JP7276838B2 - System, system control method, computer program used for system control, and its recording medium - Google Patents

Description

The present invention relates to an edge system using container orchestration technology, an edge system control method, a computer program used for controlling the edge system, and a recording medium thereof.

In a local environment such as a factory or construction site, sensors acquire information related to work equipment that performs predetermined work, monitor the acquired sensor information, and control the work equipment based on the sensor information. Systems that operate from the cloud are known. However, when the number of working devices and sensors increases in this type of system, the problem of increased network load and processing time delay arises as the amount of sensor information data and processing load increase. There are also concerns about information security. Therefore, an edge system that processes sensor information in a local environment, such as a factory, which is the terminal (edge) of the network, is being studied.

On the other hand, a technique for acquiring and analyzing sensor information in a local environment has been attracting attention in recent years, and is called edge computing or MEC (Mobile Edge Computing). At this time, the device arranged in the local environment may be called an edge terminal, MEC, or the like. By the way, in order to realize many functions in the edge terminal, it is necessary to operate a plurality of applications on an OS (Operating System) installed in the host PC. However, when a plurality of applications are operated at the same time, it is necessary to share the system resources of the OS between the applications. Therefore, the application design must be changed for each OS, which increases the development burden.

Therefore, a technique is known in which a logical section is created on the OS of the edge terminal, and a container application, in which the application main body and the environment necessary for the operation of the application are put together, is deployed within the section. According to this technology, resources on the OS can be logically separated and used by multiple containers. Therefore, the OS dependence of application design can be reduced, and the burden of system development can be reduced. Such container technology is realized by container management software installed on the OS. An example of container management software is Docker (Patent Document 1, for example).

In recent years, with the evolution of container technology and the demand for system redundancy, network connections, storage areas, and settings of host PCs have become more complicated, increasing the need to schedule multiple container applications. Therefore, the development of an orchestration tool that automatically configures these settings and manages container applications in an integrated manner is underway.

By using an orchestration tool, it is possible to easily deploy, execute, manage, and schedule multiple container applications, further reducing the burden of system development. Kubernetes, Apache Mesos, etc. are known as container orchestration tools.

WO2018/003020

Under the above-mentioned technical background, in a predetermined local environment such as a factory, it is being considered to configure an edge system using a plurality of edge terminals for operation of work equipment, analysis of sensor information, and the like. However, in a local environment such as in a factory, it is not always possible to use terminals having the same level of capability for each of the sensors and work equipment from the viewpoint of cost and the like. In that case, the efficiency of the system as a whole decreases.

Also, for this type of system, it is considered effective to use container orchestration technology, which is less dependent on hardware, etc., but the design theory of its architecture is still under development.

The present invention has been made in view of the above technical background, and its object is to provide a system having high processing efficiency even when configured with terminals of various capabilities. That's what it is.

The above problems can be solved by a system or the like having the following configuration.

In other words, a system according to an aspect of the present invention is a system configured to hierarchically connect and communicate with a plurality of terminals that operate containers using hardware resources logically allocated by orchestration technology. a parent terminal, which is one of the plurality of terminals; an alternative execution processing unit that causes the parent terminal to alternatively execute part or all of the processing to be performed in the child terminal when a processing state satisfies a predetermined condition.

According to one aspect of the present invention, processing states are monitored between terminals, and processing is alternatively executed when a predetermined condition is satisfied. Therefore, it is possible to provide a system with high processing efficiency even when terminals with various capabilities are used.

FIG. 1 is a block diagram showing the configuration of a monitoring system having an edge system according to the first embodiment. FIG. 2 is a block diagram of an edge system. FIG. 3 is a block diagram for explaining the functions of the edge system. FIG. 4 is a hardware configuration diagram of the MEC. FIG. 5 is a software configuration diagram of the MEC. FIG. 6 is a software configuration diagram of MEC when an orchestration tool is used. FIG. 7 is a schematic configuration diagram of a cluster. FIG. 8 is a table showing processing information. FIG. 9 is a flowchart showing resource management processing. FIG. 10 is a flowchart illustrating pod alternative execution processing at the alternative execution destination. FIG. 11 is a flowchart illustrating pod alternative execution processing in an alternative execution source. FIG. 12 is a block diagram of the edge system of the second embodiment. FIG. 13 is a block diagram for explaining the functions of the edge system. FIG. 14 is a schematic configuration diagram of an edge system of a first application example, which is an application example. FIG. 15 is a schematic configuration diagram of an edge system of a second application example, which is another application example.

A first embodiment of the present invention will be described below with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram showing the configuration of a monitoring system having an edge system according to an embodiment of the invention. As shown in this figure, in a monitoring system 100, an edge system 10 provided in a local environment 11 is connected to a WAN (Wide Area Network) 13 and can communicate with a terminal 14 and an image registry 15 via the WAN 13. is configured to

The edge system 10 is, for example, a system that monitors manufacturing processes, construction processes, and the like in a local environment 11 such as a factory or a construction site, and controls work equipment used in these processes. The edge system 10 is composed of a multifunction machine 12 that realizes a plurality of functions such as communication and control, and in this embodiment, it is composed of multifunction machines 12A to 12D. Also, the edge system 10 is connected to the WAN 13 by wireless communication or wired communication. The WAN 13 may be configured partially or wholly by a mobile communication network.

As will be described later, the MFPs 12A-12D have MECs that process sensor information acquired by sensors. Note that the MEC is an example of an edge terminal, and controls work equipment and collects sensor information by a container controlled using container orchestration technology.

In the edge system 10, a second generation second multifunction device 12B and a third multifunction device 12C, which are child devices, are connected to a first generation multifunction device 12A, which is a parent device. A third-generation fourth multifunction peripheral 12D, which is a grandchild machine, is connected to the second multifunction peripheral 12B.

In the present embodiment, an example in which the edge system 10 is configured by three generations of multifunction peripherals 12 including a parent machine, a child machine, and a grandchild machine will be described. 10 may be configured. Also, the number of multi-function machines 12 constituting the edge system 10 is arbitrary. Further, hereinafter, when there is no need to distinguish between the configurations of the first multifunction machine 12A, the second multifunction machine 12B, the third multifunction machine 12C, and the fourth multifunction machine 12D, these multifunction machines are collectively referred to as multifunction machines. 12. The same applies to other configurations, and when there are multiple identical configurations and there is no need to distinguish between them, the multiple configurations will be generically described with suffixes omitted.

Also, in the present embodiment, the image registry 15 is assumed to exist on the network via the WAN 13, but the configuration is not limited to this. Accordingly, the image registry 15 may be provided within the local environment 11, the edge system 10, and may be located within the MEC, for example.

A terminal 14 and an image registry 15 are connected to the WAN 13 . Edge system 10 , terminal 14 and image registry 15 can communicate with each other via WAN 13 .

The terminal 14 is a device including a display, a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), a memory, a network interface, etc. The terminal 14 is connected to the edge system 10 via the WAN 13 and receives monitoring data transmitted from the edge system 10. View results. The terminal 14 is, for example, an information terminal having a display unit such as a personal computer, a smart phone, and a tablet terminal.

Image registry 15 is, for example, a general-purpose data server. As will be described later, in the MEC, which is a part of the multi-function device 12 that configures the edge system 10, containers are executed using container orchestration technology, and the image registry 15 contains information about containers to be deployed in the MEC. image is stored.

The container image stored in the image registry 15 is transmitted to the edge system 10 via the WAN 13 and mobile communication network and deployed in the MEC of the multifunction machine 12 .

FIG. 2 is a diagram showing details of the edge system 10. As shown in FIG.

The multi-function device 12A has an MEC 21A that acquires and manages sensor information.

The MEC 21A is a device that includes a CPU and GPU, a memory, a network interface, etc., and is configured to be able to execute stored programs. The MEC 21A may be configured using a general-purpose computer, or may be a dedicated terminal.

In the MEC 21A, middleware for realizing container orchestration is installed on an OS (Operating System). Then, in the MEC 21A, a container that implements a predetermined function is deployed. As for the MEC 21A, the hardware configuration will be described later with reference to FIG. 4, and the software configuration will be described later with reference to FIG.

The MEC 21A mainly performs monitoring and machine learning of control information of the work machines 22B to 22D provided in the other multifunction machines 12B to 12D in the local environment 11 and sensor information acquired by the sensors 23B to 23D. The MEC 21A transmits monitoring results, learned models obtained by machine learning, and the like to the terminal 14 via wireless or wired communication. Also, as will be described later, the MEC 21A distributes resources among the other MECs 21B-21D.

The multifunction device 12B includes a working device 22B and a sensor 23B in addition to the MEC 21B. The work equipment 22B is equipment used in one or more of the manufacturing process and the construction process within the local environment 11. For example, the work equipment 22B may be a robot arm in a factory, a truck in a construction site, or other medium-sized or large equipment, or may be made of a semiconductor substrate. There are various types, including small devices such as display modules. The sensor 23B is a device that acquires information directly or indirectly related to the work device 22B, such as a camera or an infrared sensor. The sensor 23B outputs the acquired sensor information to the MEC 21B.

The MEC 21B mainly performs analysis and machine learning on sensor information acquired by the sensor 23B and control information of the work equipment 22B. MEC 21B may also control work equipment 22B autonomously or in response to instructions from other MEC 21 or terminal 14 . Then, the MEC 21B transmits the monitoring result using the sensor information, the learned model that is the product of machine learning, and the like to the first multifunction device 12A.

Like the multifunction machine 12B, the multifunction machine 12C includes a MEC 21C, a work machine 22C, and a sensor 23C. The MEC 21C controls the work machine 22C and acquires sensor information from the sensor 23C. The MFP 12D is configured similarly to the MFPs 12B and 12C. Note that the MEC 21A corresponds to the first generation base machine, the MEC 21B and 22C to the second generation slave machines, and the MEC 21D to the third generation grandchild machine, in addition to the multifunction machine 12 in the multifunction machines 12A to 12D. and cooperate with each other to perform processing.

FIG. 3 is a block diagram for explaining the functions of the edge system 10. As shown in FIG.

As shown in this figure, the edge system 10 includes an information processing system 31 indicated by a two-dot chain line and configured by MECs 21A to 21D in a local environment 11 indicated by a dotted line. That is, in the information processing system 31, information processing such as specific sensor information in the edge system 10 and trained models related to machine learning is performed.

In the information processing system 31, a first generation MEC 21A, second generation MECs 21B and 21C, and a third generation MEC 21D are hierarchically configured. The MECs 21 of each generation are configured so that information processing can be performed in cooperation with each other, and the MECs 21 of the same generation such as the MECs 21B and 21C of the second generation can perform information processing in cooperation with each other. configured to allow

Although the MECs 21A to 21D have the same basic hardware configuration, they do not need to be the same, and may differ in processing capability, storage capacity, and the like. In this embodiment, the MEC 21A is selected to have high information processing capability and large storage capacity, the MECs 21B and 21C are selected to have medium information processing capability and storage capacity, and the MEC 21D is selected to have information processing capability and storage capacity. The one with the small storage capacity is selected.

FIG. 4 is a hardware configuration diagram of the MEC 21. As shown in FIG.

The MEC 21 is composed of a control unit 41 composed of a CPU, a GPU, etc. for overall control, a ROM (Read Only Memory), a RAM (Random Access Memory), a hard disk or a storage, etc., and stores programs, various data, and the like. an input/output port 43 for inputting/outputting data to/from an external device; a communication unit 44 for communicating with another MEC 21; It has a display unit 45 and an input unit 46 that receives input from the outside. The control unit 41, the storage unit 42, the input/output port 43, the communication unit 44, the display unit 45, and the input unit 46 are configured to communicate with each other through bus connection. Note that this hardware configuration may be realized by a hardware circuit such as an FPGA.

Note that the MECs 21A to 21D do not have the same hardware configuration as described above. The MEC 21A has a high information processing capability of the control unit 41 and a large capacity of the storage unit 42, and the MECs 21B and 22C have an intermediate information processing capability of the control unit 41 and a medium capacity of the storage unit 42. The MEC 21D has a low information processing capability of the control unit 41 and a small capacity of the storage unit 42 .

FIG. 5 is a software configuration diagram of the MEC 21. As shown in FIG.

The MEC 21 has an operating system (OS) 52 installed on hardware 51 . In addition to general-purpose middleware 53 , a container engine 54 and an orchestration tool 55 operating together with the container engine 54 are installed in the operating system 52 .

In the MEC 21 , containers 56 are deployed and executed by container engines 54 and orchestration tools 55 . The container 56 includes middleware (MW) 58 that conforms to the operation specifications of the container engine 54 in addition to an application (APL) 57 that implements a predetermined function. Note that the middleware 58 may include a library and the like.

The hardware 51 has the hardware configuration shown in FIG. Using these hardware 51 resources, the MEC 21 can perform predetermined operations.

The operating system 52 is a basic software configuration system in the MEC 21 . The operation system 52 controls the overall operation of the MEC 21 .

The general-purpose middleware 53 is generally a functional block provided by the vendor of the operating system 52, etc., and performs the communication functions of the communication unit 44 shown in FIG. It is a functional block for realizing basic operations such as

The container engine 54 is one of middleware installed in the operating system 52 and is an engine that operates the container 56 . Specifically, the container engine 54 allocates the resources of the hardware 51 and the operating system 52 to the container 56 based on the configuration files and the like included in the middleware 58 within the container 56 .

The orchestration tool 55 is a functional block that causes the container engine 54 to allocate resources such as the hardware 51 . An orchestration tool 55 organizes one or more containers 56 into units called pods (not shown in FIG. 5), and each pod is assigned to a node (not shown in FIG. 5), which is a logically different area. Deployed. Details of the operation of the orchestration tool 55 will be described later with reference to FIG.

The container 56 includes middleware 58 such as a library as well as an application 57 that implements a predetermined function. The container 56 operates using the hardware 51 and operating system 52 resources allocated by the container engine 54 . Allocation of resources to the container 56 by the container engine 54 is performed based on a setting file or the like included in the middleware 58 in the container 56 . In this way, resource management is guaranteed by the container engine 54, thus making the operation of the container 56 less environment dependent.

FIG. 6 is a schematic configuration diagram of the MEC 21 that serves as an operating environment for the container 56 when the orchestration tool 55 is used. As described above, the hardware configuration of the MEC 21 is different for each of the MECs 21A to 21D, so the number of containers 56 and the like shown in this figure is an example.

Orchestration tool 55 manages the hardware resources allocated by container engine 54 . A logical space managed by the orchestration tool 55 is called a cluster. In this embodiment, the orchestration tool 55 forms a cluster using the hardware resources of the MECs 21A-21D.

The orchestration tool 55 manages the execution environment of the container 56 in units called nodes 61 . At the same time, the orchestration tool 55 provides a master 62 that manages the overall operation of the nodes 61 .

In the node 61, two pods 611 are deployed in the example of this figure. A pod 611 is a functional block that implements a predetermined service made up of a plurality of containers 56 , and in the example of this figure, the pod 611 has two containers 56 . A pod 611 is a unit for managing the container 56 by the orchestration tool 55 . The overall operation of pod 611 within node 61 is controlled by pod management library 612 .

The pod management library 612 includes a container runtime 6121 for allowing the pods 611 (containers 56) to use logically allocated hardware resources, an agent 6122 for receiving control from the master 62, It has a network (NW) proxy 6123 or the like that performs communication with the master 62 or the like. By means of the pod management library 612 having such a configuration, the pods 611 communicate with the pods 611 within the same node 61, the pods 611 of other nodes 61, and the like, and use hardware resources to perform predetermined operations. Realize the function.

The master 62 includes an application server 621 that deploys the pods 611, a manager 622 that manages the deployment status of the containers 56 by the application server 621, a scheduler 623 that determines which node 61 the container 56 is to be placed on, and data distribution. A data sharing unit 624 for sharing is included. Note that the master 62 can delete and deploy the pods 611 so that the processing load on each node 61 is constant, so high maintainability can be achieved.

FIG. 7 is a schematic configuration diagram of a cluster composed of MECs 21A to 21D.

As described above, each of the MECs 21A-21D has a container engine 54 (not shown in FIG. 7) and an orchestration tool 55 (not shown in FIG. 7) installed. Therefore, one cluster is formed using the hardware resources of the MECs 21A to 21D. In the cluster, one master 62 and a plurality of nodes 61 are configured, and the nodes 61 execute pods 611 composed of a plurality of containers 56 (not shown in FIG. 7). The MEC 21A is a first generation parent device, the MEC 21B and 21C are second generation child devices, and the MEC 21D is a third generation grandchild device.

The MEC 21A is provided with a master 62 and one node 61A.

In the master 62, resource management is performed by functions implemented by the orchestration tool 55. FIG. Such resource management using the orchestration tool 55 is performed according to the control status of each pod 611 by the master 62 and the like. In general, resource management in a plurality of terminals when container orchestration technology is used is performed by middleware that provides orchestration technology. However, resource management in middleware does not fully consider the actual processing state in the terminal. In this embodiment, as will be described later, monitoring and substitution processing are performed by the management pod 6111, so more appropriate resource management is realized.

The node 61A is provided with one management pod 6111 and a processing pod 6112A that implements predetermined functions.

The management pod 6111 manages the resources of the processing pod 6112 according to the processing load status of the MECs 21A-21D. Here, the processing load includes, for example, the processing load on the control unit (CPU, GPU, etc.) of each MEC 21, the capacity of the storage unit (various memories, storages, etc.), or the load on the connected network. Resource management by the management pod 6111 differs from resource management using orchestration technology by the master 62, and is performed according to the information processing conditions directly observed in each of the MECs 21. FIG. It should be noted that the status of information processing observed in the MEC 21 is summarized as processing information.

Specifically, each of the MECs 21B to 21D other than the MEC 21A is provided with one monitoring pod 6113, and the management pod 6111 receives processing information from the monitoring pod 6113 of the other MEC 21. Resource management is performed by causing another processing pod 6112 to alternatively perform the processing of . The resource management will be explained later in detail with reference to FIG. Moreover, the pods to be managed by the management pod 6111 may be arranged in the same MEC 21A.

Predetermined containers are deployed in the first processing pod 6112A in order to implement predetermined functions such as analysis of sensor information acquired by the multifunction machines 12B-12D in the MECs 21B-21D and machine learning.

In the MEC 21B, the hardware processing power is medium, three nodes 61B are configured, and two pods 611 are deployed in each node 61B, so a total of six pods 611 are deployed in the MEC 21B. . Specifically, one node 61B is provided with a monitoring pod 6113B and a first processing pod 6112B, and the other two nodes 61B are further provided with second to fifth processing pods 6112B.

The monitoring pod 6113B transmits, to the management pod 6111 of the MEC 21A, the load state of the MEC 21B in which it is provided and the processing information indicating the information processing status of the first to fifth processing pods 6112B in the MEC 21B. . Processing information transmitted by the monitoring pod 6113B will be described later with reference to FIG.

In the MEC 21C, the hardware processing capacity is moderate, but in order to ensure redundancy, two nodes 61C, which are fewer than the MEC 21B, are configured, and two processing pods 6112B are deployed in each node 61B. be. One node 61C is provided with a monitoring pod 6113B and a first processing pod 6112C, and the other node 61C is provided with second and third processing pods 6112C.

Since the MEC 21D has a relatively low hardware processing capability, one node 61D is configured, and the node 61D is provided with a monitoring pod 6113D and a first processing pod 6112D.

FIG. 8 is a table showing processing information of MEC 21B-21D received by management pod 6111 from monitoring pods 6113B-6113C. According to this table, the processing information is composed of the processing load of the hardware of each MEC 21 and the state of whether or not the first to fifth processing pods 6112B can be replaced.

For example, in the MEC 21B, the first and second processing pods 6112B directly affect the current or immediately following process of the work equipment 22B, so they cannot be replaced in the MECs 21A, 22C, and 22D that are different from themselves (alternative not possible). Further, since the third to fifth processing pods 6112B perform analysis of sensor information acquired by the sensor 23B and machine learning related to the control of the work equipment 22B, the MEC 21A, which is the master device, executes the processing instead. It is determined to be possible (substitutable by MEC21A).

Incidentally, as shown in the lower part of the drawing, regarding the MEC 21D, the first processing pod 6112D is "substituteable by the MECs 21A and 22B". This indicates that since the MEC 21D is the third generation, it can alternatively be executed in either the first generation MEC 21A or the second generation MEC 21B.

In this way, the monitoring pods 6113B to 6113D provided in the MECs 21B to 21D generate processing information including the processing load of the MEC 21 and information indicating whether the processing pod 6112 can be alternatively executed in another MEC 21. . Then, the management pod 6111 of the MEC 21A alternately executes the processing pod 6112 of the MECs 21B-21C in another MEC 21 according to the processing information transmitted from the monitoring pods 6113B-6113D of the MECs 21B-22C. Allocate resources.

FIG. 9 is a flowchart showing resource management processing performed by the management pod 6111. As shown in FIG.

In step S91, the management pod 6111 receives the processing information (FIG. 8) of the MECs 21B-21D from the monitoring pods 6113B-6113D of the MECs 21B-21D. The management pod 6111 then performs the process of step S92.

In step S92, the management pod 6111 determines whether or not there is an MEC 21 whose processing load exceeds a predetermined threshold among the MECs 21B to 21D, based on the processing information. If there is an MEC 21 whose processing load exceeds the predetermined load threshold (S92: Yes), the management pod 6111 next performs the process of step S93. If there is no MEC 21 whose processing load exceeds the load threshold (S92: No), the management pod 6111 terminates resource management processing.

It should be noted that the load threshold used for the determination in step S92 is assumed to be lower than the standard used for resource management performed in the master 62 using container orchestration technology. By doing so, resource management can be performed according to the actual processing load of the MECs 21B to 21D.

In step S93, based on the processing information, the management pod 6111 confirms whether or not the processing pod 6112 running in the MEC 21 whose processing load exceeds the load threshold can be replaced.

Then, the management pod 6111 determines whether or not there is a processing pod 6112 that can be alternatively executed in another MEC 21 among the processing pods 6112 . If there is a processing pod 6112 that can be alternatively executed in another MEC 21 (S93: Yes), the management pod 6111 then performs the processing of step S94. If all the processing pods 6112 are "non-substitutable" and there are no processing pods 6112 that can be alternatively executed in another MEC 21 (S93: No), the management pod 6111 terminates resource management processing.

In step S94, the management pod 6111 deletes the processing pod 6112 that can alternatively be executed in another MEC 21, and processes the program that realizes the processing pod 6112 that can execute the same processing as the deleted processing pod 6112. Alternate execution is performed in the node 61 of the alternate destination MEC 21 indicated in the information. Instead of deleting the processing pod 6112, the processing pod 6112 may be restarted and put into a predetermined standby state.

FIG. 10 is a flow chart showing alternative execution processing of the processing pod 6112 in step S94 of FIG. Note that the alternative execution of the processing pod 6112 is mainly performed by the management pod 6111 .

In step S101, the management pod 6111 in the MEC 21A, which is the parent device, performs initial settings to start the alternative execution process of the processing pod 6112A.

In step S102, the management pod 6111 instructs the processing pod 6112B of the MEC 21B, which is a child device, to perform alternative execution. It should be noted that, in this alternate execution instruction, the processing pod 6112B that performs the alternate execution and the MEC 21A that performs the alternate execution are indicated.

In step S103, when the management pod 6111 receives the processing-in-progress data of the processing pod 6112B transmitted from the MEC 21B in response to the alternate execution instruction, it stores the processing-in-progress data in the processing pod 6112A.

In step S104, the management pod 6111 causes the processing pod 6112A to start processing using the received in-process data in the MEC 21B, which is the child device. In this manner, alternate execution in the processing pod 6112A in the master MEC 21A is performed.

FIG. 11 is a flow chart showing the alternative execution processing of the processing pod 6112B in the MEC 21B, which is the child device that is the alternative execution source, in step S94 of FIG.

In step S111, the monitoring pod 6113B in the MEC 21B, which is a child device, receives an alternative execution instruction from the MEC 21A, which is a parent device, to the processing pod 6112B of the MEC 21B, which is a child device, as shown in step S102 in FIG. do. As described above, the alternative execution instruction indicates the processing pod 6112B that performs the alternative execution and the MEC 21A that is made alternatively executable.

In step S112, the monitoring pod 6113B stops the processing of the processing pod 6112B, which is the target of alternative execution, in response to the alternative execution instruction. Send. Note that when the MEC 21A receives this in-process data, the process of step S103 in FIG. 10 is performed.

In this way, when the processing load of the MEC 21B, which is a slave device, is high, after performing data linkage to transmit the data being processed by the processing pod 6112B to the MEC 21A, which is the parent device, the processing by the processing pod 6112B is It can be executed by processing pod 6112A of MEC 21A. In this way, resource management can be appropriately performed between the MEC 21A, which is the parent device, and the MEC 21B, which is the child device.

In the above example, the management pod 6111 of the MEC 21A alternatively executes the processing pod 6112B of the MEC 21B itself (MEC 21A), but the present invention is not limited to this. For example, it may be executed by the node 61C of the MEC 21C. By configuring in this way, resource management can be appropriately performed not only between the master device and the slave device, but also between the slave devices and between the slave device and the grandchild device. degree of freedom can be improved.

According to this embodiment, the following effects can be obtained.

In the edge system 10 of the present embodiment, the MEC 21A of the first MFP 12A, which is the parent device, has higher processing capability than the MEC 21B of the second MFP 12B, which is the child device. In such a case, the MEC 21B, which has a relatively low processing capacity, may have a high processing load and may overflow. When the control using the container orchestration technology is performed, this resource management is performed on the master 62 side, but it is performed only based on the information managed by the master 62, and the actual information processing of the MEC 21 is performed. may not be in accordance with the state of

On the other hand, in this embodiment, the MEC 21B, which is a child device, monitors the processing state of the MEC 21B in which it is arranged by the monitoring pod 6113B, and sends the processing information, which is the monitoring result, to the MEC 21A, which is a parent device. Send to pod 6111 . The MEC 21A uses the management pod 6111 to determine a predetermined condition regarding the information processing load state of the MEC 21B based on the received processing information. Then, when MEC 21B satisfies a predetermined condition, MEC 21A performs data linkage to transmit the data being processed by processing pod 6112B of MEC 21B to processing pod 6112B of MEC 21B, and is scheduled to be performed in MEC 21B. The resource management between the MEC 21A, which is the parent device, and the MEC 21B, which is the child device, is performed by alternatively executing part or all of the processing described above in the MEC 21A.

With this configuration, when it is difficult for the MEC 21B to process information, a part of the processing of the processing pod 6112B of the MEC 21B can be alternatively executed in the MEC 21A, so resource management can be appropriately performed. Therefore, the processing of the MEC 21B can be performed smoothly. Therefore, the need to use a hardware configuration with high processing capability for all the MECs 21 is reduced, the degree of freedom in designing the edge system 10 is improved, and the hardware configuration of some of the MECs 21B can be simplified. Cost can be reduced.

Further, in the container orchestration technology, resource management is performed based on the operation status of the processing pods 6112 grasped by the master 62 side. On the other hand, in this embodiment, the resource management performed by the management pod 6111 is performed according to the processing load of each MEC 21 and the actual processing state such as whether or not the processing pod 6112 can be alternatively executed. It is possible to perform appropriate resource management suitable for the actual operating state of the edge system 10 .

Furthermore, according to the edge system 10 of this embodiment, the processing information generated by the monitoring pod 6113B, which is a child device, includes the processing load of the MEC 21B. Therefore, the MEC 21A, which is the master device, uses the management pod 6111 to determine whether or not it is difficult for the MEC 21B to process information based on the processing load of the MEC 21B indicated in the processing information. When the MEC 21A determines that it is difficult for the MEC 21B to process information, the MEC 21A performs resource management by alternatively executing part or all of the processing pod 6112B of the MEC 21B. By configuring in this way, it is possible to determine whether or not it is difficult for the MEC 21B to process information according to the state in which the MEC 21B is actually performing information processing. can do

Furthermore, according to the edge system 10 of the present embodiment, the processing information generated by the monitoring pod 6113B, which is a child device, includes processing information indicating whether the processing pod 6112B can alternatively be executed in the MEC 21A. Therefore, when the MEC 21A, which is the parent device, determines that it is difficult for the MEC 21B to process information based on the management pod 6111, the MEC 21B can be replaced based on the information indicating whether the processing pod 6112B can be replaced, which is indicated in the processing information. Some or all of the processing pod 6112B is alternatively executed in the MEC 21A, which may be an alternative destination.

Among the processing pods 6112 running on the MEC 21B, there are those that must be run on the MEC 21B and those that can be run on another MEC 21 instead. Therefore, by alternatively executing the alternatively executable processing pod 6112B in the MEC 21A, resource management can be performed without affecting the processing of the processing pod 6112. FIG.

Furthermore, according to the edge system 10 of the present embodiment, the MEC 21B, which is a child device, uses the monitoring pod 6113B to determine whether the processing pod 6112 can alternatively be executed in the MEC 21A according to the operating status of the processing pod 6112B. Generate the processing information shown. For example, if the processing pod 6112B performs processing that is not directly related to the current work process of the work equipment 22B, it is determined that the other MEC 21 can alternatively execute the processing. By configuring in this way, the resource management of the edge system 10 can be performed while the operating status of each processing pod 6112B is appropriately considered.

Furthermore, according to the edge system 10 of this embodiment, the management pod 6111 of the MEC 21A may alternatively execute some or all of the processing pods 6112D of the MEC 21D in the MEC 21C. In this way, the alternative execution of the processing pod 6112D can be performed in any MEC 21 that constitutes the edge system 10, so that the degree of freedom in design can be improved.

(Second embodiment)
The second embodiment is similar to the first embodiment in that data linkage and execution of alternative processing are performed between the MECs 21 . However, in the second embodiment, the multi-function device 12 further has a base station 121 that can be used for a mobile communication network or the like. That is, an example in which each MEC 21 communicates with each other via the base station 121 will be described. Note that the same or substantially the same configuration as in the first embodiment is assigned the same reference numerals.

FIG. 12 is a diagram showing details of the edge system 10. As shown in FIG.

In addition to the MEC 21A, the MFP 12A further has a base station 121A capable of communicating with a mobile communication network. The MFP 12A has a base station 121A capable of communicating with a mobile communication network and an MEC 21A that acquires and manages sensor information. It is configured. The base station 121A is relatively large and can control the entire mobile communication network in the local environment 11. FIG.

The MFP 12B has a base station 121B in addition to the MEC 21B, work equipment 22B, and sensor 23B. Similarly, the MFPs 12C and 12D further include base stations 121C and 121D, respectively. It is assumed that the base stations 121B and 121C are medium-sized, and the base station 121D is small-sized.

Here, each of the base stations 121 of the MFP 12 can use a signal of a relatively high frequency band and perform communication (beamforming) with directivity using a plurality of antenna elements. can. Therefore, in the local environment 11, a system is configured in which mobile communication using beamforming between the base stations 121 is possible. Since the base station 121 performs communication by beamforming, the directivity to the communication target is increased, so that a low-delay and large-capacity network can be configured between the multifunction devices 12 . Note that normal wireless communication may be performed via a LAN or the like without using beamforming.

FIG. 13 is a block diagram for explaining the functions of the edge system 10. As shown in FIG.

As shown in this figure, the edge system 10 according to the present embodiment is indicated by a two-dot chain line, and in addition to an information processing system 31 composed of MECs 21A to 21D, is indicated by a one-dot chain line, and base stations 121A to 121D. A communication processing system 131 configured by a mobile communication network including 121D is provided. Communication processing of specific information such as sensor information and learned models in the information processing system 31 is mainly performed via a mobile communication network.

The communication processing system 131 is hierarchically configured with a first generation base station 121A, second generation base stations 121B and 121C, and a third generation base station 121D. The first-generation base station 121A is positioned at the highest level in the edge system 10 and is capable of high-capacity communication.

Therefore, the base station 121A is selected to have a relatively large size and large capacity. The second generation base stations 121B and 121C and the third generation base station 121D are integrated with the work machines 22B, 22C and 22D, respectively, to operate the multifunction machines 12B, 12C and 12D. Configure. If the work machines 22B and 22C are large machines such as trucks, the base stations 121B and 121C are selected to have medium processing power and medium size. If the work device 22D is a small device such as a drone, the base station 121D is selected to have a low processing capacity but a small size.

According to the edge system 10 of the second embodiment configured as described above, since mutual communication is possible via the base station 121 provided in each multifunction machine 12, low-delay and large-capacity communication is realized. This enables more reliable data linkage. Therefore, alternative execution of processing between each MEC 21 is performed with higher reliability.

In the above-described embodiment, the parent-child relationship has been described as fixed. However, the invention is not limited to such a configuration. Therefore, for example, the parent-child relationship may be dynamically changed by re-deploying. At this time, it is not always required that the ability of the parent MEC 21 is higher than that of the child MEC 21 .

Further, in the above embodiment, a multi-function machine including the MEC 21, the work equipment 22, and various sensors 23 has been described. However, the invention is not limited to such configurations. Therefore, for example, the system may be configured with only the MEC 21, or the system may be configured using a multi-function machine comprising an arbitrary combination of the MEC 21 and other devices (base station 121, work equipment 22, various sensors 23, etc.). You may

(First application example)
In the first application example, an application example of the edge system 10 to a factory will be described.

FIG. 14 is a schematic configuration diagram of an application example in which the edge system 10 is applied to a forklift system automatically operated in a factory. In this example, the local environment 11 corresponds to the inside of a factory.

The edge system 10 has a hierarchical structure. Under one parent-generation multifunction device 12A, two child-generation multifunction devices 12B (MEC 21B) and 12C are connected. Four grandchild multifunction devices 12D and 12E and two work devices 22F and 22G are connected below. A container orchestration tool is installed in each MEC 21 .

The MFP 12A has a relatively large base station 121A and a parent generation MEC 21A, and mainly performs communication with the outside of the system and high-order control.

The child generation multifunction machine 12B (MEC 21B) is connected to the MEC 21A of the multifunction machine 12A, grandchild generation MECs 21D and 21E, and grandchild generation forklifts 22F and 22G. The forklifts 22F and 22G are controlled by the MEC 21B and move by automatic operation. Note that the MEC 21B monitors the load states of the grandchild MECs 21D and 21E as in the above-described embodiment.

The multifunction device 12C corresponds to a forklift that moves by automatic operation. The multi-function device 12C includes a relatively small base station 121C, a child-generation MEC 21C, a forklift as a work machine 22C, and various sensors 23C used to detect movement and posture.

The MFPs 12D and 12E are provided with cameras 23D and 23E for monitoring the inside of the factory, and grandchild generation MECs 21D and 21E for processing information acquired from the cameras 23D and 23E.

Also, at this time, the information processing capability of the MEC 21 varies, but the higher the generation, the higher the information processing capability of the MEC 21 is used.

In such a state, when the edge system 10 operates and, for example, the processing load of the grandchild generation MEC 21D increases and exceeds a predetermined threshold, the child generation MFP 12B (MEC 21B) confirms whether or not the processing can be performed. , the processing that was originally scheduled to be performed by the MEC 21D of the grandchild generation is performed instead.

With such a configuration, it is possible to provide an edge system for a factory with high efficiency even when composed of MECs 21 of various capacities.

(Second application example)
In the second application example, an application example of the edge system 10 to a factory will be described.

FIG. 15 is a schematic configuration diagram of an application example in which the edge system 10 is applied to a factory. In this example, the local environment 11 corresponds to a factory.

In this example, the multifunction machines 12A and 12B are assumed to have a fixed base station 121. FIG. Moreover, the multi-function machines 12C and 12E do not have the working devices 22C and 22E, respectively, and only the sensors 23C and 23D function as surveillance cameras. The multi-function devices 12D and 12F are drones, respectively, and have sensors 23D and 23F, which are multifunctional sensors.

As shown in this example, MFPs 12A and 12B are first generation, and MFPs 12C-12E are second generation. The multifunction machine 12F is connected to the second generation multifunction machine 12B and the third generation multifunction machine 12D. A container orchestration tool is installed in each multi-function device 12 .

Here, the MFPs 12D and 12F collect sensor information while the working devices 22D and 22F, which are drones, move, so the amount of sensor information data and the processing load for analysis increase. However, since the MECs 12D and 12F need to be lightweight, it is necessary to use small MECs 21D and 21F with low processing capability. Therefore, the processing of the processing pods 6112D and 6112F being executed in the MECs 21D and 21F is alternatively executed by other MECs 21, whereby resource management is performed and the edge system 10 can be stably operated.

Although the embodiments of the present invention have been described above, the above embodiments merely show a part of application examples of the present invention, and the technical scope of the present invention is not limited to the specific configurations of the above embodiments. do not have.

10 edge system 11 local environment 12 compound machine 15 image registry 21 MEC (edge terminal)
22 working equipment 23 sensor 61 node 62 master 611 pod 6111 management pod 6112 processing pod 6113 monitoring pod 121 base station 100 monitoring system

Claims (12)

A plurality of terminals that operate containers using hardware resources logically allocated by orchestration technology are hierarchically connected and configured to be communicable , and generated across the plurality of terminals by the orchestration technology A system that can manage the execution environment of a container that
A parent terminal, which is one of the plurality of terminals,
a processing state monitoring unit that monitors a processing state of a child terminal that is placed below and connected to the parent terminal and that is realized by a container;
When the processing state on the child terminal satisfies a predetermined condition, the parent terminal alternatively executes part or all of the processing to be performed on the child terminal, and the alternative execution processing unit is realized by a container. and a system comprising: The processing state monitoring unit
A load state monitoring unit for monitoring the processing load state of the child terminal,
2. The system according to claim 1, wherein said predetermined condition is that the processing load of said child terminal exceeds a predetermined threshold. 3. The system according to claim 2, wherein said processing load state is an operating rate of hardware of said child terminal. The child terminal further
2. The system according to claim 1, further comprising a container control unit that, when the parent terminal executes an alternative process, controls an operation of a container related to the alternatively executed process. The child terminal further
2. The system according to claim 1, further comprising a data transmission unit that transmits data necessary for alternative processing by said parent terminal to said parent terminal. The parent terminal further
2. The system according to claim 1, further comprising an information holding unit that holds information about whether or not said parent terminal can alternatively execute a process that should be performed by said child terminal. The child terminal further comprises a sensor,
The child terminal is
The system according to claim 1, comprising a sensor information processing section that processes information acquired from the sensor. The child terminal further includes a work device,
The child terminal is
2. The system according to claim 1, comprising a work equipment control unit that controls the work equipment. The parent terminal further comprises a first base station,
The child terminal further comprises a second base station,
The system according to claim 1, wherein communication between said parent terminal and said child terminal is performed via said first base station and said second base station. A plurality of terminals that operate containers using hardware resources logically allocated by orchestration technology are hierarchically connected and configured to be communicable , and generated across the plurality of terminals by the orchestration technology A method of controlling a system capable of managing the execution environment of a container , comprising:
A parent terminal, which is one of the plurality of terminals,
a processing status monitoring step of monitoring a processing status of a child terminal arranged below and connected to the parent terminal;
an alternative execution processing step of alternatively executing part or all of the processing to be performed in the child terminal in the parent terminal when the processing state in the child terminal satisfies a predetermined condition;
A system control method, wherein an application that executes the processing state monitoring step and the alternative execution processing step is realized by the container. A plurality of terminals that operate containers using hardware resources logically allocated by orchestration technology are hierarchically connected and configured to be communicable , and generated across the plurality of terminals by the orchestration technology A computer program used to control a system capable of managing the execution environment of a container that
to the parent terminal, which is one of the plurality of terminals,
a processing status monitoring step of monitoring a processing status of a child terminal arranged below and connected to the parent terminal;
an alternative execution processing step of alternatively executing part or all of the processing to be performed in the child terminal in the parent terminal when the processing state in the child terminal satisfies a predetermined condition;
A computer program, wherein an application that executes the processing state monitoring step and the alternative execution processing step is realized by the container. A plurality of terminals that operate containers using hardware resources logically allocated by orchestration technology are hierarchically connected and configured to be communicable , and generated across the plurality of terminals by the orchestration technology A recording medium storing a computer program used for controlling a system capable of managing the execution environment of a container ,
to the parent terminal, which is one of the plurality of terminals,
a processing status monitoring step of monitoring a processing status of a child terminal arranged below and connected to the parent terminal;
an alternative execution processing step of alternatively executing part or all of the processing to be performed in the child terminal in the parent terminal when the processing state in the child terminal satisfies a predetermined condition;
A recording medium storing a computer program, wherein an application that executes the processing state monitoring step and the alternative execution processing step is implemented by the container.

Images