JPH0710069B2 - Packet communication system - Google Patents

Description

Description: TECHNICAL FIELD OF THE INVENTION The present invention relates to a packet communication network for controlling the transmission of communication packets according to the CSMA / CD method, and uses an index for evaluating the performance of the communication network from the distribution of packet transmission delay times. The present invention relates to a packet transmission delay measurement method that can be effectively obtained.

[Technical background of the invention] Recently, a plurality of information processing devices are connected to each other via a transmission path to form a local area network, and resources such as a storage device and an input / output device possessed by each of the above information processing devices. Distributed processing systems that share information and perform predetermined information processing are receiving attention. In this local area network, it is necessary that all of the information processing devices connected to the transmission line use the transmission line jointly, so that they do not interfere with each other's communication. . The CSMA / CD (Carrier Sense Multiple Access with Collision Detection) method has been widely used as one of the transmission path utilization procedures.

That is, the local area network is, for example, the first
As shown in the figure, a plurality of information processing devices 1a, 1b, to 1n are connected to a transmission line 3 via communication devices 2a, 2b to 2n called so-called stations. The transmission line 3 is composed of, for example, an optical fiber cable and a star coupler 4 that interconnects these optical fiber cables. The communication devices 2a, 2b to 2n are connected via the transmission line 3 as described above.
A data group represented as a bit string of several kilobits is mutually transmitted as a packet. In this case, each communication device
2a and 2b to 2n execute a series of procedure processing in the CSMA / CD system as shown in FIG. 2, for example, to control the transmission of the communication packet. In this packet communication control by the CSMA / CD method, carrier sense (CS) of the transmission line 3 is performed prior to the transmission of the communication packet, and when carrier sense is on, that is, some carrier signal exists on the transmission line 3. If so, the transmission of the communication packet is suspended and the transmission attempt is postponed. This postponement is commonly referred to as defer. When the carrier sense is off, transmission of the communication packet is started. At this time, a collision between the transmitted communication packet and a communication packet from another station is detected. This process is called collision detection (CD), and when no collision occurs, the transmission of the communication packet is continued as it is. Then, when the collision is detected, the transmission of the packet is stopped and the trial for the packet transmission is postponed. This postponement is commonly referred to as packoff. Since a so-called non-persistent system, in which there is almost no difference between the defer and the backoff, is adopted, the defer and the backoff will be generically described as a backoff in a broad sense.

In this way, each communication device 2a, 2b ~ 2n performs carrier sense and collision detection for a given communication packet to disperse multiple access to the transmission path for sending out the communication packet. Have control. Then, the communication packet transmitted via the transmission line 3 is transmitted to each communication device.
The communication destination is checked in 2a, 2b to 2n and taken into the corresponding communication device.

By the way, the back-off is such that when carrier sense is on when a packet is to be transmitted, or when a collision is detected after starting transmission, the transmission of the communication packet is stopped and the communication is stopped. Although the packet re-sending attempt is postponed for a randomly generated time, the control performance of the CSMA / CD system is greatly affected by how the backoff time is determined. By the way, if the setting range of the back-off time is set too small, collisions frequently occur and the effective use rate (throughput) of the transmission line 3 becomes low. On the contrary, if the back-off time is set to a large value, the ratio of the time when the transmission line 3 is not used increases, and the throughput also deteriorates. Therefore, it is very important to properly set the backoff set time range. The controllability of this type of CSMA / CD system can be generally evaluated by the following three points.

(I) Throughput This is the effective usage rate of the transmission path 3 described above.

(II) Delay time in the network This is the time from when a packet is given to the sending buffer of the communication device until it is sent successfully. The average time until the successful sending of the packet starts is the average delay time, and the success of the packet. The time when the probability that the transmission has started exceeds 99 (%) is called 99% delay time, and these are used as an evaluation measure.

(III) Stability When the load on the network increases and many communication devices each have communication packets, collisions frequently occur on the transmission path. As a result, an abnormal congestion phenomenon occurs in which packet transmission is almost impossible and recovery from the state becomes difficult. It is necessary to ensure that such a failure does not occur or that the probability of occurrence of a failure is extremely low.

Of these three points, by observing the throughput and the delay time in the network over a long period of time, knowledge information about the stability in the packet communication network can be obtained. Therefore,
In evaluating the performance of a packet communication network, it is an important issue to measure the throughput and the delay time in the network.

[Problems of Background Art] Although the throughput can be relatively easily measured by observing all the communication packets that normally pass through the packet communication network, the measurement of the delay time in the network is
For example, it is necessary to directly generate a packet and directly measure the time until the packet is sent out.
Various problems have occurred.

That is, if it is assumed that the packet used for measuring the intra-network delay time is generated based on the actual communication demand, the measurement cannot be performed when communication is not required.

On the contrary, if a measurement packet is generated irrespective of the actual communication demand, for example, at regular time intervals, the measurement packet causes an increase in communication volume and increases congestion in the communication network. This will increase the delay time. In this way, the influence on the communication network when the in-network delay time is functionally observed can be almost ignored when the above-mentioned observation is intermittently performed. However, when measuring the average delay time, it is necessary to collect a large amount of data in a short time and inevitably generate a large number of measurement packets, so the state of the communication network changes significantly. However, there is a problem that the measurement accuracy becomes poor. For this reason, it is extremely difficult to measure the average delay time with high accuracy, and it is also difficult to determine the limit of measurement accuracy.

Furthermore, to measure the 99 percent delay time, 100
Since only one packet has a delay time exceeding 99%, it is necessary to collect a larger amount of data in a shorter time than the measurement of the average delay time, which causes the same problem.

As a result, a reliable performance evaluation of the packet communication network
It was extremely difficult to carry out by the above-mentioned conventional direct method.

[Object of the Invention] The present invention has been made in view of such circumstances, and an object of the present invention is to determine the distribution of delay time in a packet communication network without increasing congestion in the network. An object of the present invention is to provide a highly practical packet transmission delay measurement method capable of highly accurately measuring and evaluating the performance of the packet communication network with high reliability.

SUMMARY OF THE INVENTION The present invention is a packet communication network in which the transmission of communication packets is controlled by the CSMA / CD system, and the back-off control time during steady operation of a plurality of stations participating in the packet communication network is the same. Each of them is controlled according to a random variable represented by the same distribution function having a parameter, that is, the back
The OFF control time range is fixedly set or common to each station according to only the state of the transmission line so that the distribution of the packet transmission delay time of each station is the same, and in this state, at least on the transmission line of the packet communication network. The ratio of the idle time of each station is observed and the packet transmission delay time of each station is measured.

The present invention is a packet communication system in which a plurality of stations are interconnected by a transmission line, and each station controls the transmission of communication packets by the CSMA / CD method, wherein each of the stations has the transmission line for each predetermined time width. First to monitor whether or not the vacant state
Of the plurality of stations, second monitoring means for monitoring whether or not the packets transmitted by any two or more stations of the plurality of stations collide with each other, and when trying to transmit the packet. The transmission of the packet is stopped when the first monitoring means detects that the transmission line is not in an idle state or when the second monitoring means detects the collision of the packet during the transmission of the packet. In this case, the back-off control time indicating the waiting time until the next attempt to start transmission of the packet is uniformly generated by generating integers in the range from 0 to a predetermined upper limit value r-1 with the same probability. The upper limit value is set randomly according to the distribution, and the initial value of the upper limit value is the same as that of the back / off control time setting means set in advance for all stations, and the upper limit value r
Parameter adjusting means for decreasing a predetermined amount when the empty state is detected by the first monitoring means and increasing a predetermined amount when a collision is monitored by the second monitoring means, A process of obtaining the probability p that the transmission line is idle based on the monitoring result of the first monitoring means for each predetermined time width, or the probability p that the transmission line is idle based on the monitoring result, A first calculation unit that performs at least one of the processes of obtaining the distribution of the number of transmission trials k required to start the successful transmission of the packet as p (1-p) ^k-1 based on the probability p; Process for calculating the distribution of packet transmission delay times, which indicates the time from the reception of a power packet to the start of successful transmission, as a composite distribution consisting of the uniform distribution and the distribution of the number of transmission attempts, or the packet The average packet transmission delay time indicating the average value of the distribution of the transmission delay time or the 99% packet transmission delay time indicating the time when the probability that the successful transmission of the packet has started exceeds 99%, the upper limit value r and the probability. and a second calculation unit that performs at least one of the processes of calculating from p.

Also, preferably, the first monitoring means, at regular time intervals,
It is characterized in that "1" is written in the case where there is a signal on the transmission path, and "0" is written in the case where there is no signal. In this case, when the data register is used up, it is preferable to return to the first bit position of the data register and continue writing.

Further, preferably, the first calculating means estimates the probability p that the transmission line is in an empty state by a ratio of the number of "0" s in the data register to the bit length of the data register. It is characterized by being.

Also, preferably, the average packet transmission delay time is (r
It is characterized in that it is calculated by +1) / 2p.

Also, preferably, the 99% packet transmission delay time is (r + 1) / 2p · {log _e C ₀ / log _e (1-p) + C ₁ } (where C ₀ and C ₁ are predetermined constants). It is characterized by being calculated by.

Further, preferably, the parameter adjusting means further decreases the upper limit value r by a predetermined amount or increases it by a predetermined amount when a specific signal or a specific packet passes through the transmission path. .

Further, preferably, the parameter adjusting means is the upper limit value r.
Is reduced by a predetermined amount, the upper limit value r is divided by 2, and when the upper limit value r is increased by a predetermined amount, the upper limit value r is multiplied by 2.

[Effect of the invention] Thus, according to the present invention, since the measurement of the delay time in the network is passively performed, the congestion of the packet communication network is not increased by this measurement, and therefore the delay in the transmission of the communication packet is reduced. It is possible to measure accurately. Moreover, since the delay time in the network is indirectly measured, unlike the conventional method in which the measurement packet is actually transmitted, the measured value does not fluctuate due to so-called variation of data, and the reliability of It is possible to obtain an effect such that a good and highly accurate measurement can be easily performed. Therefore, it becomes possible to evaluate the performance of the packet communication network easily, with high reliability and with high accuracy without hindering normal packet communication, and its practical advantage is enormous.

[Embodiment of the Invention] An embodiment of the present invention will be described below with reference to the drawings.

The control of packet transmission by the CSMA / CD method basically starts packet transmission when carrier sense is in the ON state at the time of access to the transmission path, or when carrier sense is in the OFF state and this causes a collision. If the transmission of the packet is stopped at this time, the packet is backed off for a predetermined time and then again accessed for transmission. The backoff is performed at the time of carrier sense and collision detection, and the backoff time at this time is set as follows.

That is, in this embodiment, the state of the transmission line is constantly monitored at the following two points.

(A) Collision detection (b) Transmission line idle state duration This transmission channel idle state duration detection is similar to carrier sense, but when carrier sense accesses the transmission line, While it detects whether the path is busy or idle, the above access is made to the time during which the idle state (carrier sense off) continues after the carrier sense of the transmission path is turned off. The difference is that it is detected independently.

Therefore, this empty state detection is determined by, for example, whether or not the idle time exceeds a predetermined time, for example, 800 bit time. Then, the communication device variably controls the back-off time setting range for the back-off control using the information on collision detection and the information on the empty state.

That is, when a communication packet to be transmitted is given, the communication device sets the backoff range r for the packet to 2 ^N [backoff unit time]. This backoff unit time is set with a maximum round-trip propagation delay time between communication devices as a margin, and is set to 400 bit time, for example. The above 2 ^N, as described below, regardless of the presence or absence of packet, there is constantly expanding, the adjustment of the reduction is carried out, for example, is set to a value of 2 ^3-2 about ^15. 0 within this 2 ^N [back-off unit time] range
The initial backoff time for the packet is set as a value of a uniform random number taking an integer value of (2 ^N -1). After the initial backoff time has elapsed, the packet transmission is attempted for the first time.

Then, when this transmission attempt is made, when carrier sense is in the ON state, or when collision is detected when carrier transmission is started with carrier sense being in the OFF state, a uniform random number is again generated. Is used for backoff, and the backoff range is set as follows.
That is, when a packet collision occurs between other stations or between the own station and another station on the transmission path, the backoff range r is defined as r: = min (r × 2,2 ^Nmax ). Further, when the idle state is detected for a predetermined time as the information of the transmission line during the back-off control, r: = max (r / 2,2 ^Nmin ) is set. When the backoff is performed by the carrier sense, the backoff time is set by using the backoff range r set at the time of access, and the backoff is performed by the collision detection caused by the transmission trial. At times, using the back-off range after being changed according to the transmission line situation as described above,
The backoff time is set. The update of the back-off range is executed every time when collision detection or predetermined idle state continuation detection is performed by monitoring the transmission path, completely independently of packet access.

Then, in the packet communication network configured by adopting the method as described above, the packet transmission delay time is measured as follows.

That is, each station participating in the packet communication network observes the state on the transmission line every back-off unit time, determines whether the transmission line is empty within the unit time, and records the result. In this determination, for example, when a packet signal or a collision signal passes even in a back-off unit time of 400 bit time, it is determined that the state of the transmission path in the unit time is not empty. By recording such a determination result, that is, information having empty and non-empty elements over a certain unit time, it is possible to obtain an information sequence of a certain length.

Now, if p = (number of empty lines in the sequence) / (length of sequence) is defined, it can be said that the above p represents one estimated value of the probability that the transmission path is empty.

Similarly, defining q = (the number of non-empty points in the sequence) / (the length of the sequence), q (= 1-p) represents one estimate of the probability that the transmission path is non-empty. It can be said that there is.

Considering now one station having a packet to be transmitted, in this case, the probability of successful packet transmission in the first packet transmission attempt is approximately p from the above estimation value, and the probability of failure is It can be said that it is q.
Then, the probability that the first transmission attempt fails and the second transmission attempt succeeds is given by (p · q) from the above estimated value, and similarly, the (k−1) th transmission attempt fails. However, the probability that the transmission attempt will be successful for the ^{first time} is (q ^k-1
p). This means that the number of trials until a station having a packet succeeds in sending the packet approximately follows a geometric distribution.

By the way, in this system, a random time waiting state, that is, a back-off state is always set before the first packet transmission trial. Needless to say, the number of back-offs becomes equal to the number of packet transmission attempts. Therefore, the distribution of the number of back-offs follows the distribution of the number of trials.

Here, if the distribution of the time spent for each back-off is determined, the distribution of the packet transmission delay time will be completely defined together with the above-mentioned distribution of the number of back-offs. .

On the other hand, in the above method, the delay time per back-off depends on the uniform random number within the predetermined range r, and the predetermined range r may be different for each station. However, if the situation in which all stations do not perform packet transmission attempts continues for a sufficiently long time, the back-off range r of each station contracts and converges, and finally the back-off range r of each station is reduced. It becomes the minimum respectively. The occurrence probability of such a phenomenon obviously has a positive value, and it can be said that it always occurs when the steady operation of the communication network where there is no station newly joining the packet communication network continues.

Also, for example, even if there is a new joining station, if the number of stations is not large, the back-off range r of the majority of other stations will be aligned with each other from the state where they converge to the minimum as described above. It is thought that it will change to. Therefore, it can be considered that the packet range of a representative station participating in the packet communication network is equal to the back-off range r of the station that has always been in operation after the communication network starts operating.

The above is not limited to the case where the transmission line is empty for a sufficiently long time, but is not limited to the case where packet collisions frequently occur on the transmission line for a sufficiently long time. The same happens even if there is, and in this case, the back-off range of each station will be set to the maximum value.

Thus, when the packet communication network is operated steadily, it can be considered that the distribution of the back-off times of each station is equal to the distribution of the back-off time for each station. This means that the composite distribution indicated by the distribution of the number of back-offs and the distribution of the back-off time, that is, the packet transmission delay distribution, is equal among the stations. Therefore, in this embodiment, under such a condition, the probability p that the transmission line is empty and its back
By checking the off range r and calculating the packet transmission delay distribution based on these pieces of information, the information regarding the packet transmission delay is passively obtained from the distribution and the performance of the packet communication network is evaluated.

It should be noted that there is no need to adopt the above-mentioned probabilistic method in order to align the back-off range r, for example, without fixedly setting the back-off range of each station or using the experience of each station. The back-off range may be comprehensively determined depending on only the state of the transmission path. Also,
Other methods can be adopted, and the point is that the back-off ranges of the stations participating in the packet communication network should be aligned with each other during steady operation. That is, the packet transmission delay distribution of each station may be given as the same distribution, for example, a combination of a geometric distribution having the same parameter p and a uniform distribution of integers (0 to r−1).

When such a distribution is given to each station, for example, the average delay time can be easily calculated as (r + 1) / 2p [backoff unit time] as the average of the above complex geometric distribution, and this can be calculated by the communication. It may be used as one index for network performance evaluation.

When 99% delay time is used as an index, an appropriate approximate expression, for example, (r + 1) / 2p · {log _e C ₀ / log _e (1-p) + C ₁ }, where C ₀ and C ₁ are predetermined Constant) [backoff unit time] may be used.

Next, a description will be given of a communication device in which the packet transmission delay distribution of each station is made uniform by adopting the above-mentioned method, and a packet transmission delay measuring device which is one embodiment of the method of the present invention incorporated in this communication device.

The data format of the communication packet transmission-controlled via this communication device is, for example, as shown in FIG.
As the data part, 1 to 8 (kbit) are prepared. And
In addition to these data parts, preamble, delimiter,
Information such as communication destination and sender is added.

As shown in FIG. 4, the communication device includes a memory (packet buffer) 21 for storing communication packets, an adapter control unit 22 for controlling the operation of the entire device, a reception control unit 23, and a transmission control unit 24. It is configured by the back-off control unit 25.

The back-off control unit 25 is, for example, as shown in FIG.
ROM25b that mainly stores 5a and stores its operating program,
Programmable interface circuit (PIO) 25c, 25
d, Programmable interrupt controller (P
IC) 25e, programmable timer counter (PTC) 25
It is configured by connecting f to each other via a bus 25g. The back-off control unit 25 operates a back-off time for a communication packet, issues a transmission request to the transmission control unit 23 every time the counter times out during back-off, and prompts the transmission of the communication packet. At this time, the above-mentioned back-off time range is set and controlled as described above according to the information on collision and continuation of an empty state detected by monitoring the transmission path.

FIG. 6 shows a schematic configuration of the transmission control unit 24, the operation of which is controlled by the transmission control microsequencer 24a. Then, the parallel 16-bit data provided from the memory 21 is input to the input buffer 24b, and the data is divided into upper 8 bits and lower 8 bits, and the shift register 24c
Serial conversion via. At this time, a CRC code for the above data is generated by the CRC-CCITT circuit 24d. These data and CRC code are then sent to the multiplexer 24f together with various codes from the data memory 24e such as CD zone.
Selected and is sent as a packet of the above format via the synchronization flip-flop 24g. Also,
These series of packet transmission processes are performed by the CPU interface 24h, pointer table 24i, address counter 24h.
This is performed in cooperation with the access processing of the adapter control unit 22 and the memory 21 by j.

Then, the transmission control unit 24 is the back-off control unit.
Upon receiving the transmission request from 25, the transmission of the communication packet starts only when the carrier sense is off at this time. Also, at this time, the reception control unit 23 detects a CD by collision detection.
Upon receiving the signal, the transmission of the communication packet is immediately stopped. Further, the transmission control unit 24 is adapted to receive an ACK / NAK transmission request from the reception control unit 23 and transmit an ACK / NAK packet. Then, these operations are performed by exchanging information regarding transmission with the adapter control unit 22.

Further, FIG. 7 shows a configuration example of the reception control unit 23. The overall operation of the reception control unit 23 is controlled by the reception control microsequencer 23a. The signal received via the transmission path is input to the shift register 23b, and the detector 23c performs start delimiter and ACK / NAK detection. Then, the output of the shift register 23b is transferred to the 16-bit shift register 23d, and the address comparison is performed by the address comparator 23e from the upper 8 bits.
According to this determination result, the received data is transferred to the memory 21 via the buffer register 23f. Further, receiving the output of the shift register 23b, the counter 23g checks the reception size thereof, and the CRC-CCITT circuit 23i cooperates with the comparator 23j to check the code error based on the CRC code. Then, a series of these processes is performed by the CPU interface 23k, the pointer table 23, and the address counter 23m together with the access process of the adapter control unit 22 and the memory 21.

Thus, the reception control unit 23 configured as described above performs carrier sensing based on the presence / absence of a signal on the transmission path, and also detects the presence / absence of packet collision on the transmission path. In addition, ACK / NA transmitted via the transmission line
Receiving and recognizing the K signal and detecting the timeout when there is no response. If the destination address of the received packet indicates self, upon completion of receiving the packet,
Immediately, the transmission control unit 24 is instructed to request an ACK / NAK return. Then, these series of operations are controlled while exchanging information regarding reception with the adapter control unit 22.

8 to 10 show control sequences of these respective parts. FIG. 8 shows reception control, FIG. 9 shows transmission control, and FIG.
And FIG. 10 shows the back-off control.

When transmitting a packet from the communication device configured as described above, the transmission instruction content is written in the pointer table 24i of the transmission control unit 21 under the control of the adapter control unit 21. The transmission control unit 24 transmits this transmission instruction to the backoff control unit 25. In response to this, the backoff control unit 25 confirms the packet transmission instruction and outputs a transmission request to the transmission control unit 24 when the backoff time set within the initially set range times out. . The back-off control unit 25 constantly monitors the status of the transmission path, updates the back-off range, and prepares for the next back-off process.

Then, the transmission control unit 24 having received the transmission request first checks the presence or absence of carrier sense from the reception control unit 23, and if the transmission path is free, starts the packet transmission. For the transmission of this packet, a CD element is sent out after instructing the laser element as an optical transmitter to pre-bias. This cd
When no collision is detected by the reception control unit 23 at the time of sending the zone signal, the data forming the packet is sent in order of the preamble, the start delimiter, the destination address ...
When a collision is detected during the CD zone signal transmission, the packet transmission processing is immediately stopped.

In this way, when the transmission controller 24 has finished sending one packet of data, it sends a CRC code as FCS next,
Following this, a termination delimiter is sent to terminate the packet communication. After completion of this transmission, it has the reception of ACK or NAK, and updates the pointer table 24i when receiving the notification of the reception of the ACK signal from the reception control unit 23. Then, a transmission completion interrupt is issued to the adapter control unit 22. When a NAK signal is received instead of the ACK or when there is no response for a predetermined time, the retransmission counter for the communication packet is incremented and the control is returned to the next transmission instruction.

On the other hand, packet reception control is performed as follows.
The reception control unit 23 first receives the CD zone signal and sets a carrier sense signal. Also, at this time, it is checked whether or not a collision has occurred. When a collision is detected, the reception of the communication packet is immediately stopped, and the process waits until there are no carriers on the transmission path. When the carriers on the transmission line are exhausted, the carrier sense signal is dropped to return to the initial reception waiting state. On the other hand, if the above collision does not occur, it has the reception of the start delimiter and checks whether the destination address matches its own address or the broadcast address, and if it is either of them, it is stored in the data memory 21 of the received packet. Put in. After that, check CRC error and frame error at the end of packet reception.
23l is updated and a reception completion interrupt is issued to the adapter control unit 22. At the same time, the transmission control unit 24 is instructed to transmit ACK.

Even if the above data is received correctly, if the buffer for fetching this is not prepared in the memory 21,
The NAK signal is instructed to be returned and the same communication packet is requested to be retransmitted. Also, if the destination address is different or there are many errors, ACK / NAK will not be returned. After that, the reception control unit 23 waits for the reception of ACK / NAK regardless of the destination address and the transmission address after receiving the data packet, and responds to the transmission control unit 24 with ACK, NAK, or no response. As a result, you will be notified. Then, after the end of the interframe gap, the carrier detection signal is dropped to indicate the vacancy of the transmission path.

In this way, the reception control unit 23 operates independently of the other control units and transmits the state of the transmission path to the other control units. Further, if the reception sequence is erroneous for some reason, or if the power is turned on later than the operating system to start the reception operation, it may not be possible to determine which part of the format the received data is. In this case, carrier non-detection of, for example, 15 (μsec) or more, or establishment of synchronization can be achieved by using the ACK / NAK signal as a clue.

As described above, transmission and reception of packets are controlled. The setting of the back-off time is considered to be an appropriate time, which is always updated according to the transmission path situation, independently of the packet transmission trial.

On the other hand, the packet transmission delay measuring device is configured, for example, as shown in FIG. The device is provided in a station that participates in a packet communication network, and monitors a state of the transmission line by a transmission line monitoring unit 31 and a bandage (empty or non-empty) of the transmission line determined by the transmission line monitoring unit 31. The memory 32 for storing, the back-off range calculation unit 33 for calculating the back-off range r of the station according to the state of the transmission line, and the transmission line being empty based on the information stored in the memory 32. The calculation of the packet transmission delay time in the communication network using the empty ratio p for calculating the probability p and the information of the back-off range r and empty ratio p obtained by the calculating units 32 and 34, respectively. The average delay calculation unit 35 for

The transmission line monitoring unit 31 passively monitors the state of the transmission line,
It is checked at every 0 bit time whether or not the transmission line is empty throughout the time according to the control flow shown in FIG. 12, and the judgment result is stored in the memory 32. The storage of this judgment result in the memory 32 is controlled by the pointer set in the pointer register 36, as shown in FIG. Then, this transmission path monitoring is repeatedly performed until the maximum value N of the pointer is reached, and the memory 32 stores the monitoring result of the sequence length N.

From the information sequence thus obtained in the memory 32, the vacancy rate calculation unit 34 sets, for example, when the empty state of the transmission line is “0” and the non-empty state is “1”, p = (in the memory 32 The empty rate p of the transmission path is calculated as "the number of 0" / (sequence length N of the memory 32).

Further, the back-off range calculation unit 33, like the back-off range calculation unit (Fig. 10) in a normal adaptive station, responds to the presence or absence of a collision and the continuation of idle time as shown in Fig. 14, for example. Calculates the back-off range r.

Based on these calculation results, p and r, the average delay calculation unit 35
Calculates the packet transmission delay time as (r + 1) / (2p) [backoff unit time] and outputs it as the performance evaluation value of the packet communication network.

The performance index to be measured is not limited to the average delay,
99% delay time is also acceptable. FIG. 15 shows a block diagram of a packet transmission delay measuring device in that case. Fig. 15 shows 10
The average delay calculation unit 35 in the figure is replaced with a 99% delay calculation unit 36. In this case, for example, Is a good approximation.

As described above with reference to the embodiments, according to the present invention, the transmission line state is passively monitored and the packet transmission delay is measured, so that communication is not disturbed during the measurement.
Moreover, since the packet transmission delay is indirectly measured by directly monitoring the state of the transmission line, the measurement accuracy does not vary depending on the number of samples (data), and reliable and high-accuracy measurement is simplified. It becomes possible to do. Therefore, the performance of the packet communication network can be evaluated with high reliability, and practically great effects can be obtained.

The present invention is not limited to the above embodiment. For example, the back-off range r of each station may be fixedly set, and in this case, the measurement is unnecessary. Further, the back-off range r of each station may be varied in a mutually aligned state only when a particular station transmits a special signal or transmits a packet. Needless to say, the CSMA / CD system is not limited to the above-mentioned non-persistent system, and other specifications may be defined according to the network specifications. In short, the present invention can be variously modified and implemented without departing from the scope of the invention.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration example of a local network, FIG. 2 is a diagram showing a conventional basic CSMA / CD system control sequence, and FIGS. 3 to 10 are CSMA / CD system communication devices. FIG. 3 is a diagram showing a format example of a communication packet, FIG. 4 is a schematic configuration diagram, FIG. 5 is a configuration diagram of a back-off control unit, FIG. 6 is a configuration diagram of a transmission control unit, FIG. FIG. 7 is a block diagram of the reception control unit, FIG. 8 is a diagram showing a reception control sequence, FIG. 9 is a diagram showing a transmission control sequence, FIG. 10 is a diagram showing a back-off control sequence, and FIG. 14 to 14 are views for explaining the method of one embodiment of the present invention.
FIG. 11 is a schematic configuration diagram of a packet transmission delay measuring device, FIG. 12 is a diagram showing a control sequence of a transmission line monitoring unit, FIG. 13 is a diagram showing a configuration example of a memory, FIG. 14 is a back-off range calculation FIG. 15 is a diagram showing a control sequence, and FIG. 15 is a schematic configuration diagram of a 99% transmission delay measuring device. 31 ... Transmission path monitoring unit, 32 ... Memory, 33 ... Back-off range calculation unit, 34 ... Vacancy ratio calculation unit, 35 ... Average delay calculation unit, 36 ... 99
Percentage delay calculator.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kiyoshi Yoneda 1 Komukai Toshiba-cho, Kouki-ku, Kawasaki-shi, Kanagawa Inside the Toshiba Research Institute Co., Ltd. (56) Reference JP-A-59-2463 (JP, A)

Claims (8)

[Claims] 1. A packet communication system in which a plurality of stations are interconnected by a transmission path, and each station controls transmission of communication packets by the CSMA / CD system, wherein each station performs the transmission at a predetermined time width. First monitoring means for monitoring whether or not the route is idle, and second monitoring for monitoring whether or not a collision has occurred between packets transmitted by any two or more stations of the plurality of stations. And the first monitoring means detects that the transmission line is not in an idle state when trying to send the packet, or the second monitoring means collides with the packet during transmission of the packet. When the transmission of the packet is stopped due to the detection of the back-off control time, the back-off control time, which indicates the waiting time until the start of the next transmission of the packet, is set to a range from 0 to a predetermined upper limit value r-1 An integer in each of which is randomly set according to a uniform distribution generated with the same probability, and the initial value of the upper limit value is a back-off control time setting means which is set in advance for all stations, and the upper limit value. parameter adjusting means for reducing r by a predetermined amount when the first monitoring means detects an empty state, and increasing it by a predetermined amount when the second monitoring means detects a collision, A process of obtaining a probability p that the transmission line is in an idle state based on a monitoring result of the first monitoring means for each predetermined time width,
Alternatively, the probability p that the transmission line is in an idle state is calculated based on the monitoring result, and the distribution of the number of transmission trials k required until the successful transmission of the packet is started is p (1-p) based on the probability p.
The first calculation means for performing at least one of the processes for obtaining ^k-1 ; and the distribution of the packet transmission delay time indicating the time from the reception of a packet to be transmitted by the station to the start of successful transmission, the uniform distribution And a process for calculating a composite distribution consisting of the distribution of the number of transmission attempts, or the average packet transmission delay time indicating the average value of the distribution of the packet transmission delay time or the probability that the successful transmission of the packet has started.
A packet communication system comprising: a second calculating means for performing at least one of a process of calculating a 99% packet transmission delay time indicating a time exceeding 99% from the upper limit value r and the probability p. . 2. The first monitoring means writes "1" when a signal is present on the transmission line and "0" when there is no signal in the data register at regular time intervals. The packet communication system according to claim 1. 3. The packet communication system according to claim 2, wherein when the data register is used up, the writing is continued by returning to the head bit position of the data register. 4. The first calculating means calculates the probability p that the transmission line is in an empty state when the number of "0" in the data register is data.
The packet communication system according to claim 2 or 3, wherein the estimation is performed by a ratio of the register to the bit length. 5. The average packet transmission delay time is (r + 1)
The packet communication system according to claim 1, wherein the packet communication system is calculated by / 2p. 6. A packet transmission delay time of 99% is (r + 1) / 2p. {Log _e C ₀ / log _e (1-p) + C ₁ } (where C ₀ and C ₁ are predetermined constants). The packet communication system according to claim 1, wherein the packet communication system is calculated by: 7. The parameter adjusting means further reduces the upper limit value r by a predetermined amount or increases it by a predetermined amount when a specific signal or a specific packet passes through the transmission line. A packet communication system according to claim 1. 8. The parameter adjusting means divides the upper limit value r by 2 when decreasing the upper limit value r by a predetermined amount, and sets the upper limit value r when increasing the upper limit value r by a predetermined amount. The packet communication system according to claim 1 or 7, which is multiplied by 2.