Resource control device, resource control method, and resource control program

Abstract

A resource control device (1) comprises: a prediction unit (12) that predicts a resource amount which can be used by an AI application after the elapse of a prescribed time; and a notification unit (14) that, if the predicted resource amount which can be used by the AI application is less than a resource amount which is used in processing requested by a terminal which uses the AI application, notifies the terminal in advance that the resource amount which is used by the AI application after the elapse of the prescribed time becomes insufficient.

Description

Resource Control Device, Resource Control Method, and Resource Control Program

[0001] The present disclosure relates to a resource control device, a resource control method, and a resource control program.

[0002] In a wireless communication system, a technique for predicting the amount of resources used by an application and auto-scaling the application according to the load is known.

[0003] Google Cloud, "Scaling Based on Prediction", [online], [searched on January 10, 2025], Internet <URL:https: / / cloud.google.com / compute / docs / autoscaler / predictive-autoscaling?hl=ja>

[0004] A resource control device according to an aspect of the present disclosure includes a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources available for an AI (Artificial Intelligence) application after a predetermined time has elapsed, and a notification unit that notifies the terminal in advance that the amount of resources used by the AI application will be insufficient after the predetermined time has elapsed when the predicted amount of resources available for the AI application is less than the amount of resources used in the process requested from the terminal using the AI application.

[0005] A resource control method according to an aspect of the present disclosure includes a prediction step of referring to metrics of a wireless communication network and predicting the amount of resources available for an AI application after a predetermined time has elapsed, and a notification step of notifying the terminal in advance that the amount of resources used by the AI application will be insufficient after the predetermined time has elapsed when the predicted amount of resources available for the AI application is less than the amount of resources used in the process requested from the terminal using the AI application.

[0006] A resource control program according to one embodiment of the present disclosure is a program for causing a computer to function as a resource control device, wherein the computer functions as a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources that an AI application can use after a predetermined time has elapsed, and a notification unit that, if the predicted amount of resources that the AI ​​application can use is less than the amount of resources used in processing requested by a terminal using the AI ​​application, notifies the terminal in advance that the amount of resources that the AI ​​application can use after the predetermined time has elapsed will be insufficient.

[0007] This figure shows an overview of the wireless communication system related to this disclosure. This block diagram shows the configuration of the resource control device related to this disclosure. This table shows an example of data stored in the storage unit related to this disclosure. This is an example of a table referenced by the notification unit related to this disclosure. This is a flowchart showing the processing flow executed by the resource control device related to this disclosure. This figure shows an example of processing executed by the resource control device related to this disclosure. This block diagram shows the configuration of the resource control device related to this disclosure. This is a flowchart showing the processing flow executed by the resource control device related to this disclosure.

[0008] <Embodiment 1> Hereinafter, an embodiment of the present disclosure will be described in detail with reference to the drawings. For ease of understanding, the background and problems of the present disclosure will be explained first, followed by a detailed explanation of the present disclosure.

[0009] <Auto Scaling> Auto scaling, an example of RIC (Radio Access Network (RAN) Intelligent Controller), predicts the amount of resources an application will use and performs actions such as scaling in or stopping the application. However, auto scaling may scale in or stop the application without prior notification to the application's users.

[0010] As an example, we will explain the case where autoscaling occurs in a system that integrates AI (Artificial Intelligence) applications and RAN (Wireless Network) applications on the same platform (hereinafter also referred to as "AI-RAN"). AI-RAN is a technology that utilizes the computing infrastructure of a base station to provide highly immediate services with low latency to users and devices around the base station. In AI-RAN, AI applications and machine learning applications are deployed at the network edge (terminal side) via the RAN. This configuration promotes the creation of new industries and solutions that leverage low latency and confidentiality.

[0011] In AI-RAN, applications for RAN are given high priority, while AI applications are given lower priority. As a result, resources used by RAN applications are prioritized, and if resources become insufficient, the AI ​​applications may be stopped by autoscaling. In this case, users of the AI ​​applications may not receive a notification that the AI ​​applications will stop just before they are stopped, or they may stop without any notification, which results in a poor user experience (UX).

[0012] Therefore, in AI-RAN, there is a need for technology that controls the amount of resources used while minimizing the possibility of a decrease in user experience (UX).

[0013] <Overview of Wireless Communication System 100> An overview of the wireless communication system 100 will be described with reference to Figure 1. Figure 1 is a diagram illustrating the overview of the wireless communication system 100. As shown in Figure 1, the wireless communication system 100 includes at least one terminal TE, at least one base station BS, and a core network NW.

[0014] The wireless communication system 100 is a wireless communication system that communicates using AI-RAN. In other words, in the core network NW of the wireless communication system 100, applications for RAN and AI applications are integrated on the same platform.

[0015] In the wireless communication system 100, the terminal TE communicates with the core network NW via the base station BS. For example, when the terminal TE executes an AI application, the base station BS notifies the core network NW of the request to execute the input process for the AI ​​application. The core network NW executes the requested AI application process, and the execution result is notified to the terminal TE via the base station BS.

[0016] The core network NW includes a container, a Pod containing at least one container, a Node running the Pod, a Cluster containing at least one Node, and a resource control device 1.

[0017] A container is a collection of applications, middleware, and other components.

[0018] A Pod is a managed unit and deployable unit of a container. A Pod groups at least one container and provides a shared execution environment.

[0019] Node runs a Pod and executes the containerized applications contained within that Pod.

[0020] A cluster is a combination of multiple nodes.

[0021] As an example, in the core network NW, the RAN application is made into a container, and a Pod (Pod) is deployed that groups the containers. As another example, when a terminal TE (Technical Engine) notifies the core network NW to request the execution of a process input to the AI ​​application, the core network NW makes the AI ​​application into a container, and a Pod (Pod) is deployed that groups the containers. As yet another example, when a terminal TE notifies the core network NW to request the execution of a process input to the AI ​​application, the processing request is propagated to the deployed Pod in the core network NW, and the processing is executed within the Pod.

[0022] The resource control device 1 controls the amount of resources used in the core network NW. As described above, in AI-RAN, applications for RAN have a high priority, while AI applications have a low priority. In other words, the resource control device 1 prioritizes securing resources used by applications for RAN. If the resource control device 1 predicts that the amount of resources used by AI applications will be insufficient, it notifies the terminal TE in advance to reduce the amount of resources used by the AI ​​applications.

[0023] More specifically, the resource control device 1 includes a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources available to the AI ​​(Artificial Intelligence) application after a predetermined time has elapsed, and a notification unit that, if the predicted amount of resources available to the AI ​​application is less than the amount of resources used in processing requested by the terminal using the AI ​​application, notifies the terminal in advance that the amount of resources to be used by the AI ​​application after the predetermined time has elapsed will be insufficient.

[0024] According to this configuration, the resource control device 1 refers to the amount of resources used by the RAN application and predicts the amount of resources available to the AI ​​application after a predetermined time has elapsed. If the amount of resources available to the AI ​​application after the predetermined time has elapsed is less than the amount of resources used in the processing requested by the terminal using the AI ​​application, the resource control device 1 notifies the terminal in advance to reduce the amount of resources used in using the AI ​​application after the predetermined time has elapsed.

[0025] In other words, for example, in the AI-RAN described above, the resource control device 1 first predicts the amount of resources that low-priority AI applications will be able to use after a predetermined time has elapsed, based on the RAN metrics. Therefore, the resource control device 1 can appropriately control the amount of resources used.

[0026] Next, if the resource control device 1 predicts that the resources allocated to the AI ​​application will become insufficient after a predetermined time, it notifies the terminal in advance to reduce the amount of resources used. Therefore, the resource control device 1 can reduce the possibility of a decline in user experience (UX).

[0027] In other words, the resource control device 1 can control the amount of resources used in the AI-RAN while reducing the possibility of a decrease in UX.

[0028] <Basic Configuration of Resource Control Device 1> The configuration of the resource control device 1 will be explained with reference to Figure 2. Figure 2 is a block diagram showing the configuration of the resource control device 1. As shown in Figure 2, the resource control device 1 includes a control unit 10, a storage unit 20, a communication unit 21, and an input / output unit 22.

[0029] <Control Unit 10> The control unit 10 controls each component of the resource control device 1. For example, the control unit 10 controls each component of the resource control device 1 by executing instructions written in the program. The control unit 10 is composed of, for example, arithmetic units such as a CPU (Central Processing Unit) and a GPU (Graphic Processing Unit), registers, and / or peripheral circuits. Details of the control unit 10 will be described later.

[0030] <Storage Unit 20> The storage unit 20 temporarily stores programs and data processed by programs, etc. The storage unit 20 also stores data and programs. The storage unit 20 is, for example, a combination of volatile memory such as DRAM (Dynamic Random Access Memory) and flash memory, HDD (Hard Disk Drive), or magneto-optical disk.

[0031] <Communication Unit 21> The communication unit 21 is an interface for inputting and outputting signals for communication with other devices via a network using wired or wireless communication standards. Examples of the communication unit 21 include communication chips for various communication standards such as Ethernet®, Wi-Fi®, and wireless communication standards for mobile data communication networks, as well as USB-compliant connectors.

[0032] <Input / Output Unit 22> The input / output unit 22 is an interface to an input device that receives data (for example, user input) and an output device that outputs data. Examples of input devices include a microphone, camera, eye-tracking device, keyboard, and touchpad. Examples of output devices include a speaker and liquid crystal display. The input / output unit 22 may also function as either an input device or an output device.

[0033] <Functions of the control unit 10> As shown in Figure 2, the control unit 10 includes an acquisition unit 11, a prediction unit 12, a decision unit 13, a notification unit 14, and an execution unit 15.

[0034] <Acquisition Unit 11> The acquisition unit 11 acquires data supplied from the communication unit 21 or the input / output unit 22. The acquisition unit 11 stores the acquired data in the storage unit 20.

[0035] As an example, the acquisition unit 11 acquires RAN metrics. That is, the acquisition unit 11 has the function of an orchestrator in AI-RAN. Examples of RAN metrics include, but are not limited to, the number of UEs (number of connected terminal TEs), throughput, power consumption, and information on the amount of resources used by the RAN application. Examples of information on the amount of resources used by the RAN application include, but are not limited to, CPU usage and GPU usage.

[0036] An example of the data that the acquisition unit 11 stores in the storage unit 20 will be explained with reference to Figure 3. Figure 3 is a table showing an example of the data stored in the storage unit 20.

[0037] The acquisition unit 11 acquires RAN metrics such as the number of UEs, average throughput, total throughput, CPU usage, and GPU usage at predetermined intervals, associates them with each other, and stores them in the storage unit 20. For example, at time "t-2", the acquisition unit 11 acquires the number of UEs "10", average throughput "20 Mbps", total throughput "XX Mbps", CPU usage "60%", and GPU usage "60%", and stores them in the storage unit 20, associating them with each other along with the time, as shown in Figure 3. Similarly, at time "t-1" and time "t0" (the current time), the acquisition unit 11 acquires the number of UEs, average throughput, total throughput, CPU usage, and GPU usage, associates them with each other, and stores them in the storage unit 20.

[0038] As another example, the acquisition unit 11 acquires information output from the terminal TE. For example, if the notification unit 14, which will be described later, notifies the terminal TE that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, and also asks the terminal TE to output information indicating whether or not to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed, the acquisition unit 11 will acquire information from the notifying terminal TE indicating whether or not to accept the reduction in the amount of resources used in response to the notification from the notification unit 14.

[0039] As another example, the acquisition unit 11 acquires information from at least one of the multiple terminals TE indicating a request to increase the terminal priority, as described later.

[0040] <Prediction Unit 12> The prediction unit 12 refers to the RAN metrics and predicts the amount of resources available to the AI ​​application after a predetermined time has elapsed. The predetermined time is not particularly limited and may be, for example, 10 minutes, 30 minutes, or 60 minutes. The prediction unit 12 stores information indicating the predicted amount of resources in the storage unit 20.

[0041] As an example, the prediction unit 12 predicts the amount of resources that the RAN application will use after a predetermined time has elapsed, based on the RAN metrics. The prediction unit 12 then predicts the amount of resources that the AI ​​application will be able to use after the predetermined time has elapsed as the difference between the total amount of resources that the AI ​​application and the RAN application can use and the predicted amount of resources that the RAN application will use after the predetermined time has elapsed.

[0042] An example of the process by which the prediction unit 12 predicts the amount of resources available to the AI ​​application after a predetermined time has elapsed will be explained again with reference to Figure 3.

[0043] The prediction unit 12 refers to the table shown in Figure 3 and refers to the RAN metrics from time "t-2" to time "t," namely the number of UEs, average throughput, and total throughput. The prediction unit 12 then predicts the number of UEs, average throughput, and total throughput at time "t1" and time "t2."

[0044] Next, the prediction unit 12 refers to the predicted number of UEs, average throughput, and total throughput at the time "t1", and predicts the CPU usage rate and GPU usage rate at the time "t1". Similarly for the time "t2", the prediction unit 12 refers to the predicted number of UEs, average throughput, and total throughput at the time "t2", and predicts the CPU usage rate and GPU usage rate at the time "t2".

[0045] The method predicted by the prediction unit 12 is not limited. As an example, methods using LSTM (Long Short Term Memory) and autoregressive models can be mentioned, but it is not limited to these.

[0046] With this configuration, the resource control device 1 can suitably predict the amount of resources available for use by the AI application after a predetermined time has elapsed.

[0047] <Decision unit 13> When there are a plurality of terminals TE that use the AI application, the decision unit 13 determines which terminal TE to notify to reduce the amount of resources used in the use of the AI application after a predetermined time has elapsed. The decision unit 13 stores information indicating the determined terminal TE in the storage unit 20.

[0048] As an example, the decision unit 13 sets a terminal priority for each of the plurality of terminals TE, and determines the terminal TE with a low terminal priority as the terminal TE to be notified that the amount of resources used by the AI application will be insufficient after a predetermined time has elapsed. In the present disclosure, as an example, the case where the decision unit 13 sets the terminal priority using a score and the higher the score, the higher the terminal priority will be described.

[0049] Other examples include the decision unit 13 randomly selecting a terminal TE to notify, or selecting a terminal with a long (short) usage time of the AI ​​application as the terminal TE to notify. Yet another example is the decision unit 13 selecting a terminal as the terminal TE that has a low actual usage rate of the resources requested and allocated when the application is executed. In other words, if the requested resources are excessive, resulting in a loss of opportunity for resources that could have been used by other applications, the decision unit 13 will select the terminal that excessively requested resources as the terminal TE to notify.

[0050] Furthermore, in response to the notification from the terminal TE to reduce the amount of resources used in using the AI ​​application, the decision unit 13 changes the score to be set based on the information obtained by the acquisition unit 11, which indicates whether or not the terminal TE that made the notification accepts the reduction in the amount of resources used.

[0051] An example of the processing performed by the decision unit 13 will be described later.

[0052] <Notification Unit 14> If the amount of resources available to the AI ​​application, as predicted by the prediction unit 12, is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application (in other words, the amount of resources used by the AI ​​application in the core network NW), the notification unit 14 will notify the terminal TE in advance that the amount of resources available to the AI ​​application will be insufficient after a predetermined time has elapsed.

[0053] Here, the amount of resources used in the processing requested by the terminal TE using the AI ​​application may be the amount of resources currently being used, or the amount of resources expected to be used after a predetermined time has elapsed. In other words, the notification unit 14 may compare the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed with the amount of resources that the AI ​​application is currently using, or it may compare the amount of resources that the AI ​​application is expected to use after a predetermined time has elapsed. Methods for predicting the amount of resources that the AI ​​application will use after a predetermined time have been used include, but are not limited to, methods using LSTM or autoregressive models, similar to the processing performed by the prediction unit 12 described above.

[0054] Furthermore, if there are multiple terminal TEs using the AI ​​application, the notification unit 14 notifies the terminal TE determined by the determination unit 13 that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0055] As an example, the notification unit 14 notifies the terminal TE that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, and also instructs it to output information indicating whether or not to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed.

[0056] An example of the processing performed by the notification unit 14 will be explained with reference to Figure 4. Figure 4 is an example of a table referenced by the notification unit 14.

[0057] As an example, as shown in Figure 4, we will describe a case where, among Node1 to Node3, Node1 runs the application for RAN, and a Pod containing containers A to C of the AI ​​application is deployed.

[0058] Furthermore, as shown in Figure 4, the prediction unit 12 predicts that the amount of resources available to the AI ​​application at t1 (after 10 minutes) will be 90% of the total resources available to the AI ​​application and the RAN application. Similarly, the prediction unit 12 predicts that the amount of resources available to the AI ​​application at t2 (after 20 minutes) will be 50% of the total resources, and that the amount of resources available to the AI ​​application at t3 (after 30 minutes) will be 50% of the total resources.

[0059] The notification unit 14 first determines whether the amount of resources available to the AI ​​application at t1 (10 minutes later) is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application. In other words, the notification unit 14 determines whether it is predicted that there will be insufficient resources to execute the processing requested by the terminal TE using the AI ​​application at t1 (10 minutes later).

[0060] If, at t1 (10 minutes later), the notification unit 14 determines that the amount of resources available to the AI ​​application is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application, the notification unit 14 will notify the terminal TE in advance of t1 (10 minutes later) that the amount of resources the AI ​​application will use at t1 (10 minutes later) will be insufficient. In other words, if the notification unit 14 predicts that the AI ​​application will use more than 90% of the total resources at t1 (10 minutes later), the notification unit 14 will notify the terminal TE in advance of t1 (10 minutes later) that the amount of resources the AI ​​application will use at t1 (10 minutes later) will be insufficient.

[0061] Furthermore, as described above, if there are multiple terminal TEs using the AI ​​application, the notification unit 14 will notify the terminal TEs determined by the decision unit 13 in advance of t1 (10 minutes later) that the amount of resources the AI ​​application will use at t1 (10 minutes later) will be insufficient.

[0062] On the other hand, if it is determined that the amount of resources available to the AI ​​application at t1 (10 minutes later) is greater than or equal to the amount of resources used in the processing requested by the terminal TE using the AI ​​application, then the notification unit 14 does not send any notification because there is no shortage of resources to be used in the processing requested by the terminal TE using the AI ​​application at t1 (10 minutes later).

[0063] The notification unit 14 performs the same processing for t2 (20 minutes later) and t3 (30 minutes later).

[0064] <Execution Unit 15> The execution unit 15 performs processes to reduce the amount of resources used in using the AI ​​application. Examples of processes performed by the execution unit 15 to reduce the amount of resources used in using the AI ​​application include, but are not limited to, stopping the AI ​​application and reducing or deleting the number of replica Pods created for redundancy. In this disclosure, the case in which the execution unit 15 reduces the amount of resources used in using the AI ​​application by stopping the AI ​​application will be described.

[0065] For example, the execution unit 15 determines whether the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources. If the execution unit 15 determines that the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources, it refers to the information acquired by the acquisition unit 11 and executes a process to reduce the amount of resources used by the AI ​​application.

[0066] In other words, if the amount of resources used by the AI ​​application becomes insufficient after a predetermined period of time, the system will refer to the information from the terminal TE in response to the notification and execute a process to reduce the amount of resources used by the AI ​​application.

[0067] An example of the processing performed by the execution unit 15 will be described later.

[0068] <Processing Executed by Resource Control Device 1> The flow of processing executed by Resource Control Device 1 will be explained with reference to Figures 5 and 6. Figure 5 is a flowchart showing the flow of processing executed by Resource Control Device 1. Figure 6 is a diagram showing an example of processing executed by Resource Control Device 1.

[0069] <Step S11> In step S11, the acquisition unit 11 acquires RAN metrics. The acquisition unit 11 stores the acquired RAN metrics in the storage unit 20. Step S11 may be executed multiple times at predetermined intervals.

[0070] <Step S12> In step S12, the prediction unit 12 refers to the RAN metrics acquired by the acquisition unit 11 in step S11 and predicts the amount of resources used by the RAN application after a predetermined time has elapsed.

[0071] As an example, let's assume that, as shown in the upper table of Figure 6, currently, the RAN application is using 30% of the total resources, the AI ​​application on terminal A is using 20% ​​of the total resources, the AI ​​application on terminal B is using 20% ​​of the total resources, and the AI ​​application on terminal C is using 20% ​​of the total resources.

[0072] In this case, the prediction unit 12 refers to the RAN metrics and predicts that the amount of resources used by the RAN application after 30 minutes will be 80% of the total resources, as shown in the table in the center of Figure 6.

[0073] <Step S13> In step S13, the prediction unit 12 refers to the amount of resources that the RAN application will be able to use after a predetermined time has elapsed, which was predicted in step S12, and predicts the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed. The prediction unit 12 stores the predicted amount of resources in the storage unit 20.

[0074] For example, if in step S12 the amount of resources used by the RAN application after 30 minutes is predicted to be 80% of the total resources, then in step S13 the prediction unit 12 predicts that the amount of resources available to the AI ​​application after 30 minutes will be 20%.

[0075] <Step S14> In step S14, the notification unit 14 determines whether the amount of resources available to the AI ​​application, as predicted by the prediction unit 12 in step S13, is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application.

[0076] In step S14, if it is determined that the predicted amount of resources is greater than or equal to the amount of resources used in the processing requested by the terminal TE using the AI ​​application (step S14: NO), the resource control device 1 terminates the process shown in Figure 5.

[0077] <Step S15> On the other hand, if in step S14 it is determined that the predicted amount of resources is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application (step S14: YES), in step S15 the decision unit 13 decides which terminal TE to notify that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0078] For example, in step S13, it is predicted that the amount of resources available to the AI ​​application after 30 minutes will be 20%, and as shown in the table in the center of Figure 6, the amount of resources used in the processing requested from terminal A is 20% of the total resources, the amount of resources used in the processing requested from terminal B is 20% of the total resources, and the amount of resources used in the processing requested from terminal C is 20% of the total resources.

[0079] In this case, the amount of resources used in the processing requested by terminals A, B, and C using the AI ​​application is 20 + 20 + 20 = 60% of the total resources. Therefore, the predicted amount of resources is less than the amount of resources used in the processing requested by terminal TE using the AI ​​application. Accordingly, in step S15, the decision unit 13 decides which terminal TE to notify that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0080] Here, as shown in the upper table of Figure 6, terminals A, B, and C are assigned scores of "1," "2," and "3," respectively, to indicate terminal priority. As described above, this disclosure describes a case where a higher score indicates higher terminal priority, so the terminal priority is terminal C > terminal B > terminal A.

[0081] The decision unit 13 first calculates the amount of resources to be reduced. In this example, the amount of resources available to the AI ​​application after 30 minutes is 20%, and the amount of resources used in the processing requested by terminals A, B, and C is 60%, so the amount of resources to be reduced is calculated to be 40%.

[0082] Next, the decision unit 13 determines that terminals with low terminal priority will be notified in 30 minutes that they will have insufficient resources to use in the AI ​​application, in order to reduce the amount of resources by 40%. In this example, if the amount of resources used in the processing requested by terminal A, which has the lowest terminal priority, and terminal B, which has the next lowest priority, is removed, the amount of resources can be reduced by 40%. Therefore, the decision unit 13 determines that terminals A and B will be notified in 30 minutes that they will have insufficient resources to use in the AI ​​application.

[0083] With this configuration, the resource control device 1 can determine a terminal TE with a low terminal priority as the terminal TE to which the AI ​​application will notify that it has insufficient resources.

[0084] <Step S16> In step S16, the notification unit 14 notifies the terminal TE determined by the decision unit 13 that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0085] For example, as shown in the table in the center of Figure 6, if the terminal TE determined by the determination unit 13 is terminal A and terminal B, the notification unit 14 notifies terminals A and B that the amount of resources used by the AI ​​application will be insufficient in 30 minutes.

[0086] Furthermore, the notification unit 14 notifies terminals A and B that the amount of resources used by the AI ​​application will be insufficient in 30 minutes, and also instructs them to output information indicating whether or not to reduce the amount of resources used by the AI ​​application in 30 minutes. For example, the notification unit 14 notifies terminals A and B with the message, "There is a possibility that resources will be insufficient in 30 minutes. Is it alright to stop the AI ​​application if resources become insufficient?"

[0087] <Step S17> In step S17, the acquisition unit 11 acquires information from the notifying terminal TE indicating whether or not it accepts the reduction in the amount of resources to be used, in response to the notification by the notification unit 14 in step S16.

[0088] For example, as shown in the table in the center of Figure 6, if the notification unit 14 notifies terminals A and B that the amount of resources used by the AI ​​application will be insufficient in 30 minutes, and also instructs them to output information indicating whether or not to reduce the amount of resources used by the AI ​​application in 30 minutes, the acquisition unit 11 acquires information from each of terminals A and B indicating whether or not they accept the reduction in the amount of resources used.

[0089] With this configuration, the resource control device 1 can recognize the user's intention to accept or reject a reduction in the amount of resources used.

[0090] <Step S18> In step S18, the execution unit 15 determines whether the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources.

[0091] For example, as shown in the table in the lower left of Figure 6, if the amount of resources used by the RAN application after 30 minutes is 60% of the total resources, then the amount of resources used by the AI ​​application after 30 minutes will be 40%. On the other hand, as shown in the center of Figure 6, the amount of resources used by the AI ​​application after 30 minutes is 20 + 20 + 20 = 60% of the total resources, so the execution unit 15 determines that the amount of resources used by the AI ​​application after 30 minutes is greater than the amount of available resources.

[0092] Similarly, as shown in the table in the lower right of Figure 6, if the amount of resources used by the RAN application after 30 minutes is 80% of the total resources, then the amount of resources used by the AI ​​application after 30 minutes is 20%. On the other hand, as mentioned above, the amount of resources used by the AI ​​application after 30 minutes is 60% of the total resources, so the execution unit 15 determines that the amount of resources used by the AI ​​application after 30 minutes is greater than the amount of available resources.

[0093] In step S18, if it is determined that the amount of resources used by the AI ​​application after a predetermined time has elapsed is less than or equal to the amount of available resources (step S18: NO), the resource control device 1 terminates the process shown in Figure 5.

[0094] For example, if the amount of resources used by the RAN application after 30 minutes is 30% of the total resources, then the amount of resources used by the AI ​​application after 30 minutes will be 70%. On the other hand, if the amount of resources used by the AI ​​application after 30 minutes is 60% of the total resources, then there is no shortage of resources after 30 minutes, and the resource control device 1 terminates the process shown in Figure 5.

[0095] <Step S19> In step S19, the execution unit 15 performs a process to reduce the amount of resources used in using the AI ​​application.

[0096] For example, as shown in the center of Figure 6, let's assume that in step S16, the notification unit 14 notifies terminals A and B that the amount of resources used by the AI ​​application will be insufficient in 30 minutes, and also instructs them to output information indicating whether or not to reduce the amount of resources used by the AI ​​application in 30 minutes.

[0097] In this case, for example, in step S17, the acquisition unit 11 acquires information from terminal A indicating that it will accept a reduction in the amount of resources used, as shown in the center of Figure 6, and acquires information from terminal B indicating that it will not accept a reduction in the amount of resources used.

[0098] In this case, the execution unit 15 refers to the information acquired by the acquisition unit 11 and executes a process to reduce the amount of resources used in using the AI ​​application.

[0099] For example, as shown in the table in the lower left of Figure 6, if the amount of resources used by the RAN application after 30 minutes is 60% of the total resources, then the amount of resources used by the AI ​​application after 30 minutes will be 40%. In this case, since it is necessary to reduce the total resources by 20%, the execution unit 15 stops the AI ​​application on terminal A, which has output information indicating that it accepts the resource reduction.

[0100] As another example, as shown in the table in the lower right of Figure 6, if the amount of resources used by the RAN application after 30 minutes is 80% of the total resources, then the amount of resources used by the AI ​​application after 30 minutes is 20%. In this case, since 40% of the total resources need to be reduced, the execution unit 15 first stops the AI ​​application on terminal A, which has output information indicating that it accepts the resource reduction.

[0101] Since stopping the AI ​​application on terminal A would not reduce the total resource usage by 40%, the execution unit 15 then determines whether there is a terminal TE that has output information indicating acceptance of the resource reduction. If there is a terminal TE that has output information indicating acceptance of the resource reduction, the execution unit 15 stops the AI ​​application on that terminal TE.

[0102] On the other hand, if no terminal TE outputs information indicating that it accepts the reduction of resources, the execution unit 15 stops the AI ​​application run by terminal B, which is the terminal TE with the lowest terminal priority among the terminal TEs running the AI ​​application, as shown in the table in the lower right of Figure 6. In other words, the execution unit 15 stops the AI ​​application run by terminal B even though terminal B has output information indicating that it does not accept the reduction of resources.

[0103] With this configuration, if the amount of resources used by the AI ​​application is insufficient, the resource control device 1 reduces the amount of resources used by the AI ​​application based on information obtained from the terminal TE. This makes it possible to control the amount of resources used while minimizing the possibility of a decrease in UX.

[0104] <Step S20> In step S20, the determination unit 13 refers to the information acquired by the acquisition unit 11 in step S17 and updates the score.

[0105] For example, let's consider the case where terminal A is assigned a score of "1", terminal B a score of "2", and terminal C a score of "3", as shown in the table at the top or center of Figure 6.

[0106] In this case, if the acquisition unit 11 acquires information from terminal A in step S17 indicating that terminal A accepts a reduction in the amount of resources used, the decision unit 13 sets terminal A's terminal priority to be higher. For example, as shown in the lower left or lower right of Figure 6, the decision unit 13 adds a score of "3" to terminal A's score of "1" as an incentive for outputting information indicating that terminal A accepts a reduction in the amount of resources used.

[0107] On the other hand, if in step S17 the acquisition unit 11 obtains information from terminal B indicating that it will not accept a reduction in the amount of resources used, the decision unit 13 will not change terminal B's score. For example, as shown in the lower left of Figure 6, even if the AI ​​application on terminal B is not stopped, or as shown in the lower right of Figure 6, even if the AI ​​application on terminal B is stopped, the decision unit 13 will not change terminal B's score.

[0108] With this configuration, terminal A, which outputs information indicating that it accepts a reduction in the amount of resources used, will have a higher terminal priority. Therefore, the next time resources become insufficient, terminal A is less likely to have its resource usage reduced. As a result, the resource control device 1 can reduce the likelihood of a deterioration in the user experience (UX) of terminal A.

[0109] Another example of the process by which the determination unit 13 updates the score in step S20 is when the acquisition unit 11 acquires information from at least one of the multiple terminals TE indicating a request to increase the terminal priority, the determination unit 13 sets the terminal priority of the terminal TE that output the information acquired by the acquisition unit 11 indicating a request to increase the terminal priority to be higher.

[0110] For example, if the acquisition unit 11 acquires information from terminal A indicating a request to add "5" to the terminal priority score, the determination unit 13 adds "5" to terminal A's score. In this case, the determination unit 13 may subtract "5" from terminal A's score after a predetermined time (for example, 1 hour) has elapsed, returning terminal A's score to its value before the addition. In other words, the determination unit 13 may be configured to set terminal A's terminal priority higher for a predetermined time.

[0111] An example of how the acquisition unit 11 acquires information from at least one of multiple terminals TE indicating a request to increase terminal priority is information indicating what level of incentive would be given for a terminal to accept a reduction in the amount of resources used.

[0112] For example, let's consider a case where the acquisition unit 11 acquires information from terminal A indicating that it will accept a reduction in the amount of resources used if a score of "3" is added, and information from terminal B indicating that it will accept a reduction in the amount of resources used if a score of "5" is added. In this case, the decision unit 13 identifies terminal TE that will accept a reduction in the amount of resources used by adding a smaller score. That is, the decision unit 13 identifies that terminal A will accept a reduction in the amount of resources used by terminal B with a smaller score. Then, when the AI ​​application on terminal A is stopped, the decision unit 13 adds "3" to terminal A's score.

[0113] With this configuration, the resource control device 1 can change the terminal priority according to the user's request for each terminal TE.

[0114] Alternatively, the resource control device 1 may execute step S20 after step S17, and step S18 (and step S19) after step S20.

[0115] For example, in step S17, if the acquisition unit 11 acquires information from terminal A indicating that it accepts a reduction in the amount of resources used, the decision unit 13 sets terminal A's terminal priority to be higher, regardless of whether or not resources become insufficient after a predetermined time has elapsed. In other words, the decision unit 13 may provide an incentive to terminal A of a user who has agreed to accept a reduction in the amount of resources used, for the act of agreeing itself.

[0116] <Effects of Resource Control Device 1> As described above, the resource control device 1 includes an acquisition unit 11 that acquires RAN metrics, a prediction unit 12 that predicts the amount of resources available to the AI ​​application after a predetermined time has elapsed, a determination unit 13 that determines the terminal TE to notify, a notification unit 14 that notifies the determined terminal TE that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, and an execution unit 15 that refers to information acquired by the acquisition unit 11 from the notified terminal TE indicating whether or not to accept the reduction in the amount of resources used, and executes a process to reduce the amount of resources used by the AI ​​application.

[0117] In other words, if the resource control device 1 predicts that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, it notifies the terminal TE, which has been decided to receive notifications, that the amount of resources used by the AI ​​application will be insufficient after the predetermined time has elapsed. The resource control device 1 then refers to the information from the terminal TE indicating whether or not it accepts the reduction in the amount of resources used, and executes a process to reduce the amount of resources used by the AI ​​application.

[0118] With this configuration, the resource control device 1 controls the amount of resources used appropriately. Furthermore, the resource control device 1 does not stop the AI ​​application without notification, nor does it notify the terminal TE immediately before the AI ​​application stops. Therefore, in AI-RAN, the resource control device 1 can control the amount of resources used while reducing the possibility of a decrease in UX.

[0119] <Embodiment 2> Other embodiments of the present disclosure will be described in detail below with reference to the drawings. For the sake of convenience of explanation, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0120] <Overview of Wireless Communication System 100A> An overview of wireless communication system 100A will be described below. Wireless communication system 100A has a configuration that includes resource control device 1A instead of resource control device 1 included in the wireless communication system 100 described above. Therefore, at least one terminal TE, at least one base station BS, and the core network NW other than resource control device 1A in wireless communication system 100A are as described above.

[0121] <Basic Configuration of Resource Control Device 1A> The basic configuration of the resource control device 1A will be explained with reference to Figure 7. Figure 7 is a block diagram showing the configuration of the resource control device 1A. The resource control device 1A includes a control unit 10A, a storage unit 20, a communication unit 21, and an input / output unit 22. The storage unit 20, the communication unit 21, and the input / output unit 22 are as described above.

[0122] <Control Unit 10A> The control unit 10A controls each component of the resource control device 1A. For example, the control unit 10A controls each component of the resource control device 1A by executing instructions written in a program. The control unit 10A is composed of, for example, arithmetic units such as a CPU and GPU, registers, and / or peripheral circuits.

[0123] Furthermore, as shown in Figure 7, the control unit 10A includes a prediction unit 12A and a notification unit 14A.

[0124] <Prediction Unit 12A> The prediction unit 12A refers to the RAN metrics and predicts the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed. The prediction unit 12A stores information indicating the predicted amount of resources in the storage unit 20. The RAN metrics may be stored in the storage unit 20 in advance, or the prediction unit 12A may acquire the RAN metrics, as in the acquisition unit 11 of the embodiment described above.

[0125] <Notification Unit 14A> If the amount of resources available to the AI ​​application is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application, the notification unit 14A will notify the terminal TE in advance that the amount of resources the AI ​​application will use will be insufficient after a predetermined time has elapsed.

[0126] The notification unit 14A may be configured to notify all terminal TEs if there are multiple terminal TEs, or, as in the determination unit 13 of the embodiment described above, it may be configured to determine which terminal TE to notify and notify the determined terminal TE.

[0127] <Processing performed by resource control device 1A> The flow of processing performed by resource control device 1A will be explained with reference to Figure 8. Figure 8 is a flowchart showing the flow of processing performed by resource control device 1A.

[0128] <Step S13A> In step S13A, the prediction unit 12A refers to the RAN metrics and predicts the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed. The prediction unit 12A stores information indicating the predicted amount of resources in the storage unit 20.

[0129] <Step S14A> In step S14A, the notification unit 14A determines whether the amount of resources available to the AI ​​application, as predicted by the prediction unit 12A in step S13A, is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application.

[0130] In step S14A, if it is determined that the predicted amount of resources is greater than or equal to the amount of resources used in the processing requested by the terminal TE using the AI ​​application (step S14A: NO), the resource control device 1A terminates the process shown in Figure 8.

[0131] <Step S16A> If, in step S14A, it is determined that the predicted amount of resources is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application (step S14A: YES), then in step S16A, the notification unit 14A notifies the terminal TE in advance that the amount of resources to be used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0132] <Effects of Resource Control Device 1A> As described above, the resource control device 1A includes a prediction unit 12A that refers to RAN metrics and predicts the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed, and a notification unit 14A that, if the predicted amount of resources that the AI ​​application will be able to use is less than the amount of resources used in the processing requested by the terminal TE using the AI ​​application, notifies the terminal in advance that the amount of resources that the AI ​​application will be able to use after a predetermined time has elapsed.

[0133] According to this configuration, for example, in the AI-RAN described above, the resource control device 1A first secures resources to be used by applications for the RAN with high priority. Next, the resource control device 1A allocates resources not being used by the high-priority RAN applications to the AI ​​applications. Therefore, the resource control device 1A can appropriately control the amount of resources used.

[0134] Furthermore, if the resource control device 1A predicts that the resources allocated to the AI ​​application will become insufficient after a predetermined time, it will notify the terminal TE in advance to reduce the amount of resources used. Therefore, the resource control device 1A can reduce the possibility of a decline in UX.

[0135] In other words, the resource control device 1A can control the amount of resources used in the AI-RAN while reducing the possibility of a decrease in UX.

[0136] <Example of implementation by software> The functions of each device (hereinafter referred to as "device") that constitutes the wireless communication systems 100 and 100A are programs that cause a computer to function as the device, and these programs can be implemented by programs that cause a computer to function as each control block (particularly each part included in the control units 10 and 10A) of the device.

[0137] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0138] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0139] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0140] Furthermore, each of the processes described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0141] <Summary> This disclosure contains at least the following aspects:

[0142] A resource control device according to Embodiment 1 of the present disclosure includes: a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources that an AI (Artificial Intelligence) application can use after a predetermined time has elapsed; and a notification unit that, if the predicted amount of resources that the AI ​​application can use is less than the amount of resources used in processing requested by a terminal using the AI ​​application, notifies the terminal in advance that the amount of resources that the AI ​​application can use after the predetermined time has elapsed will be insufficient.

[0143] According to the above configuration, the resource control device according to Embodiment 1 of this disclosure can control the amount of resources used in the AI-RAN while reducing the possibility of a decrease in UX.

[0144] The resource control device according to aspect 2 of the present disclosure further comprises a determination unit that, in aspect 1, if there are multiple terminals using the AI ​​application, determines which terminal to notify that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, and the notification unit notifies the terminal determined by the determination unit that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

[0145] According to the above configuration, the resource control device according to aspect 2 of this disclosure can notify any of the multiple terminals that the amount of resources used by the AI ​​application is insufficient.

[0146] In the resource control device according to aspect 3 of the present disclosure, the determination unit in aspect 2 sets a terminal priority for each of the plurality of terminals and determines the terminal with the lowest terminal priority to be the terminal that will notify the AI ​​application that the amount of resources it will use will be insufficient after a predetermined time has elapsed.

[0147] According to the above configuration, the resource control device according to aspect 3 of the present disclosure can determine terminals with low terminal priority as terminals to be notified that the amount of resources used by the AI ​​application is insufficient.

[0148] In the resource control device according to aspect 4 of the present disclosure, the notification unit in aspect 3 notifies the terminal that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed, and also notifies the terminal whether or not to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed, and further includes an acquisition unit that acquires information from the notifying terminal indicating whether or not to accept the reduction in the amount of resources used in response to the notification to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed.

[0149] According to the above configuration, the resource control device according to aspect 4 of the present disclosure can recognize the user's intention to accept or reject a reduction in the amount of resources used.

[0150] In the resource control device according to aspect 5 of the present disclosure, when the acquisition unit in aspect 4 acquires information from the notified terminal indicating acceptance of a reduction in the amount of resources to be used, the decision unit sets the terminal priority of the notified terminal to be higher.

[0151] According to the above configuration, the resource control device according to aspect 5 of this disclosure can reduce the possibility of a decrease in the user experience (UX) of the notified terminal.

[0152] In the resource control device according to embodiment 6 of the present disclosure, if the acquisition unit in embodiment 4 or 5 acquires information from at least one of the plurality of terminals indicating a request to increase the terminal priority, the determination unit sets the terminal priority of the terminal that output the information indicating a request to increase the terminal priority, which was acquired by the acquisition unit, to be increased.

[0153] According to the above configuration, the resource control device according to embodiment 6 of this disclosure can change the terminal priority according to the request of the user of each terminal.

[0154] The resource control device according to embodiment 7 of the present disclosure further comprises, in any of embodiments 4 to 6, an execution unit that determines whether the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources, and if it determines that the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources, an execution unit that refers to the information acquired by the acquisition unit and executes a process to reduce the amount of resources used by the AI ​​application.

[0155] According to the above configuration, the resource control device according to embodiment 7 of this disclosure can control the amount of resources used while reducing the possibility of a decrease in UX.

[0156] In a resource control device according to aspect 8 of the present disclosure, in any of aspects 1 to 7, the prediction unit predicts the amount of resources used by the application for the wireless communication network after a predetermined time has elapsed, based on metrics of the wireless communication network, and predicts the amount of resources that the AI ​​application can use after the predetermined time has elapsed as the difference between the total amount of resources that the AI ​​application and the application for the wireless communication network can use and the predicted amount of resources used by the application for the wireless communication network after the predetermined time has elapsed.

[0157] According to the above configuration, the resource control device according to embodiment 8 of the present disclosure can suitably predict the amount of resources available to the AI ​​application after a predetermined time has elapsed.

[0158] A resource control method according to aspect 9 of the present disclosure includes a prediction step of predicting the amount of resources that an AI application can use after a predetermined time has elapsed by referring to metrics of a wireless communication network, and a notification step of notifying the terminal in advance that the amount of resources that the AI ​​application can use after the predetermined time has elapsed will be insufficient if the predicted amount of resources that the AI ​​application can use is less than the amount of resources used in processing requested by the terminal using the AI ​​application.

[0159] According to the above configuration, the resource control method according to aspect 9 of this disclosure can control the amount of resources used in the AI-RAN while reducing the possibility of a decrease in UX.

[0160] A resource control program according to aspect 10 of the present disclosure is a program for causing a computer to function as a resource control device, wherein the computer functions as a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources that an AI application can use after a predetermined time has elapsed, and a notification unit that, if the predicted amount of resources that the AI ​​application can use is less than the amount of resources used in processing requested by a terminal using the AI ​​application, notifies the terminal in advance that the amount of resources that the AI ​​application can use after the predetermined time has elapsed will be insufficient.

[0161] According to the above configuration, the resource control program according to embodiment 10 of this disclosure can control the amount of resources used in the AI-RAN while reducing the possibility of a decrease in UX.

[0162] (Additional Notes) This disclosure is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of this disclosure.

[0163] This invention allows for optimal control of resource usage. As a result, energy efficiency can be improved through power saving. Such effects contribute to achieving goals such as Goal 7 of the United Nations' Sustainable Development Goals (SDGs), "Affordable and Clean Energy."

[0164] 1, 1A Resource control device 10, 10A Control unit 11 Acquisition unit 12, 12A Prediction unit 13 Decision unit 14, 14A Notification unit 15 Execution unit 20 Storage unit 21 Communication unit 22 Input / Output unit

Claims

1. A resource control device comprising: a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources available to an AI (Artificial Intelligence) application after a predetermined time has elapsed; and a notification unit that, if the predicted amount of resources available to the AI ​​application is less than the amount of resources used in processing requested by the terminal using the AI ​​application, notifies the terminal in advance that the amount of resources available to the AI ​​application after the predetermined time has elapsed will be insufficient.

2. If there are multiple terminals using the AI ​​application, the resource control device further comprises a determination unit that determines which terminal to notify of the insufficient amount of resources used by the AI ​​application after a predetermined time has elapsed, and the notification unit notifies the terminal determined by the determination unit that the insufficient amount of resources used by the AI ​​application after a predetermined time has elapsed.

3. The resource control device according to claim 2, wherein the determination unit sets a terminal priority for each of the plurality of terminals, and determines the terminal with the lowest terminal priority to be the terminal to notify that the amount of resources used by the AI ​​application will be insufficient after a predetermined time has elapsed.

4. The resource control device according to claim 3, wherein the notification unit notifies the terminal that the amount of resources used by the AI ​​application will be insufficient after the predetermined time has elapsed, and also notifies the terminal whether or not to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed, and further comprises an acquisition unit that acquires information from the notifying terminal indicating whether or not to accept the reduction in the amount of resources used in response to the notification to reduce the amount of resources used by the AI ​​application after the predetermined time has elapsed.

5. The resource control device according to claim 4, wherein when the acquisition unit acquires information from the notified terminal indicating acceptance of a reduction in the amount of resources to be used, the decision unit sets the terminal priority of the notified terminal to be higher.

6. The resource control device according to claim 4, wherein if the acquisition unit acquires information from at least one of the plurality of terminals indicating a request to increase the terminal priority, the determination unit sets the terminal priority of the terminal that output the information acquired by the acquisition unit indicating a request to increase the terminal priority to be increased.

7. The resource control device according to any one of claims 4 to 6, further comprising: an execution unit that determines whether the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources, and if it is determined that the amount of resources used by the AI ​​application after a predetermined time has elapsed is greater than the amount of available resources, an execution unit that refers to the information acquired by the acquisition unit and executes a process to reduce the amount of resources used by the AI ​​application.

8. The resource control device according to any one of claims 1 to 6, wherein the prediction unit predicts the amount of resources used by the application for the wireless communication network after a predetermined time has elapsed from metrics of the wireless communication network, and predicts the amount of resources that the AI ​​application can use after the predetermined time has elapsed as the difference between the total amount of resources that the AI ​​application and the application for the wireless communication network can use and the predicted amount of resources used by the application for the wireless communication network after the predetermined time has elapsed.

9. A resource control method comprising: a prediction step of predicting the amount of resources available to an AI application after a predetermined time has elapsed by referring to metrics of a wireless communication network; and a notification step of notifying the terminal in advance that the amount of resources available to the AI ​​application will be insufficient after the predetermined time has elapsed, if the predicted amount of resources available to the AI ​​application is less than the amount of resources used in processing requested by the terminal using the AI ​​application.

10. A program for causing a computer to function as a resource control device, the program comprising: a prediction unit that refers to metrics of a wireless communication network and predicts the amount of resources available to an AI application after a predetermined time has elapsed; and a notification unit that, if the predicted amount of resources available to the AI ​​application is less than the amount of resources used in processing requested by a terminal using the AI ​​application, notifies the terminal in advance that the amount of resources to be used by the AI ​​application after the predetermined time has elapsed will be insufficient.

Images