WO2024124909A1 - 一种通信方法、电子设备及存储介质 - Google Patents

一种通信方法、电子设备及存储介质 Download PDF

Info

Publication number
WO2024124909A1
WO2024124909A1 PCT/CN2023/109115 CN2023109115W WO2024124909A1 WO 2024124909 A1 WO2024124909 A1 WO 2024124909A1 CN 2023109115 W CN2023109115 W CN 2023109115W WO 2024124909 A1 WO2024124909 A1 WO 2024124909A1
Authority
WO
WIPO (PCT)
Prior art keywords
computing power
target task
computing
target
power unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2023/109115
Other languages
English (en)
French (fr)
Inventor
孙洪峰
贺保国
靳虓
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZTE Corp
Original Assignee
ZTE Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ZTE Corp filed Critical ZTE Corp
Priority to EP23902119.9A priority Critical patent/EP4590020A4/en
Publication of WO2024124909A1 publication Critical patent/WO2024124909A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • H04W28/09—Management thereof
    • H04W28/0925—Management thereof using policies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/10—Scheduling measurement reports ; Arrangements for measurement reports
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
    • G06F1/26—Power supply means, e.g. regulation thereof
    • G06F1/32—Means for saving power
    • G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
    • G06F1/3234—Power saving characterised by the action undertaken
    • G06F1/329—Power saving characterised by the action undertaken by task scheduling
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/5044—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering hardware capabilities
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/09—Supervised learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/098—Distributed learning, e.g. federated learning
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/06—Testing, supervising or monitoring using simulated traffic
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/08—Load balancing or load distribution
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/48—Indexing scheme relating to G06F9/48
    • G06F2209/485—Resource constraint
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/50—Indexing scheme relating to G06F9/50
    • G06F2209/503—Resource availability
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00—Indexing scheme relating to G06F9/00
    • G06F2209/50—Indexing scheme relating to G06F9/50
    • G06F2209/509—Offload

Definitions

  • the present application belongs to the field of communication technology, and specifically relates to a communication method, electronic equipment and storage medium.
  • AI artificial intelligence
  • AI-based training and inference engines are usually centrally deployed on a single-point computing board of a base station.
  • RAN has rich application scenarios and the computing power of a single station is fixed. Supporting numerous application scenarios will bring about the problem of insufficient single-point computing power.
  • data acquisition and AI applications may be located on different boards of the base station, resulting in the need to transmit data from other boards to a fixed location for AI training and inference, which will bring about problems such as data transmission overhead and feedback delay.
  • the purpose of the embodiments of the present application is to provide a communication method, device, electronic device and storage medium, which can solve the problem of insufficient computing power and training of RAN under the condition of large amount of business data and rich application scenarios.
  • an embodiment of the present application provides a communication method, which is executed by a first computing power unit, and the method includes: sending target startup information to a second computing power unit, wherein the target startup information carries characteristics of a target task; receiving computing power measurement information reported by the second computing power unit, wherein the computing power measurement information is used to represent the available computing power allocated to the target task by the second computing power unit; and sending the target task to the second computing power unit when the available computing power matches the characteristics of the target task.
  • an embodiment of the present application provides a communication method, which is executed by a second computing power unit, and the method includes: receiving target startup information sent by a first computing power unit, wherein the target startup information carries the characteristics of a target task; allocating available computing power to the target task according to the characteristics of the target task; and sending computing power measurement information to the first computing power unit, wherein the computing power measurement information is used to represent the available computing power.
  • an embodiment of the present application provides a communication method, which is executed by a third computing power unit, and the method includes: receiving a processing result of a target task sent by a first computing power unit; evaluating the processing result of the target task to obtain a target evaluation result; and feeding back the target evaluation result to the first computing power unit.
  • an embodiment of the present application provides a communication device, comprising: a first sending module, used to send target startup information to a second computing power unit, wherein the target startup information carries the characteristics of a target task; a first receiving module, used to receive computing power measurement information reported by the second computing power unit, wherein the computing power measurement information is used to indicate the available computing power allocated by the second computing power unit to the target task; and a second sending module, used to send the target task to the second computing power unit when the available computing power matches the characteristics of the target task.
  • an embodiment of the present application provides a communication device, the device comprising: a second receiving module, configured to receive target startup information sent by a first computing unit, wherein the target startup information carries the characteristics of a target task; an allocation module, configured to allocate the target task to the first computing unit according to the characteristics of the target task; The target task allocates available computing power; a third sending module is used to send computing power measurement information to the first computing power unit, wherein the computing power measurement information is used to represent the available computing power.
  • an embodiment of the present application provides a communication device, which includes: a third receiving module, used to receive the processing result of the target task sent by the first computing power unit; an evaluation module, used to evaluate the processing result of the target task to obtain a target evaluation result; and a feedback module, used to feed back the target evaluation result to the first computing power unit.
  • an embodiment of the present application provides an electronic device, which includes a processor and a memory, wherein the memory stores programs or instructions that can be run on the processor, and when the program or instructions are executed by the processor, the steps of the communication method described in the first aspect, the second aspect, or the third aspect are implemented.
  • an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored.
  • the program or instruction is executed by a processor, the steps of the communication method described in the first aspect, the second aspect, or the third aspect are implemented.
  • an embodiment of the present application provides a chip, which includes a processor and a communication interface, the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the steps of the communication method described in the first aspect, the second aspect, or the third aspect.
  • an embodiment of the present application provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the steps of the communication method as described in the first aspect, the second aspect, or the third aspect.
  • FIG1 is a flow chart of a communication method provided in an embodiment of the present application.
  • FIG2 is a schematic diagram of an AI distributed framework structure provided in an embodiment of the present application.
  • FIG3 is a schematic diagram of the structure of an AI training inference engine provided in an embodiment of the present application.
  • FIG4 is a schematic diagram of a distributed framework structure within a base station provided in an embodiment of the present application.
  • FIG5 is a schematic diagram of a distributed framework structure between base stations provided in an embodiment of the present application.
  • Figure 6 is a schematic diagram of a distributed framework structure between a base station and an edge computing device provided in an embodiment of the present application.
  • Figure 7 is a general architecture diagram of a 5G wireless access network RAN provided in an embodiment of the present application.
  • FIG8 is a flow chart of a communication method provided in an embodiment of the present application.
  • FIG. 9 is a flow chart of a communication method provided in an embodiment of the present application.
  • FIG. 10 is a flow chart of a communication method provided in an embodiment of the present application.
  • FIG. 11 is a flow chart of a communication method provided in an embodiment of the present application.
  • FIG. 12 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • FIG13 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • FIG. 14 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • FIG. 15 is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application.
  • first, second, etc. in the specification and claims of this application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first”, “second”, etc. are generally of the same type.
  • the number of objects is not limited, for example, the first object can be one or more.
  • “and/or” means at least one of the connected objects, and the character “/” generally means that the related objects are in an "or” relationship.
  • FIG1 shows a flow chart of a communication method provided by an embodiment of the present application.
  • the method can be executed by an electronic device or a first computing unit on the electronic device.
  • the electronic device may include: a server or a terminal device.
  • the method can be executed by software or hardware installed on the electronic device. As shown in FIG1, the method includes the following steps:
  • S101 Send target startup information to the second computing power unit.
  • the target startup information carries the characteristics of the target task.
  • the AI engine of the wireless access network RAN is usually centrally deployed on a single-point computing board of the base station.
  • the computing power of the base station under this deployment mode is fixed. Due to different scenarios or scenarios at different protocol layers, data acquisition and AI applications may be located on different boards of the base station. Centralized deployment requires that data from other boards be transmitted to a fixed location before AI training and reasoning can be performed, which brings problems such as data transmission overhead and feedback delay.
  • the wireless access network has rich application scenarios, and the computing power of a single station is fixed. If you want to support many application scenarios, the problem of insufficient single-point computing power will arise.
  • the AI engine is a framework that supports users to develop machine learning and deep learning model training operations.
  • the overall architecture of the artificial intelligence distributed AI framework is shown in Figure 2 as a master-slave mode: the main AI training and reasoning engine collaborates with multiple auxiliary AI training and reasoning engines to complete training and reasoning related tasks, and the main AI training and reasoning engine and multiple auxiliary AI training and reasoning engines communicate through a virtual eXtensible Local Area Network (VXLAN).
  • VXLAN virtual eXtensible Local Area Network
  • the main and auxiliary AI training and reasoning engines are composed of computing power management, task management, model management, and AI training and reasoning framework.
  • the computing power management is responsible for computing power measurement, evaluating AI engine computing power, and maintaining computing power status information; the model management maintains AI model addition, update, deletion management, and model loading; the task management is responsible for managing and allocating AI training tasks and reasoning tasks, and the AI training and reasoning framework is responsible for executing model training and reasoning.
  • the main AI training engine i.e., the first computing unit
  • sends a target startup message to the auxiliary AI training engine i.e., the second computing unit.
  • the target startup message carries the characteristics of the target task, which is used to notify the second computing unit to report the available computing power that can be allocated to the target task.
  • S102 Receive computing power measurement information reported by the second computing power unit.
  • the computing power measurement information is used to represent the available computing power allocated by the second computing power unit to the target task.
  • This step receives the computing power management module of the auxiliary AI training engine and reports the computing power measurement information to the main AI training engine.
  • the computing power measurement information is used to indicate the available computing power allocated by the auxiliary AI training engine to the target task.
  • a0_type indicates the type of computing power hardware, such as central processing unit (CPU), field-programmable gate array (FPGA), graphics processing unit (GPU), etc.
  • a1_flops indicates the computing power of the hardware
  • a2_load indicates the load of the computing power, including the maximum computing power used, the minimum computing power used, the average computing power used, etc.
  • a3_time indicates the time information of computing power usage.
  • the main AI training and reasoning engine receives the computing power measurement information reported by each auxiliary AI training and reasoning engine, and the task management module matches the training and reasoning task characteristics with the computing power.
  • the AI training and reasoning task characteristics cover the computing power resources required for training and reasoning, the source of training and reasoning data, the amount of data, and the real-time requirements of the task; the matching rules meet the engine that executes the training and reasoning tasks to use local data for training and reasoning to ensure the real-time nature of training and reasoning.
  • training and reasoning tasks may not be assigned to auxiliary training engines with relatively high loads.
  • the main AI training and reasoning engine sends the training and reasoning tasks to the auxiliary AI training and reasoning engines for execution based on the task matching results of the task management module.
  • a communication method provided in an embodiment of the present application is provided by sending a target startup to a second computing unit. information, wherein the target startup information carries the characteristics of the target task; receiving the computing power measurement information reported by the second computing power unit, wherein the computing power measurement information is used to represent the available computing power allocated to the target task by the second computing power unit; when the available computing power matches the characteristics of the target task, the target task is sent to the second computing power unit, which can solve the problems of insufficient computing power, large training and inference latency, and poor real-time performance of RAN when the amount of business data is large and the application scenarios are rich.
  • the first computing power unit includes a first main control board of a first base station
  • the second computing power unit includes at least one baseband board of the first base station, at least one baseband board of the second base station, and at least one edge computing device connected to the first base station.
  • the second computing power unit when the second computing power unit includes at least one baseband board of the second base station, the first base station and the second base station establish an inter-base station connection through the Xn port; when the second computing power unit includes at least one baseband board of the first base station, the first computing power unit and the second computing power unit are connected through a virtual extensible local area network VXLAN.
  • the overall architecture of the fifth generation mobile communication technology (5G) radio access network RAN is as follows: the next generation base station (Next Generation Node B, gNB) provides 5G new radio (New Radio, NR) user plane and control plane protocols; the next generation evolved base station (Next Generation Evolved Node B, ng-eNB) provides the evolved universal mobile telecommunications system (Universal Mobile Telecommunications System, UMTS) terrestrial radio access (Evolved UMTS Terrestrial Radio Access Network, E-UTRA) user plane and control plane protocols.
  • the connection between gNB and ng-eNB is through the Xn port.
  • the gNB and ng-eNB are connected to the 5G core network (5G Core, 5GC) through the NG port.
  • the 5GC includes the authentication management function (Authentication Management Function, AMF) and the user plane function (User Plane Function, UPF).
  • AMF Authentication Management Function
  • UPF User Plane Function
  • the AMF communicates with the gNB and UPF through the NG-U port respectively.
  • ng-eNB connection In the embodiment of the present application, the artificial intelligence AI distributed framework can be deployed in a distributed manner within the gNB, or in a distributed manner on the gNB and ng-eNB, and can ultimately be applied to the intelligentization of 5G wireless access networks.
  • after sending the target task to the second computing unit it also includes: receiving a processing result of the target task sent by the second computing unit.
  • after receiving the processing result of the target task sent by the second computing power unit it also includes: sending the processing result of the target task to a third computing power unit; receiving a target evaluation result fed back by the third computing power unit, wherein the target evaluation result is obtained by evaluating the processing result of the target task; and updating the processing result of the target task according to the target evaluation result.
  • the target task includes: the task of training the model; the processing result of the target task includes: the model obtained by training; the target evaluation result includes: the evaluation result obtained by evaluating the model.
  • the AI training of RAN can be carried out in a cloud manner, that is, the AI training tasks of the base station are obtained through the cloud training system, and then the training results are returned to the base station, so that the AI training of the wireless access network is no longer limited by the computing power of the base station.
  • data transmission is required between the base station and the cloud. Massive data has high requirements on the transmission bandwidth, and it will also bring large data transmission delay, resulting in poor real-time performance of online AI training reasoning. In response to task scenarios with high real-time requirements, the expected effect cannot be achieved.
  • the embodiments of the present application can collect and measure computing resources within the deployment domain by deploying an artificial intelligence distributed framework inside or between wireless base stations, and match available computing resources with AI training and reasoning tasks, completing AI reasoning and training tasks in a distributed manner, and can also share training models between base stations in a cross-site manner.
  • it can solve the problem that some models cannot be trained online due to high business load and limited idle computing power at some sites.
  • FIG8 is a flow chart of a communication method provided by an embodiment of the present application.
  • the method can be executed by an electronic device or a second computing unit on the electronic device.
  • the electronic device may include: a server or a terminal device.
  • the method can be executed by software or hardware installed on the electronic device. As shown in FIG8, the method includes the following steps:
  • S201 Receive target startup information sent by the first computing unit.
  • the target startup information carries the characteristics of the target task.
  • This step deploys the computing power management module of the auxiliary AI training engine (i.e., the second computing power unit) of different baseband boards to report the computing power measurement information of this baseband board to the main AI reasoning training engine (i.e., the first computing power unit).
  • the main AI reasoning training engine i.e., the first computing power unit
  • S202 Allocate available computing power to the target task according to the characteristics of the target task.
  • the computing power management module of the auxiliary AI training engine measures the local available computing power through calculation according to the characteristics of the target task.
  • S203 Send computing power measurement information to the first computing power unit, where the computing power measurement information is used to represent the available computing power.
  • the auxiliary AI training engine reports the available computing power to the main AI training inference engine.
  • a communication method provided in an embodiment of the present application receives target startup information sent by a first computing power unit, wherein the target startup information carries the characteristics of a target task; allocates available computing power to the target task according to the characteristics of the target task; and sends computing power measurement information to the first computing power unit, wherein the computing power measurement information is used to represent the available computing power.
  • This method can solve the problems of insufficient computing power, large training and inference latency, and poor real-time performance of RAN when the amount of business data is large and the application scenarios are rich.
  • the computing power measurement information after the computing power measurement information is sent to the first computing power unit, it also includes: receiving the target task, where the target task is sent by the first computing power unit when the available computing power matches the characteristics of the target task.
  • the method further includes: executing the target task to obtain a processing result of the target task; sending the target task to the first computing unit; The processing results of the service.
  • FIG9 shows a flow chart of a communication method provided by an embodiment of the present application. The method comprises the following steps:
  • the main AI training inference engine sends target startup information to the auxiliary AI training engines respectively.
  • the computing power management module of the auxiliary AI training engine deployed on different baseband boards reports the computing power measurement of the baseband board to the main AI inference training engine.
  • the main AI training and inference engine receives the computing power measurement information reported by each auxiliary AI training and inference engine, and the task management module matches the training and inference task characteristics with the computing power.
  • the main AI training and reasoning engine sends the training and reasoning tasks to the auxiliary AI training and reasoning engine for execution based on the task matching results of the task management module.
  • the local AI training and inference engine receives the task, starts matching the data required for the training and inference task, and starts collecting and preprocessing local data.
  • the local AI training and inference engine receives the task and uses the local pre-processed data to perform local training and inference tasks. It can also store the trained model and inference results locally through the local model management module.
  • the model and reasoning results are also synchronously sent to the model management module of the main AI training and reasoning engine for management.
  • the model management module updates the managed model according to the model management strategy.
  • FIG10 is a flow chart of a communication method provided by an embodiment of the present application.
  • the method can be executed by an electronic device or a third computing unit on the electronic device.
  • the electronic device may include: a server or a terminal device.
  • the method can be executed by software or hardware installed on the electronic device. As shown in FIG10, the method includes the following steps:
  • S301 Receive the processing result of the target task sent by the first computing unit.
  • the base station (third computing unit) that deploys the auxiliary AI training and inference engine has upper-layer applications that need to use the corresponding AI model and inference results; however, due to the high load and low idle computing power of this station, this station has not previously performed the corresponding AI model training and inference, and the upper-layer applications of this station cannot match the corresponding AI model and inference results.
  • the base station that deploys the auxiliary AI training and inference engine requests the corresponding AI model and inference results from the base station of the main AI training and inference engine.
  • the base station of the main AI training inference engine sends the corresponding AI model ⁇ inference result to the base station of the auxiliary AI training inference engine.
  • the upper-layer application of the base station of the auxiliary AI training inference engine calls this AI model ⁇ inference result to complete the upper-layer business use.
  • the auxiliary AI training inference engine evaluates the effect of the upper-layer application calling this AI model ⁇ inference result.
  • the auxiliary AI training and reasoning engine will feed back the corresponding evaluation results to the main AI training and reasoning engine, so that the main AI training and reasoning engine can decide whether to carry out the next round of model training and model reasoning tasks according to the effect evaluation strategy.
  • a communication method provided in an embodiment of the application can solve the problem of RAN in service data by receiving the processing result of a target task sent by a first computing unit; evaluating the processing result of the target task to obtain a target evaluation result; and feeding back the target evaluation result to the first computing unit.
  • There are problems such as insufficient computing power when the volume is large and the application scenarios are rich, as well as large training and inference latency and poor real-time performance.
  • the base stations are deployed in a mesh, different base stations have the same application scenarios. Different training tasks can be deployed on different base stations respectively. Each base station can share the training results, which can support AI training in all scenarios, achieve the effect of enhancing single-point computing power, and improve the processing capabilities of the wireless access network.
  • FIG11 shows a flow chart of a communication method provided by an embodiment of the present application. The method comprises the following steps:
  • the base station (third computing unit) that deploys the auxiliary AI training and inference engine has upper-layer applications that need to use the corresponding AI model ⁇ inference results; however, due to the high load and low idle computing power of this station, the station has not previously performed corresponding AI model training ⁇ inference, and the upper-layer applications of this station cannot match the corresponding AI model ⁇ inference results.
  • the base station that deploys the auxiliary AI training and reasoning engine requests the corresponding AI model ⁇ inference result from the base station of the main AI training and reasoning engine.
  • the base station of the main AI training inference engine sends the corresponding AI model ⁇ inference result to the base station of the auxiliary AI training inference engine.
  • the upper-layer application of the base station of the auxiliary AI training inference engine calls this AI model ⁇ inference result to complete the upper-layer business use.
  • the auxiliary AI training inference engine evaluates the effect of the upper-layer application calling this AI model ⁇ inference result.
  • the auxiliary AI training inference engine feeds back the corresponding evaluation results to the main AI training inference engine, so that the main AI training inference engine decides whether to carry out the next round of model training ⁇ model inference tasks according to the effect evaluation strategy.
  • the above steps are mainly based on the process of sharing AI models and inference results between base stations in a distributed framework.
  • the specific deployment principle is: base stations with relatively low load and more idle computing power deploy the main AI training and inference engine; base stations with relatively high load and less idle computing power deploy the auxiliary AI training and inference engine.
  • the trained models and inference results will be managed by the main AI training and inference engine.
  • the base stations that deploy the auxiliary AI training and inference engines can directly use the AI models and inference results managed by the main AI training and inference engine to complete the upper-level business applications.
  • the distributed deployment mode and communication method between the base station and the edge computing device can refer to the description of the above embodiment, which can achieve the same technical effect. To avoid repetition, no further details will be given.
  • the communication method provided in the embodiment of the present application can be executed by a communication device or a control module in the communication device for executing the communication method.
  • the communication device provided in the embodiment of the present application is described by taking the method for executing the communication by the communication device as an example.
  • FIG12 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • the communication device 500 includes: a first sending module 510 , a first receiving module 520 and a second sending module 530 .
  • the first sending module 510 is used to send target startup information to the second computing power unit, wherein the target startup information carries the characteristics of the target task; the first receiving module 520 is used to receive computing power measurement information reported by the second computing power unit, wherein the computing power measurement information is used to represent the available computing power allocated by the second computing power unit to the target task; the second sending module 530 is used to send the target task to the second computing power unit when the available computing power matches the characteristics of the target task.
  • a communication device includes a first sending module for sending target startup information to a second computing unit, wherein the target startup information carries the characteristics of a target task; a first receiving module for receiving computing power measurement information reported by the second computing unit, wherein the computing power measurement information is used to indicate the available computing power allocated by the second computing unit to the target task; and a second sending module for sending the target task to the second computing unit when the available computing power matches the characteristics of the target task, which can solve the problem of RAN having a large amount of business data and a high speed communication interface.
  • There are problems such as insufficient computing power in a variety of application scenarios, large training and inference latency, and poor real-time performance.
  • the first receiving module 520 is further used to: receive the processing result of the target task sent by the second computing unit.
  • the device 500 also includes: a third sending module, used to send the processing result of the target task to the third computing power unit; the first receiving module 520 is also used to: receive the target evaluation result fed back by the third computing power unit, wherein the target evaluation result is obtained by evaluating the processing result of the target task; the device 500 also includes: an updating module, used to update the processing result of the target task according to the target evaluation result.
  • a third sending module used to send the processing result of the target task to the third computing power unit
  • the first receiving module 520 is also used to: receive the target evaluation result fed back by the third computing power unit, wherein the target evaluation result is obtained by evaluating the processing result of the target task
  • the device 500 also includes: an updating module, used to update the processing result of the target task according to the target evaluation result.
  • FIG13 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • the communication device 600 includes: a second receiving module 610 , an allocating module 620 and a fourth sending module 630 .
  • the second receiving module 610 is used to receive target startup information sent by the first computing power unit, wherein the target startup information carries the characteristics of the target task; the allocation module 620 is used to allocate available computing power to the target task according to the characteristics of the target task; the fourth sending module 630 is used to send computing power measurement information to the first computing power unit, wherein the computing power measurement information is used to represent the available computing power.
  • a communication device configured to receive target startup information sent by a first computing power unit through a second receiving module, wherein the target startup information carries characteristics of a target task; an allocation module is configured to allocate available computing power to the target task according to the characteristics of the target task; and a fourth sending module is configured to send computing power measurement information to the first computing power unit, wherein the computing power measurement information is used to represent the available computing power, which can solve the problems of insufficient computing power, large training and inference latency, and poor real-time performance of RAN when the amount of business data is large and the application scenarios are rich.
  • the second receiving module 610 is further used to: receive the target task, where the target task is sent by the first computing power unit when the available computing power matches the characteristics of the target task.
  • the apparatus 600 further includes: an execution module, configured to execute the target task and obtain a processing result of the target task; and the fourth sending module 630 is further configured to: Send the processing result of the target task to the first computing power unit.
  • FIG14 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application.
  • the communication device 700 includes: a third receiving module 710 , an evaluation module 720 and a fifth sending module 730 .
  • the third receiving module 710 is used to receive the processing result of the target task sent by the first computing unit; the evaluation module 720 is used to evaluate the processing result of the target task to obtain a target evaluation result; the fifth sending module 730 is used to feed back the target evaluation result to the first computing unit.
  • the first computing power unit includes a first main control board of a first base station
  • the second computing power unit includes at least one baseband board of the first base station, at least one baseband board of the second base station, and at least one edge computing device connected to the first base station.
  • the second computing power unit when the second computing power unit includes at least one baseband board of the second base station, the first base station and the second base station establish an inter-base station connection through the Xn port; when the second computing power unit includes at least one baseband board of the first base station, the first computing power unit and the second computing power unit are connected through a virtual extensible local area network VXLAN.
  • the third computing power unit includes at least one baseband board of the first base station, at least one baseband board of the second base station, and at least one edge computing device connected to the first base station; the third computing power unit and the second computing power unit are not the same computing power unit.
  • the target task includes: a task of training a model; the processing result of the target task includes: a model obtained by training; and the target evaluation result includes: an evaluation result obtained by evaluating the model.
  • the communication device provided in the embodiment of the present application can implement the various processes implemented by the communication method embodiment described in at least one embodiment of Figures 1 to 11, and achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the communication device in the embodiments of the present application may be a device, or a component, integrated circuit, or chip in a terminal device.
  • the device may be a mobile electronic device or a non-mobile electronic device.
  • the mobile electronic device may be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer, or a portable computer.
  • the mobile electronic device may be a mobile computer, UMPC), a netbook or a personal digital assistant (PDA), etc.
  • the non-mobile electronic device may be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), an ATM or an automatic machine, etc., which is not specifically limited in the embodiments of the present application.
  • the communication device in the embodiment of the present application may be a device having an operating system.
  • the operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.
  • an embodiment of the present application further provides an electronic device 800, including a processor 801, a memory 802, a program or instruction stored in the memory 802 and executable on the processor 801, and when the program or instruction is executed by the processor 801, the communication method described in at least one of the embodiments of FIG1 to FIG11 is implemented.
  • the electronic device in the embodiment of the present application includes: a server, a terminal device, or other devices other than a terminal device.
  • the above electronic device structure does not constitute a limitation on the electronic device.
  • the electronic device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.
  • the input unit may include a graphics processing unit (GPU) and a microphone
  • the display unit may be configured with a display panel in the form of a liquid crystal display, an organic light-emitting diode, etc.
  • the user input unit includes a touch panel and at least one of other input devices.
  • the touch panel is also called a touch screen.
  • Other input devices may include, but are not limited to, a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.
  • the memory can be used to store software programs and various data.
  • the memory may mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area may store an operating system, an application program or instructions required for at least one function (such as a sound playback function, an image playback function, etc.), etc.
  • the memory may include a volatile memory or a non-volatile memory, or the memory may include both volatile and non-volatile memories.
  • the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (Erasable PROM, EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), or Flash memory.
  • Volatile memory can be Random Access Memory (RAM), Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).
  • RAM Random Access Memory
  • SRAM Static RAM
  • DRAM Dynamic RAM
  • SDRAM Synchronous DRAM
  • DDRSDRAM Double Data Rate SDRAM
  • ESDRAM Enhanced SDRAM
  • SLDRAM Synchronous Link DRAM
  • DRRAM Direct Rambus RAM
  • the processor may include one or more processing units; optionally, the processor integrates an application processor and a modem processor, wherein the application processor mainly processes operations related to the operating system, user interface, and application programs, and the modem processor mainly processes communication signals, such as a baseband processor. It is understandable that the modem processor may not be integrated into the processor.
  • An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored.
  • a program or instruction is stored.
  • the communication method described in at least one of the embodiments in Figures 1 and 2 is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
  • the processor is a processor in the electronic device described in the above embodiment.
  • the readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.
  • An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned communication method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
  • the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
  • the technical solution of the present application can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM/RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.
  • a storage medium such as ROM/RAM, a disk, or an optical disk
  • a terminal which can be a mobile phone, a computer, a server, or a network device, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Neurology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请公开了一种通信方法、电子设备及存储介质，属于通信技术领域。所述方法包括：向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。

Description

一种通信方法、电子设备及存储介质
交叉引用
本申请要求在2022年12月13日提交中国专利局、申请号为202211595595.0、发明名称为“一种通信方法、电子设备及存储介质”的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本申请属于通信技术领域，具体涉及一种通信方法、电子设备及存储介质。
背景技术
随着无线通信技术的发展，以及具有高速率、高带宽、低时延等特点的第五代移动通信技术(5th-Generation Mobile Communication Technology，5G)时代的到来，作为用户终端接入核心网媒介的无线接入网(Radio Access Network，RAN)，其计算及管理能力日益重要。人工智能(Artificial Intelligence，AI)技术的核心是使用数据、算法和代码构建的AI模型来实现控制智能化。
在相关RAN中，基于AI的训练推理引擎，通常都是集中式部署在一个基站的单点算力板上，但RAN应用场景丰富，单站下算力固定，想要支持众多的应用场景会带来单点算力不足的问题，且数据获取以及AI应用可能位于基站不同的单板上，导致其他单板数据需要传输到固定位置才可以进行AI训练以及推理，带来数据传输的开销以及反馈时延等问题。
发明内容
本申请实施例的目的是提供一种通信方法、装置、电子设备及存储介质，能够解决RAN在业务数据量大、应用场景丰富情况下的算力不足以及训练 推理时延大、实时性差的问题。
为了解决上述技术问题，本申请是这样实现的：
第一方面，本申请实施例提供了一种通信方法，由第一算力单元执行，所述方法包括：向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。
第二方面，本申请实施例提供了一种通信方法，由第二算力单元执行，所述方法包括：接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；根据所述目标任务的特征为所述目标任务分配可用算力；向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力。
第三方面，本申请实施例提供了一种通信方法，由第三算力单元执行，所述方法包括：接收第一算力单元发送的目标任务的处理结果；对所述目标任务的处理结果进行评估，得到目标评估结果；将所述目标评估结果反馈给所述第一算力单元。
第四方面，本申请实施例提供了一种通信装置，该装置包括：第一发送模块，用于向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；第一接收模块，用于接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；第二发送模块，用于在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。
第五方面，本申请实施例提供了一种通信装置，该装置包括：第二接收模块，用于接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；分配模块，用于根据所述目标任务的特征为所述 目标任务分配可用算力；第三发送模块，用于向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力。
第六方面，本申请实施例提供了一种通信装置，该装置包括：第三接收模块，用于接收第一算力单元发送的目标任务的处理结果；评估模块，用于对所述目标任务的处理结果进行评估，得到目标评估结果；反馈模块，用于将所述目标评估结果反馈给所述第一算力单元。
第七方面，本申请实施例提供了一种电子设备，该电子设备包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如第一方面或第二方面或第三方面所述的通信方法的步骤。
第八方面，本申请实施例提供了一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如第一方面或第二方面或第三方面所述的通信方法的步骤。
第九方面，本申请实施例提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如第一方面或第二方面或第三方面所述的通信方法的步骤。
第十方面，本申请实施例提供一种计算机程序产品，该程序产品被存储在存储介质中，该程序产品被至少一个处理器执行以实现如第一方面或第二方面或第三方面所述的通信方法的步骤。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请中记载的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例提供的一种通信方法的流程示意图。
图2是本申请实施例提供的一种AI分布式框架结构示意图。
图3是本申请实施例提供的一种AI训练推理引擎结构示意图。
图4是本申请实施例提供的一种基站内分布式框架结构示意图。
图5是本申请实施例提供的一种基站间分布式框架结构示意图。
图6是本申请实施例提供的一种基站和边缘计算设备间分布式框架结构示意图。
图7是本申请实施例提供的一种5G无线接入网RAN总体架构图。
图8是本申请实施例提供的一种通信方法的流程示意图。
图9是本申请实施例提供的一种通信方法的流程示意图。
图10是本申请实施例提供的一种通信方法的流程示意图。
图11是本申请实施例提供的一种通信方法的流程示意图。
图12是本申请实施例提供的一种通信装置的结构示意图。
图13是本申请实施例提供的一种通信装置的结构示意图。
图14是本申请实施例提供的一种通信装置的结构示意图。
图15是本申请实施例提供的一种电子设备的硬件结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员获得的所有其他实施例，都属于本申请保护的范围。
本申请的说明书和权利要求书中的术语“第一”、“第二”等是用于区别类似的对象，而不用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施，且“第一”、“第二”等所区分的对象通常为一类， 并不限定对象的个数，例如第一对象可以是一个，也可以是多个。此外，说明书以及权利要求中“和/或”表示所连接对象的至少其中之一，字符“/”，一般表示前后关联对象是一种“或”的关系。
下面结合附图，通过具体的实施例及其应用场景对本申请实施例提供的一种通信方法、装置、电子设备及存储介质进行详细地说明。
图1示出本申请的一个实施例提供的一种通信方法的流程示意图，该方法可以由电子设备或电子设备上的第一算力单元执行，该电子设备可以包括：服务器或终端设备。换言之，该方法可以由安装在电子设备的软件或硬件来执行，如图1所示，该方法包括如下步骤：
S101：向第二算力单元发送目标启动信息。
所述目标启动信息中携带目标任务的特征。
无线接入网RAN的AI引擎通常都是集中式部署在基站的单点算力板上，该部署方式下的基站算力是固定的；由于场景的不同或者场景处于不同的协议层，数据获取以及AI应用可能位于基站不同的单板上，采用集中式部署，导致其他单板数据需要传输到固定位置才可以进行AI训练以及推理，带来数据传输的开销以及反馈时延等问题；另外，无线接入网应用场景丰富，单站下算力固定，想要支持众多的应用场景会带来单点算力不足的问题。
AI引擎是可支持用户进行机器学习、深度学习模型训练作业开发的框架。人工智能分布式AI框架整体架构如图2所示为主从方式：主AI训练推理引擎与多个辅AI训练推理引擎协作完成训练推理相关任务，主AI训练推理引擎与多个辅AI训练推理引擎协之间通过虚拟可扩展局域网(Virtual eXtensible Local Area Network，VXLAN)进行通信。如图3所示，主\辅AI训练推理引擎由算力管理、任务管理、模型管理、AI训练推理框架组成，算力管理负责算力度量，评估AI引擎算力，维护算力状态信息；模型管理维护AI模型新增、更新、删除管理、模型加载；任务管理负责管理分配AI训练任务、推理任务，AI训练推理框架负责执行模型的训练、推理。
本步骤主AI训练引擎(也即第一算力单元)，向辅AI训练引擎(也即第二算力单元)发送目标启动信息，所述目标启动信息中携带目标任务的特征，用以通知所述第二算力单元上报可以分配给所述目标任务的可用算力。
S102：接收所述第二算力单元上报的算力度量信息。
所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力。
本步骤接收辅AI训练引擎的算力管理模块，向主AI训练引擎上报的算力度量信息，所述算力度量信息用于表示所述辅AI训练引擎分配给所述目标任务的可用算力。
具体地，例如，算力度量可以表征为如下函数：
f＝f(a0_type,a1_flops,a2_load,a3_time)
a0_type表示算力硬件类型中央处理器(Central Processing Unit，CPU)、现场可编程门阵列(Field－Programmable Gate Array，FPGA)、图形处理器(Graphics Processing Unit，GPU)等；a1_flops表示硬件的算力大小；a2_load表示算力的负荷情况，包括最大使用算力，最小使用算力，平均使用算力等；a3_time表示算力使用的时间信息。
S103：在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。
主AI训练推理引擎收到各个辅AI训练推理引擎上报的算力度量信息，由任务管理模块进行训练\推理任务特征与算力匹配，AI训练推理任务特征涵盖训练推理所需要的算力资源、训练\推理数据来源、数据量大小、任务实时性要求；匹配规则满足执行训练\推理任务的引擎使用本地数据进行训练、推理，确保训练\推理的实时性，同时对于负荷比较高的辅训练引擎有可能不分配训练\推理任务。最后，主AI训练推理引擎根据任务管理模块的任务匹配结果将训练\推理任务发送给辅AI训练推理引擎执行
本申请实施例提供的一种通信方法，通过向第二算力单元发送目标启动 信息，其中，所述目标启动信息中携带目标任务的特征；接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元，能够解决RAN在业务数据量大、应用场景丰富情况下的算力不足以及训练推理时延大、实时性差的问题。
在一种实现方式中，所述第一算力单元包括第一基站的第一主控板，所述第二算力单元包括所述第一基站的至少一个基带板、第二基站的至少一个基带板、与所述第一基站连接的至少一个边缘计算设备中的至少一者。
在一种实现方式中，在所述第二算力单元包括第二基站的至少一个基带板的情况下，所述第一基站与所述第二基站通过Xn口建立基站间连接；在所述第二算力单元包括所述第一基站的至少一个基带板的情况下，所述第一算力单元与所述第二算力单元通过虚拟可扩展局域网VXLAN连接。
如图4至6所示，本申请实施例中，人工智能AI分布式框架的部署方式有三种：1)基站内分布式部署(图4)；2)基站间分布式部署(图5)；3)基站和边缘计算设备间分布式部署(图6)。
如图7所示，第五代移动通信(5th Generation Mobile Communication Technology，5G)无线接入网RAN的总体架构图：新一代基站(Next Generation Node B，gNB)提供5G新空口(New Radio，NR)用户面和控制面协议；下一代演进基站(Next Generation Evolved Node B，ng-eNB)提供演进的通用移动通信系统(Universal Mobile Telecommunications System，UMTS)陆面无线接入(Evolved UMTS Terrestrial Radio Access Network，E-UTRA)用户面和控制面协议。gNB和ng-eNB之间连接通过Xn口。gNB和ng-eNB通过NG口与5G核心网(5G Core，5GC)连接，5GC包含认证管理功能(Authentication Management Function，AMF)和用户面功能(User Plane Function，UPF)，其中AMF通过NG-C口，UPF通过NG-U分别与gNB和 ng-eNB连接。本申请实施例中人工智能AI分布式框架既可以在gNB内分布式部署，也可以在gNB和ng-eNB上采用分布式方式部署，最终可应用于5G无线接入网智能化。
在一种实现方式中，在将所述目标任务发送给所述第二算力单元之后，还包括：接收所述第二算力单元发送的所述目标任务的处理结果。
在一种实现方式中，在接收所述第二算力单元发送的所述目标任务的处理结果之后，还包括：将所述目标任务的处理结果发送给第三算力单元；接收所述第三算力单元反馈的目标评估结果，其中，所述目标评估结果是对所述目标任务的处理结果进行评估得到的；根据所述目标评估结果，对所述目标任务的处理结果进行更新。
需要说明的是，本申请实施例中，所述目标任务包括：训练模型的任务；所述目标任务的处理结果包括：训练得到的模型；所述目标评估结果包括：对所述模型进行评估得到的评估结果。
本领域技术人员可能想到，解决RAN算力受限问题，可以将RAN的AI训练通过云化方式进行，即通过云训练系统获取基站的AI训练任务，再将训练结果返回给基站，使得无线接入网的AI训练不再受限于基站的算力，但基站和云端之间需要进行数据传输，海量数据对传输带宽要求很高，同时也会带来数据传输时延大，导致在线AI训练推理实时性较差，应对实时性要求较高的任务场景，无法达到预期的效果。
而本申请实施例通过在无线基站内部或者基站间部署人工智能分布式框架，能够搜集、度量部署域内的算力资源，并将可用的算力资源与AI训练推理任务匹配，分布式完成AI的推理训练任务，同时也可以在基站间通过跨站方式共享训练模型。一方面可以解决某些站点由于业务负荷高，空闲算力受限导致有些模型无法在线训练问题，另一方面也解决了大量数据上传到云进行训练对传输时延、传输带宽要求极高、在线推理实时性差问题，同时数据不出基站安全性也更高，提高无线接入网的处理能力。
图8示出本申请的一个实施例提供的一种通信方法的流程示意图，该方法可以由电子设备或电子设备上的第二算力单元执行，该电子设备可以包括：服务器或终端设备。换言之，该方法可以由安装在电子设备的软件或硬件来执行，如图8所示，该方法包括如下步骤：
S201：接收第一算力单元发送的目标启动信息。
其中，所述目标启动信息中携带目标任务的特征。
本步骤部署于不同基带板的辅AI训练引擎(即所述第二算力单元)的算力管理模块，向主AI推理训练引擎(即所述第一算力单元)上报本基带板的算力度量信息。
S202：根据所述目标任务的特征为所述目标任务分配可用算力。
辅AI训练引擎的算力管理模块，根据所述目标任务的特征通过计算把本地的可用算力度量出来。
S203：向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力。
辅AI训练引擎将所述可用算力上报给主AI训练推理引擎。
本申请实施例提供的一种通信方法，通过接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；根据所述目标任务的特征为所述目标任务分配可用算力；向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力，能够解决RAN在业务数据量大、应用场景丰富情况下的算力不足以及训练推理时延大、实时性差的问题。
在一种实现方式中，在所述向所述第一算力单元发送的算力度量信息之后，还包括：接收所述目标任务，所述目标任务是所述第一算力单元在所述可用算力与所述目标任务的特征匹配的情况下发送的。
在一种实现方式中，在接收所述目标任务之后，还包括：执行所述目标任务，获得所述目标任务的处理结果；向所述第一算力单元发送所述目标任 务的处理结果。
图9示出了本申请的一个实施例提供的一种通信方法的流程示意图。该方法包括如下步骤：
S001：发送目标启动信息。
主AI训练推理引擎向辅AI训练引擎分别发送目标启动信息。
S002：算力度量信息上报。
部署于不同基带板的辅AI训练引擎的算力管理模块，向主AI推理训练引擎上报本基带板的算力度量。
S003：任务与算力匹配。
主AI训练推理引擎收到各个辅AI训练推理引擎上报的算力度量信息，由任务管理模块进行训练\推理任务特征与算力匹配。
S004：任务下发。
主AI训练推理引擎根据任务管理模块的任务匹配结果将训练\推理任务发送给辅AI训练推理引擎执行。
S005：本地数据采集预处理。
本地AI训练推理引擎收到任务，启动匹配训练\推理任务所需要的数据，启动本地数据的采集与预处理。
S006：本地执行任务。
本地AI训练推理引擎收到任务，利用本地的预处理后数据，进行本地训练\推理任务，还可以将训练后的模型\推理结果通过本地模型管理模块进行本地化存储。
S007：任务执行结果上报。
本地AI训练推理引擎完成训练推理后，将模型\推理结果也同步发送给主AI训练推理引擎的模型管理模块进行管理，模型管理模块根据模型管理策略，更新所管理的模型。
以上步骤从主训练引擎与从训练引擎交互的角度，说明了本申请实施例 提供的一种通信方法，其具体步骤可以参见前述实施例中相同部分的描述，且能达到相同的技术效果，避免重复，不再赘述。
图10示出本申请的一个实施例提供的一种通信方法的流程示意图，该方法可以由电子设备或电子设备上的第三算力单元执行，该电子设备可以包括：服务器或终端设备。换言之，该方法可以由安装在电子设备的软件或硬件来执行，如图10所示，该方法包括如下步骤：
S301：接收第一算力单元发送的目标任务的处理结果。
在本地AI训练推理引擎完成训练推理后，将模型\推理结果同步发送给主AI训练推理引擎的模型管理模块进行管理。部署辅AI训练推理引擎的基站(第三算力单元)，有上层应用需要使用相应的AI模型\推理结果；但由于本站负荷较高、空闲算力较少，本站之前没有进行对应的AI模型训练\推理，本站上层应用无法匹配到相应的AI模型\推理结果。该部署辅AI训练推理引擎的基站向主AI训练推理引擎的基站请求对应的AI模型\推理结果。
主AI训练推理引擎的基站将对应的AI模型\推理结果发送给辅AI训练推理引擎的基站。
S302：对所述目标任务的处理结果进行评估，得到目标评估结果。
辅AI训练推理引擎的基站的上层应用调用此AI模型\推理结果完成上层业务使用。辅AI训练推理引擎将上层应用调用此AI模型\推理结果的效果进行评估。
S303：将所述目标评估结果反馈给所述第一算力单元。
辅AI训练推理引擎将对应的评估结果反馈给主AI训练推理引擎，以使主AI训练推理引擎根据效果评估策略决定是否进行下一轮模型训练\模型推理任务。
申请实施例提供的一种通信方法，通过接收第一算力单元发送的目标任务的处理结果；对所述目标任务的处理结果进行评估，得到目标评估结果；将所述目标评估结果反馈给所述第一算力单元，能够解决RAN在业务数据 量大、应用场景丰富情况下的算力不足以及训练推理时延大、实时性差的问题。
另外，由于基站部署为网状，不同的基站存在相同的应用场景，将不同的训练任务分别部署在不同基站上，各基站可以共享训练结果，即可支持全场景的AI训练，达到单点算力增强的效果，提高无线接入网处理能力。
图11示出了本申请的一个实施例提供的一种通信方法的流程示意图。该方法包括如下步骤：
S401：上层应用使用AI模型\推理结果。
部署辅AI训练推理引擎的基站(第三算力单元)，有上层应用需要使用相应的AI模型\推理结果；但由于本站负荷较高、空闲算力较少，本站之前没有进行对应的AI模型训练\推理，本站上层应用无法匹配到相应的AI模型\推理结果。
S402：模型\推理结果请求。
该部署辅AI训练推理引擎的基站向主AI训练推理引擎的基站请求对应的AI模型\推理结果。
S403：模型\推理结果更新。
主AI训练推理引擎的基站将对应的AI模型\推理结果发送给辅AI训练推理引擎的基站。
S404：AI模型\推理结果应用。
辅AI训练推理引擎的基站的上层应用调用此AI模型\推理结果完成上层业务使用。
S405：AI模型\推理结果应用反馈。
辅AI训练推理引擎将上层应用调用此AI模型\推理结果的效果进行评估。辅AI训练推理引擎将对应的评估结果反馈给主AI训练推理引擎，以使主AI训练推理引擎根据效果评估策略决定是否进行下一轮模型训练\模型推理任务。
以上步骤主要是基于分布式框架的基站间的AI模型\推理结果共享的流程，具体的部署原则为：负荷比较低，空闲算力比较多的基站部署主AI训练推理引擎；负荷比较高，空闲算力比较少的基站部署辅AI训练推理引擎。训练后的模型\推理结果都会在主AI训练推理引擎管理，部署辅AI训练推理引擎的基站通过直接使用主AI训练推理引擎的管理的AI模型\推理结果，完成上层业务应用。
另外，基站和边缘计算设备间分布式部署的方式及通信方法可以参见上述实施例相关的描述，其能达到相同的技术效果。避免重复，不再赘述。
需要说明的是，本申请实施例提供的通信方法，执行主体可以为通信装置，或者该通信装置中的用于执行通信方法的控制模块。本申请实施例中以通信装置执行通信的方法为例，说明本申请实施例提供的通信装置。
图12是本申请实施例提供的一种通信装置的结构示意图。如图12所示，该通信装置500包括：第一发送模块510、第一接收模块520和第二发送模块530。
第一发送模块510，用于向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；第一接收模块520，用于接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；第二发送模块530，用于在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。
本申请实施例提供的一种通信装置，通过第一发送模块，用于向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；第一接收模块，用于接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；第二发送模块，用于在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元，能够解决RAN在业务数据量大、 应用场景丰富情况下的算力不足以及训练推理时延大、实时性差的问题。
在一种实现方式中，所述第一接收模块520，还用于：接收所述第二算力单元发送的所述目标任务的处理结果。
在一种实现方式中，所述装置500还包括：第三发送模块，用于将所述目标任务的处理结果发送给第三算力单元；所述第一接收模块520，还用于：接收所述第三算力单元反馈的目标评估结果，其中，所述目标评估结果是对所述目标任务的处理结果进行评估得到的；所述装置500还包括：更新模块，用于根据所述目标评估结果，对所述目标任务的处理结果进行更新。
图13是本申请实施例提供的一种通信装置的结构示意图。如图13所示，该通信装置600包括：第二接收模块610、分配模块620和第四发送模块630。
第二接收模块610，用于接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；分配模块620，用于根据所述目标任务的特征为所述目标任务分配可用算力；第四发送模块630，用于向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力。
本申请实施例提供的一种通信装置，通过第二接收模块，用于接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；分配模块，用于根据所述目标任务的特征为所述目标任务分配可用算力；第四发送模块，用于向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力，能够解决RAN在业务数据量大、应用场景丰富情况下的算力不足以及训练推理时延大、实时性差的问题。
在一种实现方式中，第二接收模块610，还用于：接收所述目标任务，所述目标任务是所述第一算力单元在所述可用算力与所述目标任务的特征匹配的情况下发送的。
在一种实现方式中，所述装置600，还包括：执行模块，用于执行所述目标任务，获得所述目标任务的处理结果；所述第四发送模块630，还用于： 向所述第一算力单元发送所述目标任务的处理结果。
图14是本申请实施例提供的一种通信装置的结构示意图。如图14所示，该通信装置700包括：第三接收模块710、评估模块720和第五发送模块730。
第三接收模块710，用于接收第一算力单元发送的目标任务的处理结果；评估模块720，用于对所述目标任务的处理结果进行评估，得到目标评估结果；第五发送模块730，用于将所述目标评估结果反馈给所述第一算力单元。
在一种实现方式中，所述第一算力单元包括第一基站的第一主控板，所述第二算力单元包括所述第一基站的至少一个基带板、第二基站的至少一个基带板、与所述第一基站连接的至少一个边缘计算设备中的至少一者。
在一种实现方式中，在所述第二算力单元包括第二基站的至少一个基带板的情况下，所述第一基站与所述第二基站通过Xn口建立基站间连接；在所述第二算力单元包括所述第一基站的至少一个基带板的情况下，所述第一算力单元与所述第二算力单元通过虚拟可扩展局域网VXLAN连接。
在一种实现方式中，所述第三算力单元包括第一基站的至少一个基带板、第二基站的至少一个基带板、与所述第一基站连接的至少一个边缘计算设备中的至少一者；所述第三算力单元与所述第二算力单元不是同一个算力单元。
在一种实现方式中，所述目标任务包括：训练模型的任务；所述目标任务的处理结果包括：训练得到的模型；所述目标评估结果包括：对所述模型进行评估得到的评估结果。
本申请实施例提供的通信装置能够实现图1至图11至少一个实施例所述的通信方法实施例实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例中的通信装置可以是装置，也可以是终端设备中的部件、集成电路、或芯片。该装置可以是移动电子设备，也可以为非移动电子设备。示例性的，移动电子设备可以为手机、平板电脑、笔记本电脑、掌上电脑、车载电子设备、可穿戴设备、超级移动个人计算机(ultra-mobile personal  computer，UMPC)、上网本或者个人数字助理(personal digital assistant，PDA)等，非移动电子设备可以为服务器、网络附属存储器(Network Attached Storage，NAS)、个人计算机(personal computer，PC)、电视机(television，TV)、柜员机或者自助机等，本申请实施例不作具体限定。
本申请实施例中的通信装置可以为具有操作系统的装置。该操作系统可以为安卓(Android)操作系统，可以为ios操作系统，还可以为其他可能的操作系统，本申请实施例不作具体限定。
可选的，如图15所示，本申请实施例还提供一种电子设备800，包括处理器801，存储器802，存储在存储器802上并可在所述处理器801上运行的程序或指令，该程序或指令被处理器801执行时实现：图1至图11实施例中至少一个实施例所述的通信方法。需要说明的是，本申请实施例中的电子设备包括：服务器、终端设备或除终端设备之外的其他设备。
以上电子设备结构并不构成对电子设备的限定，电子设备可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置，例如，输入单元，可以包括图形处理器(Graphics Processing Unit，GPU)和麦克风，显示单元可以采用液晶显示器、有机发光二极管等形式来配置显示面板。用户输入单元包括触控面板以及其他输入设备中的至少一种。触控面板也称为触摸屏。其他输入设备可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆，在此不再赘述。
存储器可用于存储软件程序以及各种数据。存储器可主要包括存储程序或指令的第一存储区和存储数据的第二存储区，其中，第一存储区可存储操作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外，存储器可以包括易失性存储器或非易失性存储器，或者，存储器可以包括易失性和非易失性存储器两者。其中，非易失性存储器可以是只读存储器(Read-Only Memory，ROM)、可编程只读存储器(Programmable ROM，PROM)、可擦除可编程只读存储器(Erasable PROM， EPROM)、电可擦除可编程只读存储器(Electrically EPROM，EEPROM)或闪存。易失性存储器可以是随机存取存储器(Random Access Memory，RAM)，静态随机存取存储器(Static RAM，SRAM)、动态随机存取存储器(Dynamic RAM，DRAM)、同步动态随机存取存储器(Synchronous DRAM，SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM，DDRSDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM，ESDRAM)、同步连接动态随机存取存储器(Synch link DRAM，SLDRAM)和直接内存总线随机存取存储器(Direct Rambus RAM，DRRAM)。
处理器可包括一个或多个处理单元；可选的，处理器集成应用处理器和调制解调处理器，其中，应用处理器主要处理涉及操作系统、用户界面和应用程序等的操作，调制解调处理器主要处理通信信号，如基带处理器。可以理解的是，上述调制解调处理器也可以不集成到处理器中。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有程序或指令，该程序或指令被处理器执行时实现图1和图2实施例中至少一个实施例所述的通信方法，且能达到相同的技术效果，为避免重复，这里不再赘述。
其中，所述处理器为上述实施例中所述的电子设备中的处理器。所述可读存储介质，包括计算机可读存储介质，如计算机只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等。
本申请实施例另提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现上述通信方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
应理解，本申请实施例提到的芯片还可以称为系统级芯片、系统芯片、芯片系统或片上系统芯片等。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外，需要指出的是，本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能，还可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能，例如，可以按不同于所描述的次序来执行所描述的方法，并且还可以添加、省去、或组合各种步骤。另外，参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以计算机软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端(可以是手机，计算机，服务器，或者网络设备等)执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述，但是本申请并不局限于上述的具体实施方式，上述的具体实施方式仅仅是示意性的，而不是限制性的，本领域的普通技术人员在本申请的启示下，在不脱离本申请宗旨和权利要求所保护的范围情况下，还可做出很多形式，均属于本申请的保护之内。

Claims (13)

  1. 一种通信方法，由第一算力单元执行，所述方法包括：
    向第二算力单元发送目标启动信息，其中，所述目标启动信息中携带目标任务的特征；
    接收所述第二算力单元上报的算力度量信息，其中，所述算力度量信息用于表示所述第二算力单元分配给所述目标任务的可用算力；
    在所述可用算力与所述目标任务的特征匹配的情况下，将所述目标任务发送给所述第二算力单元。
  2. 根据权利要求1所述的方法，其中，在将所述目标任务发送给所述第二算力单元之后，还包括：
    接收所述第二算力单元发送的所述目标任务的处理结果。
  3. 根据权利要求2所述的方法，其中，在接收所述第二算力单元发送的所述目标任务的处理结果之后，还包括：
    将所述目标任务的处理结果发送给第三算力单元；
    接收所述第三算力单元反馈的目标评估结果，其中，所述目标评估结果是对所述目标任务的处理结果进行评估得到的；
    根据所述目标评估结果，对所述目标任务的处理结果进行更新。
  4. 一种通信方法，由第二算力单元执行，所述方法包括：
    接收第一算力单元发送的目标启动信息，其中，所述目标启动信息中携带目标任务的特征；
    根据所述目标任务的特征为所述目标任务分配可用算力；
    向所述第一算力单元发送的算力度量信息，其中，所述算力度量信息用于表示所述可用算力。
  5. 根据权利要求4所述的方法，其中，在所述向所述第一算力单元发送的算力度量信息之后，还包括：
    接收所述目标任务，所述目标任务是所述第一算力单元在所述可用算力与所述目标任务的特征匹配的情况下发送的。
  6. 根据权利要求5所述的方法，其中，在接收所述目标任务之后，还包括：
    执行所述目标任务，获得所述目标任务的处理结果；
    向所述第一算力单元发送所述目标任务的处理结果。
  7. 一种通信方法，由第三算力单元执行，所述方法包括：
    接收第一算力单元发送的目标任务的处理结果；
    对所述目标任务的处理结果进行评估，得到目标评估结果；
    将所述目标评估结果反馈给所述第一算力单元。
  8. 根据权利要求1至7任一所述的方法，其中，所述第一算力单元包括第一基站的第一主控板，所述第二算力单元包括所述第一基站的至少一个基带板、第二基站的至少一个基带板、与所述第一基站连接的至少一个边缘计算设备中的至少一者。
  9. 根据权利要求8所述的方法，其中，在所述第二算力单元包括第二基站的至少一个基带板的情况下，所述第一基站与所述第二基站通过Xn口建立基站间连接；
    在所述第二算力单元包括所述第一基站的至少一个基带板的情况下，所述第一算力单元与所述第二算力单元通过虚拟可扩展局域网VXLAN连接。
  10. 根据权利要求3或7所述的方法，其中，所述第三算力单元包括第一基站的至少一个基带板、第二基站的至少一个基带板、与所述第一基站连接的至少一个边缘计算设备中的至少一者；所述第三算力单元与所述第二算力单元不是同一个算力单元。
  11. 根据权利要求1至7任一所述的方法，其中，所述目标任务包括：训练模型的任务；所述目标任务的处理结果包括：训练得到的模型；所述目标评估结果包括：对所述模型进行评估得到的评估结果。
  12. 一种电子设备，包括处理器，存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求1至11任一项所述的通信方法的步骤。
  13. 一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如权利要求1至11任一项所述的通信方法的步骤。
PCT/CN2023/109115 2022-12-13 2023-07-25 一种通信方法、电子设备及存储介质 Ceased WO2024124909A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP23902119.9A EP4590020A4 (en) 2022-12-13 2023-07-25 COMMUNICATION METHOD, ELECTRONIC DEVICE AND STORAGE MEDIA

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211595595.0A CN118200979A (zh) 2022-12-13 2022-12-13 一种通信方法、电子设备及存储介质
CN202211595595.0 2022-12-13

Publications (1)

Publication Number Publication Date
WO2024124909A1 true WO2024124909A1 (zh) 2024-06-20

Family

ID=91402130

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/109115 Ceased WO2024124909A1 (zh) 2022-12-13 2023-07-25 一种通信方法、电子设备及存储介质

Country Status (3)

Country Link
EP (1) EP4590020A4 (zh)
CN (1) CN118200979A (zh)
WO (1) WO2024124909A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118802661B (zh) * 2024-06-24 2026-01-20 中国移动通信有限公司研究院 一种任务处理方法及装置、设备、存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021208914A1 (zh) * 2020-04-15 2021-10-21 展讯半导体(南京)有限公司 基于网络调度的算力共享方法及相关产品
WO2022028418A1 (zh) * 2020-08-04 2022-02-10 中国移动通信有限公司研究院 算力处理的网络系统、业务处理方法及设备
CN114168331A (zh) * 2021-12-07 2022-03-11 杭州萤石软件有限公司 算法部署和调度方法以及算法部署和调度装置
WO2022143744A1 (zh) * 2020-12-31 2022-07-07 维沃移动通信有限公司 信息处理方法、装置、设备及存储介质
WO2022143748A1 (zh) * 2020-12-31 2022-07-07 维沃移动通信有限公司 信息处理方法、装置、设备及存储介质
CN115373836A (zh) * 2022-05-09 2022-11-22 华为技术有限公司 计算网络、算力度量方法、调度装置及相关产品

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021208914A1 (zh) * 2020-04-15 2021-10-21 展讯半导体(南京)有限公司 基于网络调度的算力共享方法及相关产品
WO2022028418A1 (zh) * 2020-08-04 2022-02-10 中国移动通信有限公司研究院 算力处理的网络系统、业务处理方法及设备
WO2022143744A1 (zh) * 2020-12-31 2022-07-07 维沃移动通信有限公司 信息处理方法、装置、设备及存储介质
WO2022143748A1 (zh) * 2020-12-31 2022-07-07 维沃移动通信有限公司 信息处理方法、装置、设备及存储介质
CN114168331A (zh) * 2021-12-07 2022-03-11 杭州萤石软件有限公司 算法部署和调度方法以及算法部署和调度装置
CN115373836A (zh) * 2022-05-09 2022-11-22 华为技术有限公司 计算网络、算力度量方法、调度装置及相关产品

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4590020A4 *

Also Published As

Publication number Publication date
CN118200979A (zh) 2024-06-14
EP4590020A1 (en) 2025-07-23
EP4590020A4 (en) 2025-12-24

Similar Documents

Publication Publication Date Title
WO2021142609A1 (zh) 信息上报方法、装置、设备和存储介质
CN110096336B (zh) 数据监控方法、装置、设备和介质
US20200053144A1 (en) Method and system for coordination of inter-operable infrastructure as a service (iaas) and platform as a service (paas) systems
CN107515787A (zh) 资源配置方法及相关产品
CN107613107A (zh) 资源配置方法及相关产品
CN107483725A (zh) 资源配置方法及相关产品
CN107547745A (zh) 资源配置方法及相关产品
CN107894920A (zh) 资源配置方法及相关产品
CN115801569B (zh) 一种访问规则部署方法、装置、设备、介质及云平台
CN107450988A (zh) 资源配置方法及相关产品
EP2256633A2 (en) Service provider management device, service provider management program, and service provider management method
CN107621981A (zh) 资源配置方法及相关产品
CN109815202B (zh) 日志编辑方法及相关装置
CN118885180A (zh) 接入异构虚机镜像或容器集群编译的系统、方法、设备及介质
WO2024124909A1 (zh) 一种通信方法、电子设备及存储介质
CN120856413A (zh) 一种数据处理方法及相关装置
CN112527377B (zh) 应用程序生成处理方法、装置、计算机设备和存储介质
CN111767345B (zh) 建模数据同步方法、装置、计算机设备及可读存储介质
CN107479972A (zh) 资源配置方法及相关产品
CN118916130A (zh) 服务请求调度方法、装置、电子设备及存储介质
CN114928572B (zh) 一种分布式系统的流量控制方法、装置、介质及设备
CN110727511A (zh) 应用程序的控制方法、网络侧设备和计算机可读存储介质
CN113568708B (zh) 平台创建方法、装置及设备
CN116366618A (zh) 固定容器ip的网络插件方法及装置
CN115580614A (zh) 一种数据下载方法、装置、设备及计算机可读存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23902119

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2023902119

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2023902119

Country of ref document: EP

Effective date: 20250418

NENP Non-entry into the national phase

Ref country code: DE

WWP Wipo information: published in national office

Ref document number: 2023902119

Country of ref document: EP