JP6900853B2 - Device linkage server and device linkage program - Google Patents

Description

The present invention relates to a device cooperation server and a device cooperation program in an environment in which a wide variety of IoT (Internet of Things) devices are connected to a network.

With the progress of IoT technology in recent years, many IoT devices such as sensors, actuators, and robots have come to be connected to a network. IoT provides services by giving an object a communication function, connecting it to a network, and analyzing data. The scope of application of IoT is wide-ranging, and it is expected to be applied not only to manufacturing, distribution, and maintenance, which are being considered in Germany's Industry 4.0, but also to health management, medical care, agriculture, energy, and so on.

However, current IoT applications tend to be vertically integrated one-off solutions tailored to specific targets. In other words, it is often used in the SI (System Integration) business, which collectively performs sensor selection, analysis, display, action, etc., and provides it as a system. As a result, the cost of each IoT service is high, and the current situation is that a variety of services are not sufficiently available.

In order to develop and operate various IoT services at low cost, it is necessary to separate devices such as sensors and actuators from services so that devices and services can be used horizontally. .. This way of thinking is called open IoT, and it is emphasized in order to activate IoT services in the future.

For such open IoT, Tacit Computing technology has been proposed as a mechanism for freely using a device from a service (see Non-Patent Document 1). Tacit Computing is a technology that enables users to discover and use devices that hold the data they need on demand, based on the live data that the device holds at that time.

FIG. 7 is a diagram for explaining an outline of Tacit Computing.
Tacit Computing continues the service in response to the user's request by discovering and coordinating the appropriate resources for the service to the user from the three layers of the cloud layer, the network layer, and the device layer (reference numeral in FIG. 7). a). Further, in the configuration of the three layers, in order to cope with the situation that changes from moment to moment, processing is performed as much as possible in the device layer close to the site where data is generated (reference numeral b in FIG. 7). By performing the processing at a lower layer, it is possible to reduce the network traffic (reference numeral c in FIG. 7) and suppress the outflow of highly privacyable data (reference numeral d in FIG. 7).

This Tacit Computing uses live data retrieval technology and device virtualization technology as its elemental technologies.
Live data search technology is a technology for searching for a device that provides data required by a user. As an example of the IoT service, there is a service in which a plurality of sensors are installed in a facility such as a bridge to monitor the deterioration status. In this case, since it is unlikely that the deterioration status will progress rapidly, it is sufficient to upload the sensor data at a plurality of points to the cloud at a cycle of several hours or the like and analyze the change in the deterioration status by statistical software or machine learning. On the other hand, in the example of a service that provides information guidance and warning alerts to a person captured by a fixed-point camera, the person appears on the camera for only a few seconds, and only the image of the camera showing the person is displayed. It can be said that the data is meaningless to a person. Data that occurs in the device layer and changes from moment to moment is called live data.

In order to search for this live data that the user needs, in Tacit Computing, instead of waiting for the data to come up to the cloud layer, a function to analyze is placed in the lower layer and the live data is searched in the lower layer. ..
For example, suppose a friend participates in a relay race and wants the camera image of the friend to be automatically connected. In this case, for example, if a friend's bib number is requested as a search key, Tacit Computing arranges an image analysis function such as OpenCV at the gateway or network edge that houses the camera. Then, by analyzing the image at a place close to the camera, the bib number of the friend is extracted by image analysis, and the camera in which the friend is shown can be identified. In this way, live data retrieval technology is used in Tacit Computing.

Next, when the device to be used is identified by the live data search technology, it is necessary to actually use the device. IoT devices are developed by many manufacturers, and the protocol, interface, address, etc. at the time of use differ for each device. Therefore, device virtualization technology absorbs the differences in interfaces in individual devices.
For example, in the above example, based on a common request such as acquisition of a camera image, an adapter for each device is provided at a gateway or the like that accommodates the camera, and the protocol or the like is converted and then individually. Make a request according to the camera of. By using such a device virtualization technology, the user can use the device without being aware of the difference between the devices.

Junko Sasaki, 5 others, "Tacit Computing: Network Service for the Age of Open IoT / Unevenly Distributed Resources", Business Communication Co., Ltd., Business Communication, Vol.54, No.5, pp.24-25, May 2017

With the above-mentioned Tacit Computing technology, users can discover devices with the required live data and use the device virtualization mechanism to hide the interfaces of individual devices so that users can use the devices ad hoc. It becomes. However, there may be cases where the user's needs can only be met temporarily by simply discovering and using the device according to the situation at that time using Tacit Computing. Hereinafter, a specific description will be given.

Consider a tracking camera as an example of using Tacit Computing. The service using a tracking camera is to acquire images of a child such as an elementary school student on campus or while going to and from school from a camera near the child and view it on a parent's mobile terminal or the like. The service can still meet the needs of parents by allowing parents to visually check the safety of their children with a tracking camera. However, if this service is charged by setting a fee for each time, it is necessary to arrange the image processing function at the gateway accommodating the camera, the network edge, etc. in order to search the live data each time, which is costly. It will take. Rather, they continue to acquire images from cameras near their children, but parents can watch their children's images when they want to see them on their mobile terminals, and even when they are not watching them, they watch over their children by machine learning, etc. It is considered necessary to provide continuous services such as sending alerts to parents in the event of an abnormality such as when a child approaches.

As for the processing of the tracking camera, Tacit Computing finds a camera showing a child, and when the parent requests, the image of the camera is delivered to the parent's mobile terminal, etc., and the gateway or the gateway that houses the camera is used for watching. An image processing function (for example, OpenCV library, etc.) is arranged in the SSE (Subscriber Service Edge) of the network edge, and an image is cut out and analyzed from the image of the camera showing the child. Then, the result of the image analysis is summarized after being summarized into a feature vector, and in the cloud layer, an abnormal value analysis such as an approaching suspicious person is analyzed using a machine learning method (Local Outlier Factor, etc.). If there is an abnormality, an alert is sent to the parent's mobile terminal.

The tracking camera itself can be realized as an SI service using a camera and machine learning without using Tacit Computing technology. However, even when using conventional technologies such as Tacit Computing, it has been realized to discover available devices and provide services that continuously use those devices at a reasonable price, that is, at a low cost. Absent.

The present invention has been made in view of these points, and to provide a device cooperation server and a device cooperation program capable of continuously and reasonably providing a service for dynamically discovering and using an IoT device. Is the subject.

In order to solve the above-mentioned problems, the invention according to claim 1 is a device cooperation server that selects a resource from three layers of a cloud layer, a network layer, and a device layer to provide a service, and provides the service. At the time, in the optimization period indicating a predetermined trial period provided for determining the resource to be used, the usage history collection unit that collects the usage history of the service of each device belonging to the device layer, and the collected usage. With reference to the history, one or more of the three layers of the cloud layer, the network layer, and the device layer are determined as the allocation destination of the function related to the service, and the function allocation destination determination unit and the function related to the service are described. A device cooperation server characterized by providing a function distribution unit that distributes to one or a plurality of layers of the determined three layers and deletes a function related to the service in the layer that is not determined as the placement destination. did.

In this way, the device linkage server determines one of the three layers, the cloud layer, the network layer, and the device layer, as the placement destination of the functions related to the service, based on the device usage history collected during the optimization period. Can be done. As a result, the functions can be allocated to the resources of the layer related to the device actually used in the service, so that the service can be continuously provided at a reasonable price.

The invention according to claim 2 is a device cooperation server that selects a resource from three layers of a cloud layer, a network layer, and a device layer to provide a service, and determines a resource to be used when providing the service. In the optimization period indicating a predetermined trial period provided for the purpose, the optimization is performed by referring to the usage history collection unit that collects the usage history of the service of each device belonging to the device layer and the collected usage history. The function allocation destination determination unit that determines the gateway of the area to which the device used during the period belongs as the allocation destination of the function related to the service, and the function related to the service are delivered to the determined gateway and as the allocation destination. The device cooperation server is provided with a function distribution unit for deleting the function related to the service of the gateway which has not been determined.

In this way, the device cooperation server can arrange the function for realizing the service at the gateway accommodating the device used for providing the service based on the usage history of the device collected during the optimization period. Therefore, it is not necessary to allocate the function to the gateway in the area where the analysis function is unnecessary, and the server cost and the communication cost can be reduced. This makes it possible to continue to provide services at a reasonable price.

The invention according to claim 3 is a device cooperation server that selects a resource from three layers of a cloud layer, a network layer, and a device layer to provide a service, and determines a resource to be used when providing the service. In the optimization period indicating a predetermined trial period provided for the purpose, the optimization is performed by referring to the usage history collection unit that collects the usage history of the service of each device belonging to the device layer and the collected usage history. The function allocation destination determination unit that determines the network edge of the network layer connected to the gateway of the area to which the device used during the period belongs as the allocation destination of the function related to the service, and the function related to the service are determined. The device cooperation server is characterized by including a function distribution unit that distributes to the network edge and deletes the function related to the service of the network edge that has not been determined as the allocation destination.

In this way, the device linkage server arranges functions for realizing the service at the network edge connected to the gateway that accommodates the device used for service provision, based on the device usage history collected during the optimization period. can do. Therefore, it is not necessary to allocate the function to another network edge or the like that does not require the analysis function, and the server cost and the communication cost can be reduced. As a result, even if the function cannot be assigned to the gateway of the device layer, the service can be continuously provided at a reasonable price.

The invention according to claim 4 is further provided with a network setting unit for executing network settings including bandwidth allocation and priority control related to the provision of the service in the network to which the determined network edge belongs. The device cooperation server described in 3 was used.

By doing so, the device cooperation server can set the network for the network to which the network edge determined as the function allocation destination is based on the device usage history collected during the optimization period.

The invention according to claim 5 is a device cooperation server that selects a resource from three layers of a cloud layer, a network layer, and a device layer to provide a service, and determines a resource to be used when providing the service. In the optimization period indicating a predetermined trial period provided for the purpose, the optimization is performed by referring to the usage history collecting unit that collects the usage history of the service of each device belonging to the device layer and the collected usage history. A function allocation destination determination unit that determines the gateway of the area to which the device used during the period belongs, or the network edge of the network layer connected to the gateway as the allocation destination of the function related to the service, and the function related to the service. Is distributed to the determined gateway or network edge, and at the gateway and network edge other than the determined gateway or network edge, the function distribution unit that deletes the function related to the service and the collected usage history are referred to. , Extract the network edge that accommodates the most frequently used devices during the optimization period, deploy the processing function of the cloud server in the cloud that has the least delay with the extracted network edge, and use the processing function. The device linkage server is characterized by having a resource securing processing unit that secures the required resource size.

In this way, the device linkage server extracts the network edge that accommodates the device that is used most frequently based on the device usage history collected during the optimization period, and the cloud server that has the least delay with the extracted network edge. It is possible to deploy the processing function of and secure the resource size. Therefore, it is possible to reduce the communication cost and secure the resource size only for the cloud server that requires the processing function, so that the service can be continuously provided at a reasonable price.

The invention according to claim 6 is a device cooperation server that selects a resource from three layers of a cloud layer, a network layer, and a device layer to provide a service, and determines a resource to be used when providing the service. In the optimization period indicating a predetermined trial period provided for the purpose, the optimization is performed by referring to the usage history collection unit that collects the usage history of the service of each device belonging to the device layer and the collected usage history. A function allocation destination determination unit that determines the gateway of the area to which the device used during the period belongs, or the network edge of the network layer connected to the gateway as the allocation destination of the function related to the service, and the function related to the service. With reference to the function distribution unit that distributes the server to the determined gateway or network edge, and the collected usage history, the network edge that accommodates the device most frequently used during the optimization period is extracted, and the extraction is performed. The cloud layer includes a resource securing processing unit that deploys the processing function of the cloud cloud server of the cloud with the least delay between the processing function and secures the resource size required for the processing function, and the cloud layer is a GPU and a CPU. The offload setting unit is composed of a heterocloud including the heterodevices of the above, and further includes an offload setting unit that offloads a specific process that is a part of a processing function deployed on the cloud server to the GPU. Based on the genetic algorithm, the source code of the application that realizes the processing function to be deployed is analyzed, the gene length is determined from the number of repetitive processing sentences that try to parallelize the GPU, and a predetermined number of genes are obtained. Create and assign the statement of the iterative processing to the source code corresponding to the gene as a gene value whether to process in parallel with the GPU or not with the GPU, and deploy it to the verification cloud server. By doing so, performance measurement was performed, the degree of compatibility of each gene was set based on the result of the performance measurement, a predetermined number of individuals with a high value of the set degree of compatibility were selected, and the genes were selected at a predetermined crossover rate. A part of the genes is exchanged between individuals at one point, a child individual is created, each gene value of the created individual's gene is changed at a predetermined mutation rate, and the processing of the genetic algorithm is performed. Repeated for a predetermined number of generations, the source cloud that corresponds to the highest degree of conformity Calculating a de as a solution, was characterized and to Lud vice linking server that is deployed in the cloud server.

By doing so, the device cooperation server can specify the process of offloading to the GPU by using the genetic algorithm, so that the process can be speeded up.

The invention according to claim 7 is a device cooperation program for causing the computer to function as the device cooperation server according to any one of claims 1 to 6.

By doing so, each function of the device cooperation server according to any one of claims 1 to 6 can be realized by using a general computer.

According to the present invention, it is possible to provide a device cooperation server and a device cooperation program that continuously and reasonably provide a service for dynamically discovering and using an IoT device.

It is a figure which shows the Tacit Computing system which includes the device cooperation server which concerns on this embodiment. It is a functional block diagram which shows the configuration example of the device cooperation server which concerns on 1st Embodiment of this invention. It is a functional block diagram which shows the configuration example of the device cooperation server which concerns on 2nd Embodiment of this invention. It is a functional block diagram which shows the configuration example of the device cooperation server which concerns on the modification of 2nd Embodiment of this invention. It is a functional block diagram which shows the configuration example of the device cooperation server which concerns on 3rd Embodiment of this invention. It is a functional block diagram which shows the configuration example of the device cooperation server which concerns on the modification of 3rd Embodiment of this invention. It is a figure for demonstrating the outline of Tacit Computing.

Next, the device cooperation server 1 and the like in the embodiment for carrying out the present invention (hereinafter, referred to as "the present embodiment") will be described.
FIG. 1 is a diagram showing a Tacit Computing system including the device cooperation server 1 according to the present embodiment.
The Tacit Computing system according to the present embodiment is characterized by including a device cooperation server 1 in addition to the conventional Tacit Computing configuration shown in FIG. 7. The device cooperation server 1 is communicably connected to each device located in the three layers of the cloud layer, the network layer, and the device layer, and the arrangement of various functions and the optimization of processing offload for realizing the service desired by the user are optimized. It is a device that performs.

In the conventional Tacit Computing system as shown in FIG. 7, when a device is ad hocly discovered and used, it is premised that the service can be provided first, so that cost and performance are not considered. However, in order to provide services continuously and reasonably, it is necessary to reduce operating costs by improving performance and the like.

Therefore, in the Tacit Computing system including the device cooperation server 1 according to the present embodiment, optimization is executed by appropriately performing function arrangement and processing offload in each layer of the device layer, the network layer, and the cloud layer. This optimization is performed during the first ad hoc device usage period (“during the optimization period” shown below). For example, in the case of a tracking camera service, the optimization period is set with the first day as the trial use period (predetermined trial period) so that the camera image near the child to be tracked can be confirmed on the parent's mobile terminal. By collecting information on the usage history of devices, networks, and clouds during this optimization period, the device linkage server 1 can determine the optimum function layout to any one (one or more) of the device layer, network layer, and cloud layer. To decide. As a result, for example, after the trial use period (optimization period) of the first day, it becomes possible to provide the tracking camera service at a reasonable price.

Hereinafter, a configuration example in which the device cooperation server 1 according to the present embodiment continuously provides the device use in the Tacit Computing system as a reasonable service will be described as the first to third embodiments.

<First Embodiment>
The first embodiment is an example of performing optimization at the device layer.
In the Tacit Computing system including the device linkage server 1A (see FIG. 2) according to the first embodiment, the device used for providing the service is specified in the device layer based on the usage history collected during the optimization period, and the device is specified. It is characterized in that a function for realizing the service is arranged (function arrangement) in the gateway 10 (see FIG. 1) in the area to which the specified device belongs.

In the device layer, it is assumed that devices are switched at any time in order to realize the service desired by the user. For example, in the case of a tracking camera service, it is necessary to select a camera (device) in which a child's image is displayed according to the place where the child exists. At this time, if image analysis is performed on all cameras (devices) connected to the network and only the image of the camera showing the target child is used, the cost (processing cost, communication cost) becomes very high. .. Therefore, during the optimization period, the process of acquiring the position information of the child by a predetermined method and selecting the camera existing in the vicinity of the child based on the information such as the installation location of the camera indicated by the metadata of the camera is performed. Do.

Further, in the device layer, the cost (processing cost, communication cost) is greatly affected by which gateway 10 is used to allocate the function for analysis from the group of gateways 10 accommodating the device. If it is a tracking camera, for example, an image analysis function such as OpenCV is used, but the gateway 10 in the area where the child passes is specified in the optimization period, and the image analysis function is arranged for the specified gateway 10 to perform processing.

FIG. 2 is a functional block diagram showing a configuration example of the device cooperation server 1A according to the first embodiment of the present invention. The device cooperation server 1A is a device that executes optimization of function arrangement for providing services in the device layer.
As shown in FIG. 2, the device cooperation server 1A includes a control unit 11, an input / output unit 12, and a storage unit 13.

The input / output unit 12 comprises a communication interface for transmitting / receiving information to / from each device belonging to the cloud layer, the network layer, the device layer, and the like, an input device such as a touch panel and a keyboard, and an output device such as a monitor. It consists of an input / output interface for sending and receiving information between.

The storage unit 13 is composed of a hard disk, a flash memory, a RAM (Random Access Memory), and the like.
The storage unit 13 stores the device information DB 131 and the distribution function information DB 132, a program for executing each function of the control unit 11 (device cooperation program), and information necessary for processing of the control unit 11 (device cooperation program). For example, device usage history) is temporarily stored.

Here, the device information DB 131 stores the metadata of each device existing in the device layer. At least the device identification information, the device type (information indicating the function of the device such as a camera), and the device position information are stored in the metadata of this device.
Further, the distribution function information DB 132 stores a function (processing program) arranged by the device cooperation server 1A on the gateway 10 accommodating the device to be used. For example, in the case of a tracking camera service, a processing program that realizes an image analysis function such as OpenCV is stored.

The control unit 11 controls the entire device cooperation server 1A, and includes a location information acquisition unit 111, a device selection unit 112, a usage history collection unit 113, a function placement destination determination unit 114, and a function distribution unit 115. ..
Further, the control unit 11 is realized by, for example, a CPU (Central Processing Unit) (not shown) deploying and executing a program (device cooperation program) stored in the storage unit 13 in a RAM.

The position information acquisition unit 111 acquires the position information of the object of the service (in the case of the service of the tracking camera, the child to be tracked) by a predetermined method. As a predetermined method, the position information obtained by the position detection function attached to the object of the service is received. For example, a child is made to hold a mobile terminal to detect position information by GPS, and the position information acquisition unit 111 receives the position information. In addition, non-IP (for example, NFC (Near Field Communication) Bluetooth, etc.) may be used for the IoT device, and the communication pattern of the IoT device including the non-IP device is analyzed and the location information is obtained. You may try to get it.
Further, the position information acquisition unit 111 outputs the position information acquired by the position information acquisition unit 111 to the device selection unit 112 during the optimization period.

The device selection unit 112 is a device for searching the device information DB 131 in the storage unit 13 based on the position information acquired by the position information acquisition unit 111 and realizing the service desired by the user. Select a device located near the indicated location.

The usage history collecting unit 113 acquires information (usage history) indicating the usage status of the device used in the service desired by the user during the optimization period. In this usage history, for example, in the case of a tracking camera service, information indicating whether or not the child to be tracked was shown during the optimization period is collected together with the identification information of the camera (device).

The function allocation destination determination unit 114 refers to the usage history collected by the usage history collection unit 113, and determines the gateway 10 in the area where the analysis function is arranged based on the history of the devices used during the optimization period.

The function distribution unit 115 extracts and distributes the function (processing program) stored in the distribution function information DB 132 in the storage unit 13 to the gateway 10 to which the function determined by the function arrangement destination determination unit 114 is to be arranged. .. For example, in the case of a tracking camera service, the function distribution unit 115 distributes a processing program that realizes an image analysis function such as OpenCV to the gateway 10 determined by the function placement destination determination unit 114 as the placement destination.
During the optimization period, the function distribution unit 115 extracts and distributes the function (processing program) stored in the distribution function information DB 132 to the gateway 10 accommodating the device selected by the device selection unit 112. Then, among the gateways 10, the function allocation destination determination unit 114 may delete the analysis function of the gateway 10 that has not been determined as the gateway 10 in the area where the analysis function is arranged.

As described above, according to the device cooperation server 1A according to the first embodiment, the device used for providing the service in the device layer is specified based on the usage history collected during the optimization period, and the specified device is specified. A function for realizing the service can be arranged in the gateway 10 accommodating the device.
Therefore, it is not necessary to arrange the function for the gateway 10 in the area where the analysis function is unnecessary. As a result, it is possible to prevent an increase in unnecessary traffic and reduce communication costs. Further, for the gateway 10 that accommodates only unused devices, it is possible to reduce the server cost associated with unnecessary processing. As a result, according to the device cooperation server 1A according to the first embodiment, it is possible to continuously and reasonably provide a service for dynamically discovering and using an IoT device.

<Second Embodiment>
The second embodiment is an example of performing optimization at the network layer.
In the Tacit Computing system including the device linkage server 1B (see FIG. 3) according to the second embodiment, the device used for providing the service is specified in the device layer based on the usage history collected during the optimization period, and the device is specified. It is characterized in that a function for realizing the service is arranged (function arrangement) at a network edge 20 (see FIG. 1) belonging to a network layer connected to a gateway 10 in an area to which the specified device belongs.

In the Tacit Computing system, as described in FIG. 7, in principle, processing is performed at a lower layer. Therefore, it is desirable to arrange the functions in the gateway 10 in the area to which each device belongs as much as possible, but depending on the type and performance of the gateway 10, it may not be possible to operate the processing program that realizes the service. In such a case, the device cooperation server 1B according to the second embodiment arranges functions at the network edge 20 of the network layer.
If the network edge 20 is an NGN (Next Generation Network), it is an edge router that connects the access network and the core network, and is, for example, an SSE (Subscriber Service Edge) or the like.

FIG. 3 is a functional block diagram showing a configuration example of the device cooperation server 1B according to the second embodiment of the present invention. The device cooperation server 1B is a device that executes optimization of function arrangement for providing services at the network layer.
The difference from the device cooperation server 1A shown in FIG. 2 is that the function placement destination determination unit 114 and the function distribution unit 115 become the function placement destination determination unit 114B and the function distribution unit 115B in the device cooperation server 1B of FIG. That is the point. A configuration having the same functions as the device linkage server 1A will be given the same name and reference numeral, and the description thereof will be omitted.

The function placement destination determination unit 114B refers to the usage history collected by the usage history collecting unit 113, and is connected to the gateway 10 in the area to which the used device belongs based on the history of the devices used during the optimization period. The network edge 20 of the network layer is determined as the network edge 20 in which the analysis function is arranged.
The function allocation destination determination unit 114B uses the technology of the information-oriented network (content-oriented network) to extract the edge router adjacent to the gateway 10 in the area to which the device used belongs, and arranges the analysis function at the network edge. It may be decided as 20. For more information on information-oriented networks, see, for example, Non-Patent Document 2 (Hitoshi Asaeda, "Possibilities and Research Issues Brought by Information-Oriented Networks," National Institute of Information and Communications Technology, Vol.61 No.2, 2015).

The function distribution unit 115B extracts and distributes the function (processing program) stored in the distribution function information DB 132 in the storage unit 13 to the network edge 20 to which the function determined by the function arrangement destination determination unit 114B is arranged. To do.
During the optimization period, the function distribution unit 115B provides the function (processing program) stored in the distribution function information DB 132 to the network edge 20 connected to the gateway 10 accommodating the device selected by the device selection unit 112. The analysis function of the network edge 20 that has not been determined as the network edge 20 in which the analysis function is arranged may be deleted in the function arrangement destination determination unit 114B so as to be extracted and distributed.

As described above, according to the device cooperation server 1B according to the second embodiment, the device used for providing the service in the device layer is specified and the specified device is specified based on the usage history collected during the optimization period. A function for realizing the service can be arranged at the network edge 20 of the network layer connected to the gateway 10 accommodating the device.
Therefore, even if the function cannot be arranged at the gateway 10 in the area to which the device used for providing the service belongs, the function can be arranged at the network edge 20 of the network area connected to the gateway 10. Further, it is possible to eliminate the need to arrange the function with respect to the other network edge 20 that does not require the analysis function. As a result, it is possible to prevent an increase in unnecessary traffic and reduce communication costs. Further, with respect to the network edge 20 connected to the gateway 10 accommodating only unused devices, it is possible to reduce the server cost associated with unnecessary processing. As a result, according to the device cooperation server 1B according to the second embodiment, it is possible to continuously and reasonably provide a service for dynamically discovering and using an IoT device.

<Modified example of the second embodiment>
As a modification of the second embodiment, a function of executing network settings such as bandwidth allocation and priority control may be provided in the network layer according to the characteristics of the service. For example, as shown in FIG. 4, the network setting unit 116 is additionally provided in the configuration of the device cooperation server 1B shown in FIG. 3, and is configured as the device cooperation server 1C.
The network setting unit 116 sets bandwidth allocation and priority control when providing services at the network layer. For example, in the case of NGN, the function of securing the bandwidth is realized by RACF (Resource and Admission Control Function).
By doing so, the device cooperation server 1C can also set the network for providing the service based on the usage history of the optimization period.

<Third Embodiment>
The third embodiment is an example of executing optimization in the cloud layer.
In the Tacit Computing system including the device linkage server 1D (see FIG. 5) according to the third embodiment, the device used for providing the service is specified in the device layer based on the usage history collected during the optimization period, and the device is specified. A function for realizing the service is arranged (function is arranged) at the gateway 10 in the area to which the specified device belongs or the network edge 20 belonging to the network layer connected to the gateway 10. Further, the device cooperation server 1D selects the DC30 (see FIG. 1) having a small delay with the network edge 20 accommodating the frequently used devices, and secures the resource size of the processing function in the cloud server of the DC30. It is a feature.

For example, in the case of a tracking camera service, an image is summarized by an arranged processing program at the gateway 10 or the network edge 20 to extract a feature vector. Then, the cloud server of DC30 acquires the feature vector and analyzes it by a method such as machine learning to analyze the approaching suspicious person and output an alert. In this case, the device cooperation server 1D selects a DC30 having a small delay with the network edge 20 accommodating frequently used devices, and has a function (for example,) of processing data generated in real time provided in the cloud server of the DC30. , Deploy stream processing features like Apache Storm and Apache Spark) to ensure proper resource size.

FIG. 5 is a functional block diagram showing a configuration example of the device cooperation server 1D according to the third embodiment of the present invention. The device cooperation server 1D is a device that executes optimization for providing services in the cloud layer.
The difference from the device cooperation server 1A shown in FIG. 2 is that the function placement destination determination unit 114 and the function distribution unit 115 become the function placement destination determination unit 114D and the function distribution unit 115D in the device cooperation server 1D of FIG. The point is that the resource securing processing unit 117 is newly provided. A configuration having the same functions as the device linkage server 1A will be given the same name and reference numeral, and the description thereof will be omitted.

The function placement destination determination unit 114D refers to the usage history collected by the usage history collection unit 113, and based on the history of the devices used during the optimization period, the gateway 10 in the area where the analysis function is placed, or its gateway. The network edge 20 of the network layer connected to 10 is determined as the placement destination of the analysis function. If the analysis function can be arranged in the gateway 10, the function arrangement destination determination unit 114D determines the gateway 10 as the arrangement destination. However, if the gateway 10 cannot be placed as the placement destination because the processing program cannot be placed in the gateway 10, the network edge 20 connected to the gateway 10 is determined as the placement destination of the analysis function.

The function distribution unit 115D provides a function (processing program) stored in the distribution function information DB 132 in the storage unit 13 to the gateway 10 or the network edge 20 to which the analysis function determined by the function arrangement destination determination unit 114D is arranged. Extract and deliver.
Here, the processing program distributed by the function distribution unit 115D is, for example, in the case of a tracking camera service, in addition to the function of image recognition (for example, face recognition) of a person (child) to be tracked, image recognition. It has a function to perform a process of summarizing an image determined to show a person (child) to be tracked and extracting a feature vector.

The resource securing processing unit 117 refers to the usage history collected by the usage history collecting unit 113, extracts the network edge 20 accommodating the device most frequently used during the optimization period, and delays with the extracted network edge. Deploy the processing function required for the service (for example, the function to analyze the approaching suspicious person from the feature vector and output an alert) to the cloud of DC30, which has the least number of resources. Further, the resource securing processing unit 117 secures the resource size necessary for executing the processing.
Regarding the resource size to be secured, the resource reservation processing unit 117 may secure a resource size preset according to the service, or sets an initial value as a required resource size and uses the resource. The predetermined value may be set to be increased depending on the situation.

By doing so, according to the device cooperation server 1D according to the third embodiment, the device used for providing the service at the device layer is specified based on the usage history collected during the optimization period, and the gateway is used. An analysis function (processing program) can be arranged at 10 or the network edge 20. Further, the device cooperation server 1D deploys the processing function required for the service to the cloud server of DC30 having the least delay with the network edge 20 accommodating the device most frequently used, and secures the resource size. Therefore, the operating cost can be reduced as compared with the case where the processing function of the cloud of all DC30 is deployed and the resource size is secured. As a result, according to the device cooperation server 1D according to the third embodiment, it is possible to continuously and reasonably provide a service for dynamically discovering and using an IoT device.

<Modified example of the third embodiment>
A modified example of the third embodiment will be described.
FIG. 6 is a functional block diagram showing a configuration example of the device cooperation server 1E according to a modified example of the third embodiment of the present invention. The difference from the device cooperation server 1D shown in FIG. 5 is that the offload setting unit 118 is further provided. The configuration having the same function as the device cooperation server 1D will be given the same name and reference numeral, and the description thereof will be omitted.

In recent years, in the cloud, servers equipped with heterogeneous HW (hardware) (hereinafter referred to as "hetro device") such as GPU (Graphics Processing Unit) and FPGA (Field Programmable Gate Array) are increasing. For example, FPGA is also used in Bing search of Microsoft (registered trademark). In this way, by utilizing the Hetro device, for example, offloading matrix calculation etc. to GPU and offloading specific processing such as FTT (Fast Fourier Transform) calculation to FPGA, high performance can be achieved. It has been realized. The offload setting unit 118 of the device cooperation server 1E shown in FIG. 6 is characterized in that the offload setting for the specific processing is set for the cloud server deployed by the resource securing processing unit 117. Hereinafter, two methods (method 1) and (method 2) executed by the offload setting unit 118 will be described.

(Method 1 of offload setting section)
In the method 1, the logic of specific processing such as matrix calculation is stored in advance in the storage unit 13 of the device cooperation server 1E as a pattern of offloading to a hetero device such as a GPU or FPGA. Then, when the request from the user includes a specific process, the offload setting unit 118 extracts the pattern (offload process description, for example, OpenCL, etc.) stored in the storage unit 13 to the cloud server. Make settings.

(Method 2 of offload setting section)
The method 2 is a method in which the offload setting unit 118 offloads the iterative statement in the specific process to a hetero device such as a GPU on the spot. In order to speed up processing by GPU or the like, it is necessary to parallelize an appropriate region, and in order to autonomously extract an appropriate region, for example, a method such as a genetic algorithm (GA) is used. By the method of a genetic algorithm (hereinafter referred to as "GA"), the possibility of parallelization of repeated statements of a specific process is tried and repeated in a verification environment, and the extracted optimum parallel region is obtained by a PGI compiler or the like (PGI is a registered trademark). Is specified as a parallel area by. As a result, the offload setting unit 118 performs the offload setting using the compiler. Hereinafter, a case of offloading to the GPU will be described in more detail.

GPU does not guarantee latency, but throughput can be increased by parallel processing. There are many repetitive processes in application servers and DB servers that execute applications run by users in the cloud. Therefore, the speed is increased by autonomously offloading the repeated statements of the application to the GPU.
However, this speedup requires parallelization of appropriate parts (part of application processing), and there are cases where performance is not achieved by simply parallel processing the repetitive parts. In this regard, there is an attempt to brute force the parallelism of for statements using the PGI compiler. However, brute force takes a lot of time, and in the cloud business, there is a problem that the start of use by the user is delayed.

Therefore, in the device cooperation server 1E of the present embodiment, in order to autonomously extract an appropriate offload area, the offload setting unit 118 repeats the performance test in the verification environment using GA. At that time, only the top and second level for statements are targeted for the combination of GA. The processing of the offload setting unit 118 makes it possible to search for a high-performance offload area in a short time.
This GA is one of the combinatorial optimization methods that imitates the evolutionary process of living organisms, and proceeds with the flow of initialization, evaluation, selection, crossover, mutation, and termination determination. Here, an example using the Simple GA method will be described. Simple GA is a simplified GA in which the number of genes is only 1,0, roulette selection, one-point crossover, and mutation reverse the value of one gene.

In initialization, the source code of the application to be deployed is analyzed, and the for statement and while statement of the top stage (top level) and the second stage (second level) are mapped to the gene sequence. Here, the gene length is determined from the number of repetitive processing sentences that try GPU parallel processing, and is set to 1 when parallel processing is performed by GPU and 0 when not. A predetermined number of individuals M is prepared for the gene, the value (gene value) of one gene is all 1 (all for sentences are GPU offloaded), one gene is all 0 (all CPU processing), and the rest are randomly selected. Allocate 1,0.
In the evaluation, the source code corresponding to the gene is built, deployed on the verification cloud server, and the benchmark performance is measured. Then, the goodness of fit of the gene with good performance is set high.
In selection, a gene with a high goodness of fit is selected in a predetermined number of individuals based on the goodness of fit. Here, roulette selection and elite selection of the highest goodness-of-fit gene are performed according to the goodness of fit.
In crossover, at a predetermined crossover rate Pc, selected individuals are exchanged at one point, which is a part of genes, to create offspring individuals.
In mutation, each value of an individual's gene is changed from 0 to 1 or 1 to 0 at a predetermined mutation rate Pm.
In the termination determination, the gene having the highest goodness of fit is used as the solution after repeating the treatment T times, which is a predetermined number of generations.
Then, the offload setting unit 118 deploys to the cloud server whose placement is determined by the resource securing processing unit 117 with the pattern of the source code having the highest performance corresponding to the gene of the highest goodness of fit.

By doing so, according to the device cooperation server 1E according to the modified example of the third embodiment, the cloud server of DC30 having the least delay with the network edge 20 accommodating the device most frequently used can be provided as a service. It is possible to deploy the necessary processing functions, secure the resource size, and further speed up the processing by performing offload processing on the hetro device (for example, GPU).

As described above, in the device cooperation servers 1A to 1E according to the present embodiment, it is possible to realize the optimum arrangement and processing offload of various functions in each of the device layer, the network layer, and the cloud layer. However, depending on the usage pattern of the service, for example, where it is appropriate to arrange the function may change. Therefore, the device cooperation server 1 may model the cost and performance when executing each process in each layer, and calculate where it is appropriate to place the process. The costs here are, for example, a server cost for securing resources and a network cost based on generated traffic. Performance is, for example, throughput or delay.
The device cooperation server 1 holds this cost and performance as parameters, and determines where to place the processing function and which resource is appropriate to secure the performance under the constraint condition that the cost is less than or equal to the predetermined value. By solving the combination optimization problem for the purpose of "increasing the cost" and "reducing the cost under the constraint condition that the performance exceeds the predetermined value", the function allocation and the resource securing are performed.
In addition, in order to be able to find a solution to this combinatorial optimization problem during a limited optimization period, a method of a genetic algorithm (GA) may be used.

By doing so, the device cooperation server 1 can allocate functions to the optimum layer and secure resources in consideration of all three layers (device layer, network layer, and cloud layer). As a result, the service desired by the user can be continuously provided at a reasonable price.

1,1A, 1B, 1C, 1D, 1E Device linkage server 10 Gateway 11 Control unit 12 Input / output unit 13 Storage unit 20 Network edge 30 DC (data center)
111 Location information acquisition unit 112 Device selection unit 113 Usage history collection unit 114, 114B, 114D Function placement destination determination unit 115, 115B, 115D Function distribution unit 116 Network setting unit 117 Resource reservation processing unit 118 Offload setting unit 131 Device information DB
132 Distribution function information DB

Claims (7)

It is a device cooperation server that selects resources from the three layers of cloud layer, network layer, and device layer to provide services.
A usage history collection unit that collects usage history related to the service of each device belonging to the device layer in an optimization period indicating a predetermined trial period provided to determine resources to be used when providing the service.
A function allocation destination determination unit that determines one or more of the three layers of the cloud layer, the network layer, and the device layer as the allocation destination of the functions related to the service by referring to the collected usage history.
A function distribution unit that distributes the function related to the service to one or a plurality of layers of the determined three layers, and deletes the function related to the service in the layer that is not determined as the placement destination.
A device cooperation server characterized by being equipped with. It is a device cooperation server that selects resources from the three layers of cloud layer, network layer, and device layer to provide services.
A usage history collection unit that collects usage history related to the service of each device belonging to the device layer in an optimization period indicating a predetermined trial period provided to determine resources to be used when providing the service.
A function allocation destination determination unit that refers to the collected usage history and determines the gateway of the area to which the device used during the optimization period belongs as the allocation destination of the function related to the service.
A function distribution unit that distributes the function related to the service to the determined gateway and deletes the function related to the service of the gateway that has not been determined as the placement destination.
A device cooperation server characterized by being equipped with. It is a device cooperation server that selects resources from the three layers of cloud layer, network layer, and device layer to provide services.
A usage history collection unit that collects usage history related to the service of each device belonging to the device layer in an optimization period indicating a predetermined trial period provided to determine resources to be used when providing the service.
A function allocation destination that refers to the collected usage history and determines the network edge of the network layer connected to the gateway of the area to which the device used during the optimization period belongs as the allocation destination of the function related to the service. The decision department and
A function distribution unit that distributes the function related to the service to the determined network edge and deletes the function related to the service of the network edge that has not been determined as the placement destination.
A device cooperation server characterized by being equipped with. In the network to which the determined network edge belongs, a network setting unit that executes network settings including bandwidth allocation and priority control related to the provision of the service is provided.
The device cooperation server according to claim 3, further comprising. It is a device cooperation server that selects resources from the three layers of cloud layer, network layer, and device layer to provide services.
A usage history collection unit that collects usage history related to the service of each device belonging to the device layer in an optimization period indicating a predetermined trial period provided to determine resources to be used when providing the service.
With reference to the collected usage history, the gateway in the area to which the device used during the optimization period belongs or the network edge of the network layer connected to the gateway is set as the placement destination of the function related to the service. Function placement destination decision unit to decide, and
A function distribution unit that distributes the function related to the service to the determined gateway or network edge and deletes the function related to the service at the gateway and network edge other than the determined gateway or network edge.
With reference to the collected usage history, the network edge accommodating the most frequently used devices during the optimization period is extracted, and the processing of the cloud server in the cloud with the least delay with the extracted network edge. A resource allocation processing unit that deploys the function and secures the resource size required for the processing function,
A device cooperation server characterized by being equipped with. It is a device cooperation server that selects resources from the three layers of cloud layer, network layer, and device layer to provide services.
A usage history collection unit that collects usage history related to the service of each device belonging to the device layer in an optimization period indicating a predetermined trial period provided to determine resources to be used when providing the service.
With reference to the collected usage history, the gateway in the area to which the device used during the optimization period belongs or the network edge of the network layer connected to the gateway is set as the placement destination of the function related to the service. Function placement destination decision unit to decide, and
A function distribution unit that distributes functions related to the service to the determined gateway or network edge, and
With reference to the collected usage history, the network edge accommodating the most frequently used devices during the optimization period is extracted, and the processing of the cloud server in the cloud with the least delay with the extracted network edge. It is equipped with a resource allocation processing unit that deploys the function and secures the resource size required for the processing function.
The cloud layer is composed of a hetero cloud including a hetero device of GPU and CPU.
Further provided with an offload setting unit that offloads specific processing that is a part of the processing function deployed on the cloud server to the GPU.
The offload setting unit is based on a genetic algorithm.
The source code of the application that realizes the processing function to be deployed is analyzed, the gene length is determined from the number of repetitive processing sentences that try to parallelize the GPU, and a predetermined number of genes are created.
Performance by assigning whether to perform parallel processing on the GPU or not in parallel processing on the GPU to the source code corresponding to the gene as a gene value and deploying to the verification cloud server for the statement of the iterative processing. Make measurements and
Based on the results of the performance measurement, the goodness of fit of each gene is set.
Select a predetermined number of genes with a high goodness-of-fit value, and select them.
At a predetermined crossover rate, some genes are exchanged between selected individuals at one point to create offspring individuals.
By changing each gene value of the gene of the created individual at a predetermined mutation rate,
Wherein the processing of the genetic algorithm, repeatedly a predetermined number of generations, and calculates the source code corresponding to the highest matching degree as the solution, features and to Lud vice linking server that is deployed in the cloud server. A device cooperation program for causing a computer to function as the device cooperation server according to any one of claims 1 to 6.

Images