the migration event fc:- update **:control, at a block 4713; 2) accesses its ro'iting table to ident ify the forwardiihg pathway, at the block 4705; 3) forwards the retrieved migration information, at the block 4707; and 4) returns to the block 4701 to await another event.

After receiving and logging (counting the occurrence ~r20 of) a processing request at the block 4701, the access server branches via an event block 4715 to determine 0 whether the requested processing can be performed locally or not at a block 4717. If not, the access server forwards thse processing request at the block 4707 per routing table instruction at the block 4705. Afterwards, the access server returns to the await another event at the blc-P. 4701.

-180-- However, if the access server determines that it can perform the requested processing at the block 4717, the access server performs the processing at a block 4719, generates a response at a block 4721, routes the response back to the requesting network device at the block 4707 per routing table instruction at the block 4705, and, finally, returns to the block 4701 to await another event.

Upon receiving and logging a data or code request at the block 4701, the access server vectors via an event block 4723 to determine whether the requested data or code is stored locally at a block 472'. If so, the access server branches to a block 4727 to retrieve the *o data or code from storage. Thereafter, the data or code is forwarded at the block 4707 pe: routing table

S

instruction at the blocK 4705. Once forwarded, the access server branches to the block 4701 to await another C g event.

At the block 4725, if the access server determines that the requested data or code is not stored locally, the access server considers whether it should migrate the See.

data at a block 4729. The access server analyzes the C. overall link cost, the size of the requested data or see.., code, the frequency of such requests, available local 25 storage resources (some of which it may determine to .o recapture by purging other locally stored data, code or processing resources).

-181- Specifically, if sufficient local resources are (or can be made) available, the access server determines the weighted average frequency of the requests for that data or code. The frequency is then multiplied by a predetermined fraction of the overall link cost for retrieving the data or object to the access server from the current source. The resulting number is then compared to a migration threshold number, for example If, at the block 4729, the access server determines that zhe threshold number is greater than the resulting number, the access server, deciding not to migrate, branches to route the data/code request per routing table

C

*instruction at the blocks 4705 and 4707. Alternatively, if the access server determines that the threshold number is equal or less than the resulting number at the block 4729, the access server decides to migrate. Thus, at a block 4731, the access server creates and sends a migration request (instead of merely forwarding the 20 data/code request) and awaits delivery of the requested 0.

code or data. Upon receipt, at a block 4733, the access server stores the data or code. Thereafter, the data/code is retrieved at the block 4727 for muting to the requesting network device via the blocks 4705 and we 25 4707. Once routing is complete, the access server again 0 returns to the block 4701 to await another event.

Finally, upon receiving a time out event signifying the periodic need to perform migration evaluation and -182processing, the access server branches to execute migration procedures at a block 4737, as described in more detail below.

Fig. 47b is a flow diagram utilized by the access servers of Figs. 4'a-b to manage the migration of data and program code from a source st-rage and/or processing device toward an end-point device. More specifically, the exemplary flow diagram illustrates the migration and purging procedures represented by the block 4737 of Fig.

47a.

Upon encountering a time out event (occurring every minutes), an czcess server begins the illustrated procedure of Fig. 47b. At a block 4751, the access server retrieves a data/code entry from its routing table 9 15 for which it provides local storage. At a block 47-3, the current count recorded (indicating the number of 0* requests for that data/code entry during the current time out interval) is multiplied by two thirds and added to one third the value of the previously recorded weighted frequency. The access server records the result as the new weighted frequency in the routing table entry.

This weighting of frequency constitutes an "aging" of the data/code routing table entry.

At a block 4755, fifty percent of the overall cost of the link, from the access server to another source of the locally stored data/code, is multiplied by the newly recorded weighted frequency. The access server compares the results of the multiplication with a -183hysteresis threshold at a block 4757. The hysteresis threshold is also referred to herein as a purging threshold. In premises network locations, for example, the hysteresis threshold is set at five units below the migration threshoJd of the block 4729 in Fig. 47a.

However, the migration and hysteresis thresholds may need be modified in alternate network embodiments, such as may be found in vehicular network installations.

If the hysteresis threshold is exceeded, the access server determines that it should continue to store the data/code, and branches tc a block 4759 to determine whether there are any remaining entries for locally stored data/code which have not yet been considered for purging. Alternatively, if the hysteresis threshold is 15 not exceeded, the access server determines that the data/code item should be purged, and does so at a block 4761. Thereafter, the access server branches to the block 4759.

If, at the block 4759, other data/code items which *20 have not yet been considered for purging, the access server repeats the purging consideration of the blocks 4751 through 4759 until all locally stored data/code items have been considered. At that point, the access server branches to block 4763 to begin migration and S. 25 purging consideration of processing resources.

00. First, the access server retrieves a routing table entry relating to processing resources, supporting program code and any associated data. At a block 4765, the access server ages the entry, performs the aforementioned weighted frequency averaging. Thereafter, fifty percent of the overall link cost is multiplied with the new weighted frequency at a block 4767. If the entry indicates local storage of the processing resources at a block 4769, the access server compares the results with th hysteresis threshold at a block 4771. If above the hysteresis threshold, the access server continues to store the processing resources, branching to consider any remaining processing resource entries at a block 4775. Otherwise, the access server purges the stored resources at a block 4773 before considering any remaining entries at the block 4775.

S..oo: Alternately, if the routing table entry indicates 15 that the processing resources are not stored locally, at a block 4777, the access server determines whether it has been configured with the hardware necessary to perform S'9' the processing. If not, the access server branches to the block 4775 to consider process other entries.

Otherwise, at a block 4779, the access server compares .the migration threshold with the result, 50% of the link cost multiplied by the new weighted frequency. If the result does not exceed the migration threshold, the 9* g access server branches to the block 4775 to consider other entries. If the result exceeds the migration threshold, the access server formulates and routes a migration request for the processing resources, awaits 0 9o -185the responsive delivery and stores the resources locally at a block 4781, before branching to the block 4775.

At the block 4775, if the access server determines that other processing resource entries have not be considered for purging or migration, it repeatedly branches back to the block 4763 to carry out the consideration cycle until complete. Thereafter, the migration/purging procedure ends, and the access point returns to the block 4701 of Fig. 47a to await the occurrence of another event.

Fig. 48 is a schematic diagram of the access servers of Figs. 46a-b illustrating an exemplary circuit layout which supports the functionality described in reference to Figs. 47a-b. In particular, a typical access server, 15 an access server 4801, is configured with transceiver circuitry 4803 and associated antenna 4805 for participating in the premises, peripheral and/or wide area networks. In addition, another transceiver, a transceiver circuit 4807, and associated antenna 4809 might be added, for example, to support WAN or cellular *communications. Although not shown, interface circuitry for other wireless or wired communication links may he S. included in the access server configuration when needed.

Processing circuitry 4811 provides at least three processing functions for the access server by managing or 0* 0performing: 1) communication processing functionality; 2) migration and purging; and 3) local resource processing.

wherein incoming communications. Although in most -186embodiments, the processing circuitry 4811 comprises a single microprocessor, it may comprise several.

Moreover, if the processing circuitry 4811 is not configured to perform migration and local resource processing., the illustrated access device operates as an access point.

The processing circuitry 4811 utilizes a memory 4813 for short term and long term bulk storage. The memory 4813 comprises hard drive storage, dynamic RAM (ranc:om access memory), flash memory, and ROM (read only memory).

However, all other types of memory circuits or devices might alternately be used.

Specific hardware configurations needed to accommodate specialized processing requests are represented by a circuit/device block 4815. However, *e*O such hardware need not be present to service relatively basic processing requests. Additionally, access servers may either be battery powered although, if the network configuration permits, AC (alternating current) power is preferred.

Fig. 49a is a specific exemplary embodiment of an access server in a multi-hop communication network utilized for remote processing of 1-D (one dimensional) .or 2-D (two-dimensional) code information. In this embodiment, a code reader 4901 is used to capture and transmit code information for further processing, including decoding, by a remote access server in a premises LAN. Specifically, a user brings the code -187reader 4901, which preferably is a CCD (charged coupled device) type reader, into a reading relationship with a 2-D code 4903 located on a container 4905. Light reflected from the code 4903 is received by the code reader 4901 and directed onto the CCD located within the reader to "capture" the code image.

To enable the CCD to operate properly, however, it may first be necessary for the reader to focus the image on the CCD. Such focusing can, for example, be performed by conventional techniques known in the camera art. As another example, one or more spotter beams are presently used to ensure that the user is holding the reader the proper distance from the code to enable the CCD to urOr s S* properly capture the image.

15 Once captured, the code image may then be digitized S* within the reader to create a digital signal 4 representative of the code image, which is then transferred, via RF transmissions, to other network devices for further processing. Alternatively, the reader 4901 may transmit a modulated analog 3ignal representative of the code image to other network devices for further processing.

In any event, the code reader 4901, an end-point 0 device, forwards the code imace signal downstream in the premises LAN to the first access server in the network that has the capability of decoding the signal into the usable information represented by the code 4903. As discussed &bove, any one or all of the access devices -188- 4907-4913 may be an access server and contain the digital signal processing circuitry necessary to decode the code image signal. For example, the network may be designed such that the access device 4907 is an access server which performs decoding for all code readers, such as the code reader 4901, being used in a designated area. If, however, the access device 4907 is merely an access point, or is an access server but does not have decoding capability, tnen the access device 4907 relays the code image signal downstream.

More specifically, and as discussed more completely above, the code reader 4901 sends a processing request downstream to the access device 4907. If the access device 4907 is an access point, the processing request is 15 simply relayed downstream to the access device 4909. If the access device 4907 is an access server, it looks up S" in its table to determine whether it has the capability to perform the type of processing requested, i.e., decooing. If it does, the access device 4907 sends an acknowledge and the code reader 4901 forwards the code image signal to the access device 4907 for decoding.

Once decoded, the :nformation may be re-transmitted to the code reader 4901 for display on a screen (not shown) In addition, or alternatively, the access device 4907 may send a good read signal to the code reader 4901 to indicate to the user that the reading operation has resulted in a valid reading. The decoded information may -189also be transmitted to a host computer 4915 or other network device for further processing.

If the access device 4907 does not find decode capability listed in its table, it forwards the processing request downstream to access device 4909.

Likewise, if access device 4909 is an access point or an access server without decode capability, the processing request is forwarded downstream to the access device 4913. Once access device 4913 receives the processing request, it also examines its table to determine whether it, or any device upstream of it (such as, for example, access device 4911), has the capability to service the Sprocessing request. If it does locate such capaoility,

C

Sit sends an acknowledge upstream to the code reader 4901 15 which forwards the code image signal to the access device 4913 for decoding thereby or for routing to the upstream -i access device having that capability.

If the access device 4913 does not locate decode capability in its table, it forwards the processing 20 request to host computer 4915 for decoding thereby or so that the host computer 4915 can locate a device having the capability to service the processing request. Of course, as mentioned above, the network could be 0 configured such that each one of the access devices 4907- 25 4913 is an access server having the circuitry necessary for decoding.

While a CCD type code reader is preferred with respect to the embodiment of Fig 49a, other types of code -190readers, including laser scanners, are also contemplatedi.

Furthermore, while the above description places the decoding circuitry in a device external to the code reader 4901, the code reader 4901 may house such decoding circuitry and may transmit decoded data to external network devices for further processing. However, there are many advantages to placing the decoding circu4-try external to the code reader 4901. For example, because the code reader is a portable -evice and likel'; batterypowered, power conservation as well as reader size and .weight become important design considerations. By .placing the decoding circuitry in a device external to @005the reader 49 1 the re d ruses le spower a d my b smaller and lighter than if the decode circuitry is placed in the code reader 4901. Further, in an environment where numerous code readers are used, placing the decode circuitry in one or a few externa.1 devices 5 0**rather than all readers, which are often dropped by users, redaces the chances that che decoae circuitry will *:20 be damaged. in addition, suc a configuration rei'ices the amount of circuitry used and consequently results in lower reader manufacturing costs.

In add-Ition, the cow. reader 4929 is configured to collect signature, printed text and handwriting images for further processing. Although further processing can be performed on-board, within the reader 4929, in one embodiment it occurs within an access server.

-191- 0 Either way, such processing first involves the identification of the type of information contained within the image. If the user does not simplify the process by identifying the type of image captured, automatic identification is invoked. This occurs by first attempting to identify the image as a 2-D code. If this fails, the processing involves an attempt at character recoqnition to idcntify any printed text that might exist within the image. If no text is found, an .1o: 0 analysis is performed to determine whether the image is a handwritten signature. Finally, if all else fail-F, the me memo: image is generically classified as d picture. Several •0o90m examples of pictures include: images of bakery shelf space Jn a given store for subsequent collection and m e 0 evaluation of ones competition; '-0ges of broken equipment for transmission to remote exoerts for service advice; and images of meter displays fzr billing verification.

After identification, euch type of data receives vet oe

S

20 further processing. Decoded 2-D code informatis., is 0:00*0 forwarded and acknowledged. Handwritten signatures are .coz.ared with known authentic counterparts. Other types o£ images nay be associatively forwarded, stored, displayed and/or acknowledged.

Fig. 49b is an alternate embodiment of Fig. 49a wherein communication between the 2-D code reader and the access devicesr takes hie form of modulated infrared transmissions. Specifically, as discussed above with -192respect to Fig. 49a, a user uses a code reader 4917 to read a 2-D code 4939 on a container 4921. Tne uzer then points the code reader 4917 at an infrared transceiver 4923 of an access device 4925 and transmits a processing request to the access device 4925 using infrared transmissions. To facilitate receipt of th'e infrared transmissions by the infrared transceiver 4923, the reader may disperse its transmissions, say, for example, four inches over a distance of ten feet. Such dispersion i allows a iser to be less accurate in aiming the code .0 reader 4917 at the infrared transceiver 4923. The o00oo.

infrared transceiver 4923 may be, for example, a 00 phototransistor/photodiode pair.

00 0 As above, if the access device 4925 is simply an access point, the processing request is simpi% relayed downstream, via eitner RF or infrared transmissions, to a further accet device downstream. If the access device 0 4925 is an access ;erver, it looks up ii its table to determine whether it has the capability to perform the type of processing reauested. If "t does, the access device 4925 sends an acknowledge via infrared 0 transmissions to the code reader 4917 and the code reader 0o 4917 forwards the code image signal to the access device 4925 via infrared transmissions for decoding. The access device 4925 may then transmit the decoded information to the code reader 4917 for display on a screen and/or forward the decoded information to a host computer 4927 for further processing.

-193- If the access device 4925 does not ffind decode capability listed in its table, the access device 4925 forwards the processing request to one of the access tiavices 4924, 4926, or 4928 to locate such decoding capability similarly as discussed above with respect to Fig. 49a. When such a device is located, the code reader, via infrared transmissions, performs a batch forwarding of the stored image data to the access device 4925 for eventual decoding by one of the access devices 4924, 4926, or 4928 or by a host computer 4927 or another device in the premises LAN whichever it; the first *device located that has the decodina capability) In this embodiment, communication between access devices may be achieved using either RF or infrared transmissions.

Furthermore, a user may ch~oose to direc'tly com-municate withi any specific access de; ice in the network ,imply by pointing the code reader 4917 at that device and 00 transmittinq a processing request.

Fig. 49c is an alternate embodiment of Fig. 49a .:20 wherein indirect communication between the 2-D code foe0of: reader and the ac,.ess 3ervers takes place via hclstering .0or docking access servers. Specifically, as discussed 00 above with respect to Fig. 49a, a user uses a code reader 00 4929 to read a 2-D code on a container. The user then places the reader 4929 in a holster access device 4931.

The user may support the holster access device 4931 by a shoulder strap 4933 and belt 4935 to facilitate portability.

-194- In one embodiment, the holster access device 4931 may be configured to perform decoding so that when the code reader 4929 is placed inside the holster access device 4931, the code reader 4929 may transmit the code image data to the holster access device 4931 for immediate decoding thereby. Alternatively, if the holster access device 4931 does not house the necessary decoding circuitry, the holster access device 4931 transmits a processing request downstream to one of access devices 4937-4943 to locate such decoding o* capability similarly as discussed above with respect to Fig. 49a.

In a scenario where numerous codes 4945 are to be read successively by the code reader 4929, the code 15 reader 492; may store the read image data and perform a batch transmission to the holster access device 4931 for immediate decoding thereby if the holster access device 4931 is configured with decoding circuitry. In another embodiment where the holster access device 4931 is not so 20 configured, the code reader 4929 transmits a processing request to the holster access device 4931 via infrared transmissions. The holster access device 4931 in turn 99 forwards the processing request downstream via RF 9. transmission to one of the access devices 4937-4943 to locate such decoding capability similarly as discussed above with respect to Fig. 49a. When such a device is located, the code reader, via the holster access device 4931, performs a batch forwarding of the read image data -195for eventual decoding by one of the access devices 4927- 4943 or by a host computer 4947 or another device in the premises LAN whichever is the first device located that has the capability).

In an alternate embodiment, batch transmission of stored image data may be performed via a docking access server 4949. When a user has completed his code reading tasks, he docks the code reader 4929 in a bay 4951 of the docking access server 4949. Other users, when their tasks are completed, may similarly dock their code readers in other bays of the docking access server 4949.

S a In one embodiment, similarly as discussed above with respect to the holster access device 4931, nce a code reader is docked in the docking access server 4949, the Oe 15 code reader performs a batch transmission of its stored code image data to the docking access server 4949 for immediate decoding thereby if the docking access server 4949 is configured with decocing circuitry. In another embodiment where the docking access server 4949 is not so 20 configured, the code reader 4929 transmits a processing request to the docking access server 4949 via infrared transmissions. The docking access server 4949 in turn *9 °0 forwards the processing request downstream via RF transmission to one of the access devices 4937-4943 to locate such decoding capability similarly as discussed above with respect to Fig. 49a. When such a device is located, the code reader, via the docking access server 4949, performs a batch forwarding of the stored image 96data for eventual decoding by one of the access devices 4937-4943 or by a host computer 4947 or another device in the premises LAN whichever is the first device located that has the decoding capability).

In the embodiments of Figs. 49b or 49c wherein a number of codes are read and the captured image data is stored within the code reader for batch transmission at a later time, it may be desirable to configure the network such that decoding is performed first within the code reader. Specifically, when a user successively S reads a plurality of codes, a user can ensure that each **reading operation is successful or valid when the decoding is done immediately within the reader and the Suser is provided some sort of good read acknowledgement S by the reader. On the other hand, if the image data is simply stored for later decoding by an off-site device, the user cannot be sure that each reading operation resulted in a valid read. Such a situation may not be a problem, however, if the code and reader are highly 20 reliable or if simple information, such as a signature, is being read which may not require a validity determination.

SFig. 50 is a schematic diagram similar to that shown 0* in Fig. 48 which illustrates the circuit layout used in an access server of Fig. 49 to process the 2-D code information. Specifically, in an access point 5001, a processing circuitry 5003 manages 2-D code processing functionality as indicated by a block 5005. Although -197migration processing functionality is also present, in some embodiments such as those which use a single access server, the migration processing need not be present.

In a memory 5007, the access point 5001 also stores a database of known 2-D images in an image database 5009.

To further support 2-D code processing, digital signal processing circuitry 5011 has been added.

As configured, the signal processing circuitry 5011 assists the exact decoding of 2-D images, and may also be 10 used in the image comparison process of received 2-D images with the database 5009 of stored images.

Figs. 51a-b are flow diagrams illustrating the *0 oper-tion of the 2-D code processing access servers of Figs. 49-50. In Fig. 51a, when the access server receives image data via its LAN transceiver, it first attempts at a block 5101 to exactly identify the code information from the received code image data.

Specifically, the access server uses its code processing circuitry to perform an analysis of the received image 20 data using a decoding algorithm specifically designed for Sdecoding the type of code which was read. A number of 2- D code types exist, includirg, for example PDF-417, Maxicode, etc., which have specific corresponding decoding algorithms or rules.

After its analysis is complete, the access server next determines whether the exact identification was successful at a block 5103. Determining whether an identification was successful often depends on the type -198of code used. If enough redundancy is built into the code, then the loss of a number of bits of data resulting from, for example, a partially blurred image may not be fatal to a successful exact identification. If, on the other hand, the type of 2-D code being read is less "tolerant," then even the loss of a single bit might result in a failed exact identification.

In any event, if the exact identification is successful, at a block 5105, the access server sends the 10 identified code information to a predetermined destination for further processing, and acknowledges the successful identification. If the exact identification is not successful, however, the access point performs a fur'her analysis of the image data to attempt to identify the correspcnding code information.

At blocks 5107 and 5109, the access server compares the received image to stored images located in its image

*S

database and attempts to locate the closest or best match. Although grey scale considerations and image- .e 20 shifting correlation techniques are contemplated, in a relatively simple embodiment, such a comparison involve a process of scaling the received image to correspond to the stored images, then performing an "exclusive OR" of the received image with the stored images. More exact matches will yield an overall sum value nearer to zero.

After the access point completes its comparison and has identified the closest or best match between the received image data and the stored images, the access -199point then determines at a block 5111 if the overall value resulting from the best match comparison is above a predetermined accuracy threshold. Such an accuracy threshold may vary depending on, again, the type of code that was read, and the level of importance associated with a good read. If the overall value is below the predetermined threshold, the access server, as above, sends the identified code information (corresponding to the best match stored image) at a block 5105 to a 1 0 predetermined destination for further processing, and acknowledges the successful decode.

If the overall value of the best match comparison is above the predetermined threshold, then, at a block 5113, the access server forwards a bad read or retry message to the code reader to indicate to the user to re-read the code.

Fig. 51b is similar to Fig. 51a except that the comparison of the received image with stored images is performed before any exact identification is attempted.

9 20 Specifically, the access server first compares the received image to the stored images at a block 5115, identifies the closest match at a block 5117, and determines whether the overall value of the comparison is above a predetermined accuracy threshold at a block 5119.

If the overall value is below the threshold, the code information relating to the best match stored image is simply forwarded at a block 5121 to a pre-determined destination for further processing.

-200- If the overall value is below the threshold, then the access server attempts the exact identification and determines success at blocks 5123 and 5125. If such exact identification is successful, then the access device forwards the code information at block 5121. If it is not successful, the access device forwards a retry message to the code reader at block 5127.

Fig. 52 illustrates the structuring of 2-D code information so as to support a hierarchical recognition strategy as used by the access server of Figs. 49-50. In the image database of an access server, each known image are stored and hierarchically organized in sections. d g So*, Each section of image contains information relating to a wa specific category of information. For example, as shown, images may include a main category followed by further and further sub-categories. Thus, the image database stores all of the images in the main or first category at a top level in the hierarchy. Under each main category Si image, the image database stores only those sub-category 20 images which coexist with the main category image on known complete 2-D code images. Similarly, under each sub-category image, the image database only stores subsub-category images which coexist with the main category image and the sub-category image.

Fig. 53 is a diagram illustrating an exemplary 2-D code 5301 wherein the hierarchical structure of Fig. 52 is implemented. From left to right, top to bottom, the illustrated 2-D code provides image portions of -201categories separated by five bit line borders, sucn as a border 5303. As shown, the main category image represents "grocery". The sub-category represents "beans", and so on for the further sub-categories.

Using such a hierarchical categorization, the access server can more rapidly perform the process of image comparisons. For example, at a main category level in the hierarchy, a grocery image, an office supply image and general merchandize image might be the only three types of main category images known to the access point.

O* 0 If after comparing the received and the stored main Sa categorization images, no acceptable match is found, the 0 attempted comparison ends without ever having to compare the remainder of the potentially thousands of remaining images stored in the image database. Similarly, if a main level match is found with the stored office supply image, no comparison need be made with the plethora of remaining images under the grocery image main category.

g O i Further detail of the efficiency of such a ,e 20 hierarchical organization can be found below in reference go to Fig. 54. In addition, although the 2-D code illustrated in Fig. 53 is not necessarily a current 2-D 8 e 0 code standard, the principle of hierarchical organization can be utilized in current 2-D code standards to take advantage of the image comparison efficiencies involved.

Fig. 54 is a flow diagram illustrating the functionality of the access server of Figs. 49-50 in carrying out tha hierarchical recognition strategy of -202- Figs. 52 and 53. The access server begins the hierarchical image comparison process, and, at a block 5401, extracts from the received 2-D image a first subcategory image portion, the main category image indicating "groccry" for example. At a block 5403, the access server compares the extracted image with each of the main category images stored in the image database.

If the closest comparison fails to fall within an accuracy threshold at a block 5405, the access server indicates that the comparison has failed at a block 5407, and ends the process.

Otherwise, if the comparison is within the accuracy threshold at the block 5405, the comparison process continues with the access server checking to see if there 15 are any further sub-categories at a block 5409. Because other sub-categories exist, the access point branches to repeat the process beginning at the block 5401. This time, the access server extracts from the received image the image portion relating to the first sub-category (beans) for comparison at the block 5403 with only those first known sub-category images having "grocery" as the main category.

Again if no match within the threshold is found, the access point vectors to indicate failure at the block 5407, and terminates the process. However, if a subcategory match is found, the access point oranches to handle the sub-sub-category in a similar way. 'If, at the block 5409 after successfully repeating the comparison a -203number of times, the access point concludes that there are no further sub-categories to compare, the access point delivers the 2-D code information stored in the image database and associated with the matching stored image, and successfully ends the code identification process.

The known image database is supplemented by exact decoding as illustrated for example in Fig. 51b, wherein any successful exact decode is used to provide both categorized image and information portions for subsequent decoding through comparison. In addition, although the hierarchical structuring described herein offers many advantages, it need not be implemented to carry out the *e*q comparison process.

15 Moreover, it will be apparent to one skilled in the art having read the foregoing that various modifications and variations of this communication system according to S0 0 the present invention are possible and is intended to include all those which are covered by the appended claims.

O

r or

O

•D

-204-

Claims (1)

140. 14 14 14 14114 14 *1 14 0 1. A communication network operating between a premises and a vehicle, comprising: a data source located at a premises; a terminal device within a vehicle; a first communication link between said data source and said terminal device; a vehicular network comprising a portable computing device and said terminal device; said portable computing device and said terminal device communicating via a second communication link, said second communication link comprising a wireless communication link; and said terminal device being configured to store data delivered from said data source, and, upon communication from said portable computing device, selectively forwarding the stored data to said portable computing device. 2. The communication network of claim 1 wherein said terminal device monitors the flow of data to the portable computing device, and, based on such monitoring, said terminal device selectively migrates data into local storage. 3. The communication network of claim 2 wherein said terminal device monitors the flow of program code to the portable computing device, and, based on such monitoring, said terminal device selectively migrates program code into local storage. 4. The communication network of claim 2 wherein said terminal device monitors processing requests from the portable computing device, and, based on such monitoring, said terminal device selectively migrates programming resources into local storage. 14 14 14 1 1414 0*14401 14 14 1 141 14 14 L. 1 0 14 S {N:\LIBK]O 132:MXL 206 A communication network substantially as herein described with reference to any one of the embodiments of the invention shown in the accompanying drawings. DATED this Twenty-first Day of December 1998 Norand Corporation Patent Attorneys for the Applicant SPRUSON FERGUSON 0e 0 0 s -0 0 [NALIBKIO1 1 32:MXL