Computing power scheduling management method, baseband processing unit and communication system

Abstract

The invention provides a computing power scheduling management method, a baseband processing unit and a communication system. The method comprises the following steps: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal comprises computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; scheduling the task data according to the service matching request, the task data acquisition signal and a preset task scheduling algorithm, and returning a task result corresponding to the task data and a docking related signal; the docking related signal is a signal for continuously acquiring task data or a docking backup service proxy signal. According to the invention, computing power support can be flexibly provided for task data, and the resource utilization rate of the base station is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a computing power scheduling management method, a baseband processing unit and a communication system. Background Art

[0002] With the continuous advancement of communication technology, base stations have become more powerful and can undertake more and more tasks. The hardware specifications of base stations are usually planned according to peak capacity, but in actual operation, the load of base stations is often much smaller than the peak load, causing hardware equipment to be idle and thus causing resource waste. Summary of the invention

[0003] The present invention provides a computing power scheduling management method, a baseband processing unit and a communication system to solve the defect of waste of base station computing power resources in the prior art. The present invention can flexibly provide computing power support for task data and improve the resource utilization rate of the base station.

[0004] The present invention provides a computing power scheduling management method, which is applied to a service agent, and the method includes: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; scheduling task data according to the service matching request, the task data acquisition signal and a preset task scheduling algorithm, and sending back a task result and a docking-related signal corresponding to the task data; the docking-related signal is a signal for continuing to obtain task data or a signal for docking a backup service agent; the signal for continuing to obtain task data is a signal that prompts the current service agent to continue to obtain task data when the computing power service resources of the current service agent are valid; the signal for docking a backup service agent is a signal that prompts the backup service agent to determine the task result corresponding to the task data when the computing power service resources of the current service agent are invalid.

[0005] According to a computing power scheduling management method provided by the present invention, the service matching request includes a first service registration request; the first service registration request includes a first task type and a first task agent identifier; the service matching response includes a first service registration response; generating a service matching response based on the service matching request includes: generating the first service registration response in a first state based on the first task type and the first task agent identifier; the first state is that there is a service instance corresponding to the first task type locally; the first service registration response is used to send back the first service matching identifier.

[0006] According to a computing power scheduling management method provided by the present invention, the service matching request includes a second service registration request and a service establishment request; the second service registration request includes a second task type and a second task agent identifier; the service matching response includes a service resource response and a second service registration response; generating a service matching response based on the service matching request includes: generating the service resource response in a second state based on the second task type and the second task agent identifier; the second state is that there is no local service instance corresponding to the second task type; the service resource response is used to send back the local service resource state; receiving the service establishment request, and generating the second service registration response based on the service establishment request, the second task type and the second task agent identifier; the second service response is used to send back the second service matching identifier.

[0007] According to a computing power scheduling management method provided by the present invention, the task data is scheduled according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and the task result corresponding to the task data is sent back, including: when the computing power budget information is a positive value, the first task data is acquired according to the first service matching request and the first task data acquisition signal, and the first task result corresponding to the first task data determined by the first computing power service resource is sent back; the computing power budget information is a positive value, indicating that the first computing power service resource is valid; when the computing power budget information is a non-positive value, the acquisition of the second task data is stopped based on the second task data acquisition signal, and when there is a second computing power service resource, the second task data is re-acquired, and the second task result corresponding to the second task data determined by the second computing power service resource is sent back; the computing power budget information is a non-positive value, indicating that the second computing power service resource is invalid.

[0008] The present invention also provides a computing power scheduling management method, which is applied to task agents, and the method includes: sending a service matching request; when a service matching response and a task data acquisition signal are received, determining the task data according to the service matching response and the task data acquisition signal, and receiving the task result and docking-related signals corresponding to the task data; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking-related signal is a signal for continuing to obtain task data or a signal for docking with a backup service agent; the signal for continuing to obtain task data is a signal that prompts to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal for docking with a backup service agent is a signal that prompts to determine the task result corresponding to the task data through a backup service agent when the computing power service resources of the current service agent are invalid.

[0009] According to a computing power scheduling management method provided by the present invention, the service matching response is generated by a plurality of service agents; and the sending window information of the task data is dynamically adjusted along with the computing power budget information.

[0010] The present invention also provides a baseband processing unit, including a service agent and a task agent; the service agent is used to execute the steps of the above-mentioned computing power scheduling management method, and the task agent is used to execute the steps of the above-mentioned computing power scheduling management method.

[0011] The present invention also provides a communication system, comprising a plurality of communication clusters, each of which comprises a plurality of the above-mentioned baseband processing units.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a computing power scheduling management method as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements any of the computing power scheduling management methods described above.

[0014] The present invention provides a computing power scheduling management method, a baseband processing unit and a communication system, the method comprising: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; scheduling the task data according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and sending back the task result and docking related signal corresponding to the task data; the docking related signal is a signal for continuing to acquire task data or a docking backup service proxy signal. The present invention can flexibly provide computing power support for task data, thereby improving the resource utilization of the base station. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of a computing power scheduling management method provided by the present invention.

[0017] Figure 2It is a specific flow chart of a computing power scheduling management method provided by the present invention.

[0018] Figure 3 It is a schematic diagram of the principle of a baseband processing unit provided by the present invention.

[0019] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Considering that traditional communication base stations are dedicated devices that can only handle the signal processing tasks of the base station itself. With the development of 5G technology, vRAN (Virtualized Radio Access Network) was introduced. The processing functions on the original BBU (Base Band Unit) were separated from the dedicated hardware through software-defined methods and run on general-purpose servers, which laid the foundation for running general-purpose computing tasks on the BBU (i.e., endogenous computing power). In addition, the high-specification hardware of the base station is usually planned according to the peak capacity, but in actual operation, the load of the base station is often much less than the peak load, causing the hardware equipment to be idle, which provides the necessity for endogenous computing power.

[0022] In the embodiment of the present invention, a video stream inference scenario for a 5G computing power endogenous system is taken as an example for explanation.

[0023] Considering the existing technology, the main challenges of real-time inference of video streams on BBU are: the communication protocol stack load is highly volatile, which may preempt the resources of the inference module at any time, resulting in inference interruption and failure to meet SLO requirements; the video stream will bring both communication load (acting on the communication protocol stack) and computing load (acting on the inference module), and the two will increase or decrease at the same time, causing resource preemption conflicts.

[0024] Please refer to Figure 1 , Figure 1 A flowchart of a computing power scheduling management method provided by the present invention.

[0025] The present invention provides a computing power scheduling management method, which is applied to a service agent, and the method comprises: 101: receiving a service matching request; 102: Generate a service matching response according to the service matching request and a preset service matching algorithm; 103: Generate a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; 104: Schedule the task data according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and send back the task result and docking-related signal corresponding to the task data; the docking-related signal is a signal for continuing to obtain task data or a signal for docking with a backup service agent; the signal for continuing to obtain task data is a signal for prompting to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal for docking with a backup service agent is a signal for prompting to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

[0026] In order to solve the technical problems existing in the prior art, the present invention provides a computing power scheduling management method. UE (User Equipment) generates a video stream and uploads the video through the base station to which it is connected (by default, a UE is only connected to one base station, and the corresponding relationship depends on the location of the UE). For the convenience of description, the base station is recorded as a "directly connected base station" and its BBU is recorded as a "directly connected BBU". At this time, the video bit stream will first be decoded by the decoder (task agent) on the directly connected BBU and converted into a series of video frames (task data, that is, a series of coherent pictures). Then these video frames will first be cached in the decoder.

[0027] In order to be forwarded immediately to the service agent that matches the decoder, the task agent needs to establish a matching relationship with the service agent. Specifically, the task agent will send a service matching request (which may include the globally unique decoder ID (task agent identifier) ​​in the cluster, which is used to return the task result) to the service agent. After receiving the service matching request, the service agent generates a service matching response based on the service matching request and the preset service matching algorithm, thereby establishing a matching relationship with the task agent.

[0028] After the matching relationship is established for the first time, the service agent will send a task data acquisition signal to the decoder. The task data acquisition signal includes computing power budget information, that is, the number of video frames that can be processed at present (depending on the number of computing power service resources that can work locally). After receiving the task data acquisition signal, the decoder starts to forward the video frames to the service agent that matches the decoder. The service agent can maintain a cache queue for different video inference models to store video frames. As long as the video stream uses the same model for inference, it will be placed in the same queue. The service agent sends the local status to the decoder in the form of computing power budget information for congestion control. The service agent monitors the status of local computing power service resources and promptly informs the decoder to re-match when the computing power service resources fail (return the docking backup service agent signal, which is a signal that prompts the backup service agent to determine the task result corresponding to the task data when the computing power service resources of the current service agent fail). When the computing power service resources of the current service agent are valid, the service agent returns a signal to continue to obtain task data, prompting the current service agent to continue to obtain task data.

[0029] The service agent adjusts the size of the thread pool (the thread pool maintained for different inference models) of the local computing service resources, and the threads in the thread pool will occupy the idle CPU core of the BBU server (that is, the part not used by the communication protocol stack) for inference calculation. The service agent allocates task data to the computing service resources in a pull manner, that is, when a computing service resource completes the processing of the previous task, it will actively pull tasks from its corresponding queue in the service agent. After the computing service resource processes a frame of image, the service agent will return the task result and docking-related signals to the corresponding decoder according to the decoder ID of the frame. After receiving the result, the decoder returns the result to the UE and deletes the corresponding cached frame in the decoder queue. The present invention can flexibly provide computing power support for task data, improve the resource utilization of the base station, and can deploy video inference tasks in a computing power endogenous manner in the BBU server cluster, and maintain the corresponding SLO (Service Level Objective, service level target, such as 30fps).

[0030] It should be noted that: The method of the present invention can provide idle computing resources for various application services, such as FL (Federated Learning) applications, face recognition applications, etc.

[0031] This embodiment is illustrated using a 5G (5th Generation Mobile Communication Technology) computing network system. Those skilled in the art will appreciate that, in other application scenarios, the present invention may also be applicable to other mobile communication networks, such as future networks, i.e., new generation networks, such as B5G (Beyond 5G) or 6G (6th Generation Mobile Communication Technology) networks.

[0032] Computing power is a general term for data processing capabilities and services. In a narrow sense, computing power refers to computing resources composed of a variety of chips and components. In a broad sense, computing power also includes upper-level services after computing power encapsulation.

[0033] A base station refers to a radio transceiver station that transmits information to mobile phone terminals through a mobile communication exchange center within a certain radio coverage area. It is mainly used to transmit and receive wireless communication signals to ensure that mobile devices can communicate normally.

[0034] BBU is mainly responsible for processing baseband signals, which carry user data (such as voice, text messages, network data, etc.) and control information.

[0035] vRAN virtualizes the functions of traditional radio access network (RAN) through software-defined approach.

[0036] Endogenous computing power is a concept that organically integrates computing power into every aspect of the system architecture, allowing computing power to exist naturally in the system as if it were innate, thereby achieving efficient allocation and utilization of computing resources. This embodiment specifically refers to general computing in the base station BBU server.

[0037] Video streaming refers to the transmission of video data in the form of a continuous data stream. The user end can receive and play the video content while the data is being transmitted, rather than waiting for the entire video file to be downloaded before playing it.

[0038] Video reasoning is a technology that analyzes and infers video data. It is not just a simple recognition of the images in the video, but also uses the time series characteristics of the video, such as the movement trajectory of the object, the sequence of actions, and other information, to infer higher-level semantic information through pre-trained deep learning models (such as 3D convolutional neural networks, etc.).

[0039] SLO is a clear and measurable goal for service quality agreed upon between a service provider and a user or between different departments within an organization. SLO is mainly used to define the performance standards that a service should achieve in multiple dimensions.

[0040] Computing service resources may refer to data processing capabilities that can support application or business requirements, such as the processing capabilities of various processors, processing hardware, etc. (e.g., CPU, GPU, etc.). Of course, computing resources may correspond to various processors, processing hardware, etc. that provide processing capabilities, which will not be described in detail in the present invention.

[0041] The service agent may include but is not limited to at least one of a network controller and an edge computing controller.

[0042] Various devices that utilize computing power, such as terminal devices, may be "UE", which generally refers to terminals such as mobile phones and security cameras, and is a device that generates video streams, and the present invention does not make any special limitation here.

[0043] The communication system may include several communication clusters, each of which includes multiple baseband processing units. Each baseband processing unit includes a decoder (task agent), a service agent, and a computing service resource. The communication between the devices in the communication system can be performed through any appropriate signal, protocol, channel, path, form, etc. known in wireless communication, as long as the computing power related information, computing power scheduling related information, and any other suitable data, etc. can be transmitted securely. The establishment of the computing power network, the scheduling of computing power, etc. can be signaled by a specific signal, which is, for example, a broadcast signal and / or a dedicated signal on a specific channel from a controller and / or a computing power provider, and can take an appropriate form known in wireless communication, which will not be described in detail here.

[0044] Please refer to Figure 2 , Figure 2 A schematic diagram of a specific process of a computing power scheduling management method provided by the present invention.

[0045] As a preferred embodiment, the service matching request includes a first service registration request; the first service registration request includes a first task type and a first task agent identifier; the service matching response includes a first service registration response; based on the service matching request and a preset service matching algorithm, a service matching response is generated, including: based on the first task type and the first task agent identifier, a first service registration response is generated in a first state; the first state is that there is a service instance corresponding to the first task type locally; the first service registration response is used to send back the first service matching identifier.

[0046] In this embodiment, when the user device initiates a request, its directly connected BBU will start a decoder process for it. If there is no corresponding inference instance on the BBU, a thread pool of computing service resources will be started for it and a queue will be allocated for it in the service agent. Otherwise, it will directly match the existing local queue. That is, by default, the user's video stream is first inferred on the directly connected BBU.

[0047] If the computing resources available on the directly connected BBU are insufficient, the decoder will broadcast the first service registration request, which includes the first task type and the first task agent identifier. Each service agent will listen on a specific port of UDP. After receiving the first service registration request, if there are computing service resources for the corresponding inference model locally (there are service instances corresponding to the first task type locally), the local IP and the listening port of the corresponding queue will be reported to the broadcast decoder (the first service registration response is generated in the first state, and the first service registration response is used to send back the first service matching identifier). After receiving the replies from each service agent, if there is a service registration response, the decoder will directly establish a matching relationship with the service agent.

[0048] As a preferred embodiment, the service matching request includes a second service registration request and a service establishment request; the second service registration request includes a second task type and a second task agent identifier; the service matching response includes a service resource response and a second service registration response; a service matching response is generated according to the service matching request and a preset service matching algorithm, including: generating a service resource response in a second state according to the second task type and the second task agent identifier; the second state is that there is no local service instance corresponding to the second task type; the service resource response is used to send back the local service resource state; a service establishment request is received, and a second service registration response is generated according to the service establishment request, the second task type and the second task agent identifier; the second service response is used to send back the second service matching identifier.

[0049] In this embodiment, if the computing resources available on the directly connected BBU are insufficient, the decoder will broadcast a second service registration request, which includes a second task type and a second task agent identifier. Each service agent will listen on a specific port of UDP. After receiving the second service registration request, if there are computing power service resources of the corresponding reasoning model locally (there are service instances corresponding to the second task type locally), a service resource response will be generated in the second state, and the service resource response is used to send back the local service resource state. After receiving the replies from each service agent, if there is no service registration response, but a service establishment request is received, the decoder will establish a request to the service agent with the most sufficient resources. The service agent will start the corresponding computing power service resources and allocate the service agent queue according to the service establishment request, the second task type and the second task agent identifier, and then send back the second service registration response to the decoder. The second service response is used to send back the second service matching identifier (local IP and the listening port of the corresponding queue), thereby establishing a matching relationship.

[0050] As a preferred embodiment, the task data is scheduled according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and the task result corresponding to the task data is sent back, including: When the computing power budget information is a positive value, the first task data is acquired according to the first service matching request and the first task data acquisition signal, and the first task result corresponding to the first task data determined by the first computing power service resource is returned; the computing power budget information is a positive value indicating that the first computing power service resource is valid; When the computing power budget information is a non-positive value, the acquisition of the second task data is stopped based on the second task data acquisition signal, and when there are second computing power service resources, the second task data is reacquired, and the second task result corresponding to the second task data determined by the second computing power service resources is returned; the computing power budget information is a non-positive value indicating that the second computing power service resource has failed.

[0051] In this embodiment, after the matching relationship is established for the first time, the service agent will send a task data acquisition signal to the decoder. The task data acquisition signal includes computing power budget information, that is, the number of video frames that can be processed currently (depending on the number of computing power service resources that can work locally). If the computing power budget information is a positive value, it means that the first computing power service resource is valid, and the first task data is acquired according to the first service matching request and the first task data acquisition signal (new frames can be sent), and the first task result corresponding to the first task data determined by the first computing power service resource is sent back. If the computing power budget information is a non-positive value, for example, if it is 0, it means that the computing power service resources will not increase, and if it is a negative value, it means that the computing power service resources will decrease, and the video frame sending expectation will be reduced. If the BBU server where the computing power service resource is located has a surge in the load of the communication protocol stack, resulting in a large number of computing power service resources being preempted, the service agent sends -9999 (second task data acquisition signal), so that the window value at the decoder is negative, indicating that resources are scarce, and the acquisition of the second task data is stopped. After the computing power resources are subsequently restored, the service agent resends the task data acquisition signal, resets the sending window, reacquires the second task data, and returns the second task result corresponding to the second task data determined by the second computing power service resource.

[0052] The present invention also provides a computing power scheduling management method, which is applied to task agents, and the method includes: sending a service matching request; when a service matching response and a task data acquisition signal are received, determining the task data according to the service matching response and the task data acquisition signal, and receiving the task result and docking related signals corresponding to the task data; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking related signal is a continue to obtain task data signal or a docking backup service agent signal; the continue to obtain task data signal is a signal that prompts to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the docking backup service agent signal is a signal that prompts to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

[0053] As a preferred embodiment, the service matching response is generated by a plurality of service agents; the sending window information of the task data is dynamically adjusted according to the computing power budget information.

[0054] In this embodiment, the decoder maintains a sending window for each service agent registered with it. Upon receiving the task data acquisition signal, the sending window size of the service agent is modified, and the video frame is forwarded. The decoder window value is reduced by 1 for each forwarded frame. The service agent subsequently affects the sending window by sending a signal to continue to acquire task data along with the result of the previous frame. The decoder sends frames according to the sending window sizes of different service agents. When the sending window of the service agent of the same BBU is large enough, it only sends frames locally; otherwise, it sends frames in proportion to the size of different sending windows. In order to ensure that the decoder can quickly switch to a new service agent when a service agent stops urgently, the decoder usually matches at least two service agents, one of which can only serve as a backup task agent and send a frame every 2 seconds, so that the service agent includes a signal to continue to acquire task data as a heartbeat packet when returning to maintain a matching relationship.

[0055] The baseband processing unit provided by the present invention is described below. The baseband processing unit described below and the computing power scheduling management method described above can be referenced to each other.

[0056] Please refer to Figure 3 , Figure 3 A schematic diagram of the principles of a baseband processing unit provided by the present invention.

[0057] The present invention also provides a baseband processing unit, including a service agent and a task agent; the service agent is used to execute the steps of the above-mentioned computing power scheduling management method, and the task agent is used to execute the steps of the above-mentioned computing power scheduling management method.

[0058] The communication system provided by the present invention is described below. The communication system described below and the computing power scheduling management method described above can be referenced to each other.

[0059] The present invention also provides a communication system, including a plurality of communication clusters, each of which includes a plurality of the above-mentioned baseband processing units.

[0060] Figure 4 An example of a structural diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor (processor) 410 , a communication interface (Communications Interface) 420 , a memory (memory) 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other through the communication bus 440 . The processor 410 can call the logic instructions in the memory 430 to execute the computing power scheduling management method, which is applied to the service agent, and the method includes: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to the computing power service resources; scheduling the task data according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and sending back the task results and docking-related signals corresponding to the task data; the docking-related signal is a signal to continue to obtain task data or a signal to dock with a backup service agent; the signal to continue to obtain task data is a signal to prompt to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal to dock with a backup service agent is a signal to prompt to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid. A computing power scheduling management method is implemented and applied to a task agent, the method comprising: sending a service matching request; upon receiving a service matching response and a task data acquisition signal, determining the task data according to the service matching response and the task data acquisition signal, and receiving a task result corresponding to the task data and a docking-related signal; the task data acquisition signal comprises computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking-related signal is a signal for continuing to obtain task data or a signal for docking a backup service agent; the signal for continuing to obtain task data is a signal for prompting to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal for docking a backup service agent is a signal for prompting to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

[0061] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0062] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the computing power scheduling management method provided by the above-mentioned methods, which is applied to a service agent, and the method includes: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to the computing power service resources; scheduling the task data according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and sending back the task results and docking-related signals corresponding to the task data; the docking-related signal is a signal for continuing to obtain task data or a signal for docking a backup service agent; the signal for continuing to obtain task data is a signal that prompts the current service agent to continue to obtain task data when the computing power service resources of the current service agent are valid; the signal for docking a backup service agent is a signal that prompts the backup service agent to determine the task result corresponding to the task data when the computing power service resources of the current service agent are invalid. A computing power scheduling management method is implemented and applied to a task agent, the method comprising: sending a service matching request; upon receiving a service matching response and a task data acquisition signal, determining the task data according to the service matching response and the task data acquisition signal, and receiving a task result corresponding to the task data and a docking-related signal; the task data acquisition signal comprises computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking-related signal is a signal for continuing to obtain task data or a signal for docking a backup service agent; the signal for continuing to obtain task data is a signal for prompting to continue to obtain task data through the current task agent when the computing power service resources of the current service agent are valid; the signal for docking a backup service agent is a signal for prompting to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

[0063] On the other hand, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the computing power scheduling management method provided by the above-mentioned methods, which is applied to a service agent, and the method includes: receiving a service matching request; generating a service matching response according to the service matching request and a preset service matching algorithm; generating a task data acquisition signal based on the service matching response; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to the computing power service resources; scheduling the task data according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and sending back the task results and docking-related signals corresponding to the task data; the docking-related signal is a signal for continuing to obtain task data or a signal for docking with a backup service agent; the signal for continuing to obtain task data is a signal that prompts the current service agent to continue to obtain task data when the computing power service resources of the current service agent are valid; the signal for docking with a backup service agent is a signal that prompts the backup service agent to determine the task result corresponding to the task data when the computing power service resources of the current service agent are invalid. A computing power scheduling management method is implemented and applied to a task agent, the method comprising: sending a service matching request; upon receiving a service matching response and a task data acquisition signal, determining the task data according to the service matching response and the task data acquisition signal, and receiving a task result corresponding to the task data and a docking-related signal; the task data acquisition signal comprises computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking-related signal is a signal for continuing to obtain task data or a signal for docking a backup service agent; the signal for continuing to obtain task data is a signal for prompting to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal for docking a backup service agent is a signal for prompting to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

[0064] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0065] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A computing power scheduling management method, characterized in that: Applied to a service agent, the method comprises: receiving a service matching request; Generate a service matching response according to the service matching request and a preset service matching algorithm; Based on the service matching response, a task data acquisition signal is generated; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; The task data is scheduled according to the service matching request, the task data acquisition signal and the preset task scheduling algorithm, and the task result and docking-related signal corresponding to the task data are sent back; the docking-related signal is a signal for continuing to obtain task data or a signal for docking with a backup service agent; the signal for continuing to obtain task data is a signal for prompting to continue to obtain task data through the current service agent when the computing power service resources of the current service agent are valid; the signal for docking with a backup service agent is a signal for prompting to determine the task result corresponding to the task data through a backup service agent when the computing power service resources of the current service agent are invalid.

2. The computing power scheduling management method according to claim 1, characterized in that: The service matching request includes a first service registration request; the first service registration request includes a first task type and a first task agent identifier; the service matching response includes a first service registration response; Generating a service matching response according to the service matching request and a preset service matching algorithm includes: generating the first service registration response in a first state according to the first task type and the first task agent identifier; The first state is that there is a service instance corresponding to the first task type locally; the first service registration response is used to return a first service matching identifier.

3. The computing power scheduling management method according to claim 1, characterized in that: The service matching request includes a second service registration request and a service establishment request; the second service registration request includes a second task type and a second task agent identifier; the service matching response includes a service resource response and a second service registration response; Generating a service matching response according to the service matching request and a preset service matching algorithm includes: generating the service resource response in a second state according to the second task type and the second task agent identifier; The second state is that there is no local service instance corresponding to the second task type; the service resource response is used to send back the local service resource state; The service establishment request is received, and the second service registration response is generated according to the service establishment request, the second task type and the second task agent identifier; the second service response is used to send back the second service matching identifier.

4. The computing power scheduling management method according to any one of claims 1 to 3, characterized in that: The step of scheduling the task data according to the service matching request, the task data acquisition signal and a preset task scheduling algorithm, and returning a task result corresponding to the task data, includes: When the computing power budget information is a positive value, first task data is acquired according to the first service matching request and the first task data acquisition signal, and a first task result corresponding to the first task data determined by the first computing power service resource is returned; the computing power budget information is a positive value indicating that the first computing power service resource is valid; When the computing power budget information is a non-positive value, the acquisition of the second task data is stopped based on the second task data acquisition signal, and when there is a second computing power service resource, the second task data is reacquired, and the second task result corresponding to the second task data determined by the second computing power service resource is returned; the computing power budget information being a non-positive value indicates that the second computing power service resource has failed.

5. A computing power scheduling management method, characterized in that: Applied to a task agent, the method comprises: Send service matching request; When a service matching response and a task data acquisition signal are received, the task data is determined according to the service matching response and the task data acquisition signal, and the task result and docking-related signals corresponding to the task data are received; the task data acquisition signal includes computing power budget information; the computing power budget information is dynamically managed according to computing power service resources; the docking-related signal is a continue to acquire task data signal or a docking backup service agent signal; the continue to acquire task data signal is a signal prompting to continue acquiring task data through the current service agent when the computing power service resources of the current service agent are valid; the docking backup service agent signal is a signal prompting to determine the task result corresponding to the task data through the backup service agent when the computing power service resources of the current service agent are invalid.

6. The computing power scheduling management method according to claim 5, characterized in that: The service matching response is generated by a plurality of service agents; the sending window information of the task data is dynamically adjusted according to the computing power budget information.

7. A baseband processing unit, characterized in that: It includes a service agent and a task agent; the service agent is used to execute the steps of the computing power scheduling management method described in any one of claims 1 to 4, and the task agent is used to execute the steps of the computing power scheduling management method described in any one of claims 5 to 6.

8. A communication system, characterized in that: It comprises several communication clusters, each of which comprises a plurality of baseband processing units as claimed in claim 7.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the computing power scheduling and management method as described in any one of claims 1 to 6 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computing power scheduling and management method as described in any one of claims 1 to 6 is implemented.

Images