Task allocation method, computer-readable storage medium, electronic apparatus and computer program product

Abstract

Provided in the embodiments of the present disclosure are a task allocation method, a computer-readable storage medium, an electronic apparatus and a computer program product. The method comprises: receiving a task request sent by a second entity for requesting the execution of a target task; on the basis of computing power information of a plurality of third entities, determining a target entity that meets computing power required for executing the target task; and sending the task request to the target entity for processing. The problem in the related art of some network elements having heavy loads but low computing power, while some network elements have light loads but high computing power can be solved, and it can be ensured that tasks are allocated to entities having sufficient computing power, thereby improving the efficiency and success rate of task execution, and avoiding the waste of computing power resources.

Description

Task allocation methods, computer-readable storage media, electronic devices and computer program products

[0001] Cross-references to related applications

[0002] This disclosure is based on and claims priority to Chinese Patent Application No. 2025100182756, filed on January 2, 2025, entitled “Task Allocation Method, Computer-Readable Storage Medium, Electronic Device and Computer Program Product”, and incorporates the entire contents of that patent application by reference. Technical Field

[0003] This disclosure relates to the field of communications, and more specifically, to a task allocation method, a computer-readable storage medium, an electronic device, and a computer program product. Background Technology

[0004] In the future 6G core network, network elements will handle a large number of AI-related tasks (model training, inference, testing, etc.). Different tasks consume different amounts of computing power and have different priorities. The relevant technologies lack coordination measures for artificial intelligence / machine learning (AI / ML) tasks. In the future 6G network, this may lead to problems such as some network elements having a heavy load but low computing power, while others have a low load but high computing power. Summary of the Invention

[0005] This disclosure provides a task allocation method, a computer-readable storage medium, an electronic device, and a computer program product to at least solve the problem in the related art that some network elements have high load but low computing power, while others have low load but high computing power.

[0006] According to one embodiment of this disclosure, a task allocation method is provided, applied to a first subject, comprising:

[0007] Receive a task request from a second entity to execute the target task;

[0008] The target entity that meets the computing power requirements for executing the target task is determined based on the computing power information of multiple third entities;

[0009] The task request is sent to the target entity for processing.

[0010] According to another embodiment of this disclosure, a task allocation method is provided, applied to a first subject, comprising:

[0011] Receive the first information sent by the third entity and / or the second entity;

[0012] Based on the first information, the task request demand and / or computing power demand for different time periods in the future are predicted to obtain task processing suggestions and / or task request suggestions.

[0013] Send the task processing suggestion and / or the task request suggestion to the third entity and / or the second entity.

[0014] According to yet another embodiment of this disclosure, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program is configured to perform the steps in any of the above method embodiments when it is run.

[0015] According to yet another embodiment of this disclosure, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0016] According to yet another embodiment of this disclosure, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps in any of the above method embodiments. Attached Figure Description

[0017] Figure 1 is a schematic diagram of the 5G core network architecture based on relevant technologies;

[0018] Figure 2 is a schematic diagram of a 5G system architecture based on ADRF in related technologies;

[0019] Figure 3 is a flowchart of a task allocation method according to an embodiment of the present disclosure;

[0020] Figure 4 is a flowchart of NWDAF computing power status reporting and NWDAF selection according to an embodiment of the present disclosure;

[0021] Figure 5 is a flowchart of instant NWDAF matching according to an embodiment of the present disclosure;

[0022] Figure 6 is a flowchart of a task allocation method according to an embodiment of the present disclosure;

[0023] Figure 7 is a flowchart of the overall planning of future time period tasks according to an embodiment of the present disclosure;

[0024] Figure 8 is a structural block diagram of a task allocation device according to an embodiment of the present disclosure;

[0025] Figure 9 is a structural block diagram of a task allocation device according to an embodiment of the present disclosure. Detailed Implementation

[0026] The embodiments of this disclosure will be described in detail below with reference to the accompanying drawings and examples.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0028] Figure 1 is a schematic diagram of the 5G core network architecture based on related technologies. As shown in Figure 1, it includes:

[0029] 1) User Equipment (UE).

[0030] 2) Radio Access Network (RAN): The RAN manages radio resources, transmits user data received through N3 to the UE, and transmits user data from the UE through the N3 interface. The RAN maps QoS traffic between DRB (Dedicated Radio Bearer) and PDU sessions.

[0031] 3) Access and Mobility Management Functions (AMF): This function includes registration management, connection management, reachability management, and mobility management. It also performs access authentication and authorization. The AMF is a NAS security endpoint that forwards SM NAS information between the UE and the SMF.

[0032] 4) Session Management Function (SMF): This function includes the following: session establishment, modification, and release; UE IP address allocation and management (including optional authorization functions); UP function selection and control; downlink data notification, etc. The SMF controls the UPF through N4 association. The SMF provides the UPF with a PDR (Packet Detection Rule) instructing how to detect user data traffic; and provides FAR (Forwarding Action Rule), QER (Quality of Service Enforcement Rules), and URR (Usage Reporting Rules) instructing the UPF how to perform user data traffic forwarding, QoS processing, and usage reporting on user data traffic detected using the PDR.

[0033] 5) User plane function (UPF): This function includes the following: acting as an anchor point for intra / inter-radio access type mobility, packet routing and forwarding, traffic usage reporting, user plane QoS processing, downlink packet buffering, and downlink data notification triggering. A GTP-U (GPRS (General Packet Radio Service) Tunneling Protocol User Plane) tunnel is used for the N3 interface between the RAN and the UPF. The GTP-U tunnel operates on a per-PDU session basis. For downlink traffic, the UPF binds the QoS traffic within the downlink traffic PDU (Protocol Data Unit) session GTP-U tunnel using the FAR received from the SMF. For uplink traffic, the RAN transmits user plane traffic to a QoS stream recognized by the UE.

[0034] 6) Policy Control Function (PCF): The PCF provides QoS policy rules to control plane functions for enforcement. The PCF translates AF requests into PCC rules applicable to PDU sessions.

[0035] 7) Unified Data Management (UDM): UDM performs 3GPP AKA (Authentication and Key Agreement) authentication credential generation, access authorization based on subscription data, UE service element registration management (e.g., storing AMF for UEs, storing SMF for UE PDU sessions), and subscription management. UDM accesses UDR (Unified Data Repository) to retrieve UE subscription data and store UE context in UDR. UDM and UDR can be deployed together.

[0036] The Network Data Analysis Function (NWDAF) is a 5GC network element located in the control plane that performs statistical data and machine learning tasks within the 5GS. NWDAF can interact with different entities for various purposes:

[0037] Data is collected based on event subscriptions provided by AMF, SMF, UPF, PCF, UDM, NSACF (Network Slice Application Function), AF (Application Function) (directly or through NEF (Network Exposure Function)) and OAM (Operations, Administration and Maintenance).

[0038] Optionally, the DCCF (Data Collection and Coordination Function) can be used for analysis and data collection.

[0039] Retrieve information from data repositories (e.g., retrieve UDRs related to users via UDM or retrieve PFD information via NEF (PFDF));

[0040] Collect location information data from the LCS system;

[0041] Optionally, information can be stored and retrieved from the ADRF (Analytics Data Repository Function).

[0042] Optionally, data can be analyzed and collected from the MFAF (Management Framework Adapter Function); information about network elements can be retrieved (e.g., network element-related information can be retrieved from the NRF (Network Repository Function); and analysis can be provided to consumers on demand.

[0043] Provides batch data related to the analysis ID, provides accuracy information for the analysis ID, and provides ML model accuracy information or ML model accuracy degradation indicators.

[0044] A single instance or multiple instances of NWDAF can be deployed in a Public Land Mobile Network (PLMN). NWDAF can include the following logical functions:

[0045] The Analytics Logic Function (AnLF) is a logic function in NWDAF used to perform inference, derive analytical information (i.e., derive statistical data and / or predictions based on analytical consumer requests), and expose analytical services (i.e., Nnwdaf_AnalyticsSubscription or Nnwdaf_AnalyticsI network elements).

[0046] Model Training Logic Function (MTLF): A logic function in NWDAF used to train machine learning (ML) models and expose new training services (such as providing pre-trained ML models).

[0047] An NWDAF can contain an MTLF or an AnLF, or both of these logical functions.

[0048] The Data Collection and Control Function (DCCF) is also a network element on the 5GC control plane. DCCF is responsible for coordinating the collection and distribution of data requested by network element consumers. It prevents data sources from processing multiple subscriptions to the same data and prevents the sending of multiple notifications containing the same information due to incoordination of data consumer requests.

[0049] DCCF applies to at least one of the following:

[0050] NWDAF requests data from a data source (such as for computational analysis);

[0051] Requesting analysis of network element consumers from NWDAF data sources;

[0052] Network element consumers that request data from ADRF data sources;

[0053] ADRF that receives data from network data sources.

[0054] Figure 2 is a schematic diagram of the 5G system architecture of ADRF according to related technologies. As shown in Figure 2, the 5G system architecture allows ADRF to store and retrieve collected data and analytics. The 5G system architecture allows ADRF to store and retrieve collected data and analytics, supporting the following options:

[0055] ADRF exposes the Nadrf service for storing and retrieving data for other 5GC network functions (such as NWDAF), and data can be accessed through the Nadrf service. The Nadrf service is a network service interface provided by ADRF in the 5G network architecture.

[0056] Based on network function requests or configurations on the DCCF, the DCCF can determine the ADRF and interact with it directly or indirectly to request or store data. Interaction methods may include:

[0057] Directly: DCCF requests data to be stored in ADRF via the Nadrf service or Ndccf_DataManagement_Notify (e.g., when ADRF requests DCCF to perform a data collection notification). Additionally, DCCF retrieves data from ADRF via the Nadrf service.

[0058] Indirectly: DCCF requires the message frame to store data in ADRF via the Nadrf service or the Nmfaf_3daDataManagement_Co element igure. The message frame may contain one or more adapters for conversion between 3GPP-defined protocols.

[0059] The internal logic of the message frame is not within the 3GPP scope; only the interface between MFAF and other network functions defined by 3GPP is within the 3GPP scope.

[0060] Consumer network functions can be specified in the request sent to DCCF, and data provided by the data source needs to be stored in ADRF.

[0061] ADRF stores data received directly from Nadrf_DataManagement_StorageRequest sent by the network function, or from Ndccf_DataManagement_Notify / Nmfaf_3caDataManagement_Notify or Nnwdaf_DataManagement_Notify sent by DCCF, MFAF or NWDAF.

[0062] This embodiment provides a task allocation method running on a network architecture. Figure 3 is a flowchart of the task allocation method according to an embodiment of this disclosure. As shown in Figure 3, the method is applied to a first entity, and the process includes the following steps:

[0063] Step S302: Receive a task request from the second entity to execute the target task;

[0064] Step S304: Determine the target entity that meets the computing power requirements for executing the target task based on the computing power information of multiple third entities;

[0065] Step S306: Send the task request to the target entity for processing.

[0066] By following the steps above, we can solve the problem in related technologies where some network elements have high load but low computing power, while others have low load but high computing power. This ensures that tasks are assigned to entities with sufficient computing power, thereby improving the efficiency and success rate of task execution.

[0067] The first subject can be a task coordination subject, the second subject can be a consumer network element, the third subject can be an NWDAF, and the target subject is one of the third subjects, but is not limited to this.

[0068] The embodiments disclosed herein are applicable to large-scale distributed computing environments, such as cloud computing and edge computing, and can effectively manage computing resources and avoid resource waste.

[0069] Preferably, determining the third entity that meets the computing power requirements for executing the target task based on the computing power information of multiple third entities includes: sending a discovery request to a fourth entity, wherein the discovery request carries service type and computing power information; receiving a list of task execution entities from the fourth entity, wherein the list of task execution entities includes computing power information of multiple third entities; and matching a target entity that meets the computing power requirements for executing the target task from the list of task execution entities based on the computing power requirements carried in the task request. The fourth entity can be an NRF (Network Data Function), and the list of task execution entities can specifically be a NWDAF (Network Data Function List). In this way, the first entity can quickly locate a suitable task execution entity, improving the accuracy and speed of task allocation, which is particularly suitable for scenarios that require task allocation across multiple computing nodes, such as the Internet of Things (IoT) and big data analysis. Further, the task execution entity is a Network Data Function (NWDAF).

[0070] In some embodiments, after step S302 above, the computing power required for the target task can be estimated based on at least one of the priority information, service type, and task type carried in the task request. By estimating the computing power requirements of the task, resources can be allocated more rationally, avoiding task delays or failures caused by insufficient computing power. This is suitable for scenarios that require resource pre-allocation, such as online games and real-time video processing.

[0071] Furthermore, at least one of the priority information, service type, and task type is input into a pre-trained target inference model to obtain the computing power required for the target task, as output by the target inference model. This method utilizes machine learning techniques to more accurately estimate the computing power requirements of a task, and is suitable for scenarios with complex and variable computing power requirements, such as artificial intelligence and deep learning.

[0072] In this embodiment of the disclosure, step S306 may specifically include: sending a task allocation message to the target entity, wherein the task allocation message includes a task request. This approach ensures that the task request is accurately conveyed to the target entity, improving the accuracy of task execution. It is suitable for scenarios requiring precise task scheduling, such as autonomous driving and smart healthcare.

[0073] In some embodiments, if there are multiple task requests, the task allocation message carries priority information for each request; and / or if the same task request is assigned to multiple different third entities, the task allocation message carries the same task completion deadline information. This approach ensures that high-priority tasks are processed first, while setting task completion deadlines guarantees timely completion. It is suitable for scenarios requiring the processing of a large number of tasks and ensuring service quality, such as e-commerce and online education.

[0074] In some embodiments, after step S306 above, a task response sent by the target entity can also be received, wherein the task response includes accepting the request, or rejecting the request and a reason for rejection. This method can monitor the status of task allocation in real time and adjust the task allocation strategy in a timely manner, and is suitable for scenarios that require real-time task monitoring and management, such as data centers and cloud computing platforms.

[0075] Furthermore, the rejection reason may be at least one of the following: insufficient computing power or insufficient memory. Understanding the rejection reason helps in optimizing resource allocation and avoiding similar problems in the future. This is applicable to scenarios that require resource optimization and management, such as high-performance computing and virtual reality applications.

[0076] In some embodiments, in response to a task response to an acceptance request, the task processing result sent by the target subject is received. Alternatively, a third subject may directly return the task processing result to a second subject. This approach ensures timely feedback of task processing results, improves the transparency and traceability of task processing, and is suitable for scenarios requiring feedback of task processing results, such as financial transactions and logistics management.

[0077] In other embodiments, an update request for updated computing power information is received from one or more third entities; the computing power information of one or more third entities is updated according to the update request. This approach enables real-time updates of computing power information, ensuring the accuracy and real-time nature of computing resources, and is suitable for scenarios requiring dynamic resource management, such as dynamic load balancing and resource optimization.

[0078] In other embodiments, a deletion request or deregistration request for deleting computing power information is received from one or more third entities; the computing power information of one or more third entities is deleted according to the deletion request or deregistration request. This approach can promptly clean up invalid computing power information, avoid errors in resource allocation, and is suitable for scenarios that require resource cleanup and management, such as cloud services and virtualization environments.

[0079] In other embodiments, the result of deleting computing power information is indicated to one or more third entities. This approach ensures transparency in computing power information management, facilitates resource self-management and optimization by third entities, and is suitable for scenarios requiring transparent resource management, such as enterprise-level cloud services and resource management platforms.

[0080] Furthermore, computing power information can include computing power status, which helps to allocate resources more accurately and avoid waste of computing power resources. It is suitable for scenarios that require monitoring of computing power status, such as data center management and high-performance computing clusters.

[0081] The following example illustrates the embodiments of this disclosure, with the first entity being the task collaboration entity, the second entity being the consumer network element, and the third entity being the NWDAF.

[0082] Figure 4 is a flowchart of NWDAF computing power status reporting and NWDAF selection according to an embodiment of the present disclosure. As shown in Figure 4, it includes:

[0083] S401, NWDAF internally has changes in computing power, and decides to send or update the computing power status;

[0084] S402, NWDAF sends a computing power status registration or update (e.g., computing power shortage, computing power idle, computing power busy, etc.) to NRF or the task coordination entity;

[0085] S403, NRF / Task Coordination Entity stores NWDAF computing power status;

[0086] S404, the consumer NF sends an NWDAF discovery request to the NRF or task coordinator, indicating whether it is necessary to discover high-computing-power NWDAFs to execute high-computing-power AI / ML tasks, and the number and priority information of AI / ML tasks.

[0087] S405, NRF matches the appropriate NWDAF based on the computing power status sent by NWDAF and the computing power information and priority information in the model / analysis discovery request, and returns one or more NWDAFs to the model / analysis consumer;

[0088] S406, NWDAF requests NRF or the task coordinator to delete the computing power status / deregister;

[0089] S407, NRF or the task coordinating entity returns an operation result indication.

[0090] Figure 5 is a flowchart of real-time NWDAF matching according to an embodiment of the present disclosure, as shown in Figure 5, including:

[0091] S501, one or more consumer NFs send one or more AL / ML task requests to the task coordinating entity, which include priority information, service type (e.g., analytics ID), and AI / ML training task type (e.g., model training, model testing, analytics generation, etc.). If this message contains multiple task requests, the model / analysis consumer NFs are labeled with absolute priority information (priority order).

[0092] S502, the task coordination entity estimates the required computing power for the requested task based on the priority information, service type, task type, and other information received in step S501 (the required computing power can be generated based on model inference, and the inference input data is the information received in step S501), and sends an NWDAF discovery request to the NRF. This discovery request includes the service type (e.g., analysis ID) and computing power information / computing power status.

[0093] S503, NRF returns a list of NWDAFs, which includes one or more NWDAFs.

[0094] S504, the task coordination entity will match the appropriate NWDAF from the NWDAF list based on factors such as computing power and priority of the task request received in step S501.

[0095] In step S505, the task coordinating entity sends an AL / ML task to the NRF. The task coordinating entity then distributes the task requests received in step S501 to one or more NWDAFs via a message. This message contains consumer address information, priority information, service type (e.g., analytics ID), AI / ML training task type (e.g., model training, model testing, analytics generation, etc.), and task completion deadline information. If the message distributing tasks to the same NWDAF contains multiple task requests, the task coordinating entity marks the absolute priority information (priority order).

[0096] In this case, a single task (e.g., model training, model testing, analysis generation) can be decomposed and assigned to multiple different NWDAFs for processing. In this case, the task coordination entity needs to send the same task completion deadline information to each NWDAF.

[0097] S506, NWDAF sends a task request to the task coordinating entity, either accepting or rejecting the task, along with a reason (e.g., insufficient computing power or insufficient memory). This message may contain information about the remaining computing power or an update on the computing power status after NWDAF accepts or rejects the task.

[0098] S507 and NWDAF are used for task processing (e.g., model training, model testing, analysis and generation).

[0099] After completing the task, S508 NWDAF returns the processing results, which may include analysis results, trained models, model test results, etc.

[0100] This embodiment of the disclosure achieves efficient allocation of tasks and computing resources through intelligent analysis and matching, avoiding waste of computing resources and improving task processing efficiency and response speed. In particular, by predicting task request demands and computing power requirements at different times in the future, resource planning and allocation are performed in advance, further enhancing the system's flexibility and stability. This allows for better handling of sudden task requests, ensuring service quality and user experience. Furthermore, this method supports dynamic updating and management of computing power information, ensuring the real-time nature and accuracy of computing resources, and providing strong technical support for task allocation in large-scale network environments. The application of this technology not only improves resource utilization and reduces operating costs but also enhances service quality and user experience. It is applicable to various scenarios requiring efficient resource allocation and management, such as 5G networks, the Internet of Things, big data analytics, and artificial intelligence. In these scenarios, intelligent task allocation can achieve optimal resource utilization, improve overall system performance, and provide a solid technical foundation for future large-scale network applications.

[0101] Another aspect of this disclosure provides a task allocation method. Figure 6 is a flowchart of a task allocation method according to an embodiment of this disclosure. As shown in Figure 6, the method is applied to a first entity and includes:

[0102] Step S602: Receive first information sent by the third subject and / or the second subject;

[0103] Step S604: Based on the first information, predict the task request demand and / or computing power demand for different time periods in the future to obtain task processing suggestions and / or task request suggestions;

[0104] Step S606: Send task processing suggestions and / or task request suggestions to the third entity and / or the second entity.

[0105] Through the above steps S602 to S606, the rational planning of network resources and the efficient allocation of real-time tasks can be achieved. This enables the forward-looking and proactive allocation of tasks and computing resources, better copes with potential future task peaks, and ensures the continuity and stability of services. It is suitable for scenarios that require future resource planning, such as network traffic prediction and resource demand prediction.

[0106] Among them, the first subject executing the above steps can be a task coordination subject, the second subject can be a consumer network element, the third subject can be an NWDAF, and the target subject is one of the third subjects, but is not limited to these.

[0107] In this embodiment of the disclosure, step S602 may include at least one of the following:

[0108] S6021, receive task prediction information generated by the second subject, wherein the task prediction information carries predicted task request time information, predicted task request quantity information or predicted task type information.

[0109] Based on historical data and current network conditions, machine learning algorithms are used to predict AI / ML task request patterns at different times in the future. This prediction information may include: Task request timing: predicting peak and off-peak periods for AI / ML task requests on a future day or week.

[0110] Task request count: Predicts the number of requests for each type of task (such as model training, model inference, data analysis, etc.) within a specific future time period.

[0111] Task type: Prediction requests will primarily come from which types of tasks? For example, requests for expected model training tasks will increase significantly.

[0112] S6022, receive task planning information generated by the second subject, wherein the task planning information carries planned task request time information, planned task request quantity information or planned task type information.

[0113] After obtaining task prediction information, task planning information may be generated, typically based on the predicted task requirements and network resource allocation strategies. Planning information may include:

[0114] Planned task request times: Based on forecasts, set the optimal execution time for each task to optimize resource utilization;

[0115] Planned number of task requests: Based on resource availability, plan the number of tasks that can be processed in each time period;

[0116] Planned task types: Based on the degree of matching between computing power type and task type, plan which types of tasks will be carried out at which times.

[0117] S6023, Receive computing power prediction information generated by a third entity, wherein the computing power prediction information carries time information or predicted computing power information;

[0118] The task coordinator predicts available computing resources in the network for different time periods based on historical data and current network load. Computing resource prediction information may include:

[0119] Time information: Predicting the time points when computing power changes, for example, predicting whether computing power demand will decrease or increase during specific periods of the day;

[0120] Predicted computing power information: Predicts the total amount or remaining amount of computing power (CPU (Central Processing Unit), GPU (Graphics Processing Unit), memory, etc.) available in the network within a specific future period.

[0121] S6024, receive computing power planning information generated by a third entity, wherein the computing power planning information carries time information or planned computing power information.

[0122] Based on computing power forecasts, the task coordination entity plans computing power allocation and task scheduling strategies for different time periods in the future. Planning information may include:

[0123] Time information: The execution time planned for each task to make full use of the network's high-efficiency computing power during peak periods;

[0124] Planned computing power information: The computing power resources planned for each task or task type to ensure that the task can be executed with sufficient computing power support.

[0125] The above methods can collect more comprehensive forecasting and planning information, providing richer data support for resource allocation. They are suitable for scenarios that require multi-dimensional resource planning, such as large data centers and cloud computing platforms.

[0126] In this embodiment of the disclosure, step S604 may include at least one of the following: determining task processing suggestions for different time periods in the future based on task prediction information; and determining request suggestions for different time periods in the future based on computing power prediction information. In this way, the first entity can plan computing power resources in advance to ensure sufficient resources to cope with peak task periods. Simultaneously, it can adjust task request strategies based on computing power resource predictions to avoid excessive consumption of computing power resources. This approach is suitable for scenarios requiring dynamic matching of resources and tasks, such as online games and real-time video processing.

[0127] The following example illustrates the embodiments of this disclosure, with the first entity being the task collaboration entity, the second entity being the consumer network element, and the third entity being the NWDAF.

[0128] Figure 7 is a flowchart of the overall planning of future time period tasks according to an embodiment of the present disclosure. As shown in Figure 7, it includes:

[0129] S701, the Consumer Network Function (NF, as the task requester) generates (can use the model to generate and predict) relevant information about task requests that will occur at different times in the future (e.g., the number of AI / ML tasks requested in a certain time period in the future, the type of AI / ML tasks (e.g., model training / inference & analysis generation / testing), and the type of AI-related analysis service (e.g., analysis ID)) and sends the task request prediction to the task coordinating entity.

[0130] S702, NWDAF (as the task completer) generates (can use the model to generate predictions) available computing power information for different future time periods, and sends the available computing power predictions to the task coordination entity.

[0131] S703, the task coordination entity performs task processing, specifically by statistically analyzing the relevant information of task requests that will occur at different times in the future (listed in step S701) received from different network elements in step S701, or by training a model and performing inference to generate task processing suggestions.

[0132] S704, the task coordination entity sends task processing suggestions for future time periods to NWDAF (e.g., how much computing power to reserve during what time period, what type of AI / ML tasks are likely to be received during what time period, how many or few (or task quantity predictions) AI / ML tasks are likely to be received during what time period, and the types of tasks, and what time period may be affected by computing power shortages).

[0133] S705, the task coordination entity sends task processing suggestions for future time periods to the consumer network element (as the task requester) (e.g., suggestions on when to request tasks, the number of tasks to request, and the type of AI / ML tasks to request; when there will be a shortage of computing power in the future, which will cause the task processing speed to slow down; and when there will be sufficient computing power in the future, which will result in a faster task processing speed).

[0134] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this disclosure.

[0135] This embodiment also provides a task allocation device for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0136] Figure 8 is a structural block diagram of a task allocation device according to an embodiment of the present disclosure. As shown in Figure 8, the device is applied to a first main body and includes:

[0137] The first receiving module 82 is configured to receive a task request sent by the second subject to execute the target task;

[0138] The determination module 84 is configured to determine the target entity that meets the computing power requirements for executing the target task based on the computing power information of multiple third entities;

[0139] The first sending module 86 is configured to send the task request to the target subject for processing.

[0140] Figure 9 is a structural block diagram of a task allocation device according to an embodiment of the present disclosure. As shown in Figure 9, the device is applied to a first main body and includes:

[0141] The second receiving module 92 is configured to receive first information sent by the third subject and / or the second subject;

[0142] Prediction module 94 is configured to predict the task request demand and / or computing power demand for different time periods in the future based on the first information, and obtain task processing suggestions and / or task request suggestions;

[0143] The second sending module 96 is configured to send the task processing suggestion and / or the task request suggestion to the third subject and / or the second subject.

[0144] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0145] Embodiments of this disclosure also provide a computer-readable storage medium storing a computer program configured to perform the steps in any of the above method embodiments when executed.

[0146] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0147] Embodiments of this disclosure also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0148] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0149] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.

[0150] It is obvious to those skilled in the art that the modules or steps of this disclosure described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this disclosure is not limited to any particular combination of hardware and software.

[0151] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Various modifications and variations can be made to this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A task allocation method, applied to a first subject, comprising: Receive a task request from a second entity to execute the target task; The target entity that meets the computing power requirements for executing the target task is determined based on the computing power information of multiple third entities; The task request is sent to the target entity for processing.

2. The method according to claim 1, wherein, Based on the computing power information of multiple third-party entities, the third-party entities that meet the computing power requirements for executing the target task include: A discovery request is sent to a fourth entity, wherein the discovery request carries service type and computing power information; Receive the task execution entity list of the fourth entity, wherein the task execution entity list includes computing power information of multiple third entities; Based on the computing power requirements carried in the task request, a target entity that meets the computing power requirements for executing the target task is matched from the list of task execution entities.

3. The method according to claim 2, wherein, The task execution entity is the Network Data Analysis Function (NWDAF).

4. The method according to claim 1, wherein, After receiving a task request from a second entity to execute a target task, the method further includes: Estimate the computing power required for the target task based on at least one of the priority information, service type, and task type carried in the task request.

5. The method according to claim 4, wherein, Estimating the computing power required for the target task based on at least one of the priority information, service type, and task type carried in the task request includes: Input at least one of the priority information, the service type, and the task type into a pre-trained target inference model to obtain the computing power required for the target task as output by the target inference model.

6. The method according to claim 1, wherein, Sending the task request to the target entity for processing includes: A task allocation message is sent to the target entity, wherein the task allocation message includes the task request.

7. The method according to claim 6, wherein, If there are multiple task requests, the task allocation message carries priority information for each task request; and / or If the same task request is assigned to multiple different third entities, the task assignment message carries the same task completion time limit information.

8. The method according to claim 1, wherein, After sending the task request to the target entity for processing, the method further includes: Receive a task response sent by the target entity, wherein the task response includes accepting the request, or rejecting the request and a reason for rejection.

9. The method according to claim 8, wherein, The reason for rejection is at least one of the following: insufficient computing power or insufficient memory.

10. The method according to claim 8, wherein, The method further includes: In response to the task response being an acceptance request, the task processing result sent by the target entity is received.

11. The method according to claim 1, wherein, The method further includes: Receive update requests for updated computing power information from one or more third entities; The computing power information of the one or more third entities is updated according to the update request.

12. The method according to claim 11, wherein, The method further includes: Receive deletion requests or deregistration requests for the computing power information from one or more third entities; The computing power information of the one or more third entities shall be deleted according to the deletion request or the deregistration request.

13. The method according to claim 12, wherein, The method further includes: Instruct the one or more third entities to perform the operation of deleting the computing power information.

14. The method according to any one of claims 1 to 13, wherein, The computing power information includes the computing power status.

15. A task allocation method, applied to a first subject, comprising: Receive first information sent by a third entity and / or a second entity; Based on the first information, the task request demand and / or computing power demand for different time periods in the future are predicted to obtain task processing suggestions and / or task request suggestions. Send the task processing suggestion and / or the task request suggestion to the third entity and / or the second entity.

16. The method according to claim 15, wherein, Receiving the first information sent by the third subject and / or the second subject includes at least one of the following: Receive task prediction information generated by the second subject, wherein the task prediction information carries predicted task request time information, predicted task request quantity information, or predicted task type information. Receive task planning information generated by the second entity, wherein the task planning information carries planned task request time information, planned task request quantity information, or planned task type information; Receive computing power prediction information generated by the third entity, wherein the computing power prediction information carries time information or predicted computing power information; The system receives computing power planning information generated by the third entity, wherein the computing power planning information carries time information or planned computing power information.

17. The method according to claim 16, wherein, Based on the first information, the task request demand and / or computing power demand for different future time periods are predicted, and task processing suggestions and / or task request suggestions are obtained, including at least one of the following: Based on the task prediction information, determine the task processing suggestions for different time periods in the future; The request recommendations for different time periods in the future are determined based on the computing power prediction information.

18. A computer-readable storage medium storing a computer program, wherein, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1 to 14, 15 to 17.

19. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the steps of the method according to any one of claims 1 to 14, 15 to 17.

20. A computer program product comprising a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1 to 14, 15 to 17.

Images