Communication method and apparatus

Abstract

A communication method and an apparatus. The method comprises: a first network element receives a first message from a second network element, the first message being used for requesting a machine learning model, and the first message comprising energy consumption indication information; and the first network element sends model information of at least one machine learning model to the second network element, the at least one machine learning model being determined on the basis of the energy consumption indication information and energy consumption used for running the at least one machine learning model. The method indicates, when the second network element requests a machine learning model from the first network element, the energy consumption indication information that the machine learning model needs to satisfy, so that the first network element determines at least one machine learning model on the basis of the energy consumption indication information. Therefore, when running a machine learning model, the second network element can consider energy consumption of each machine learning model, so as to prevent the machine learning model from generating high energy consumption during running processes, thus implementing network energy saving.

Description

Communication method and device

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office of the People's Republic of China on December 29, 2023, with application number 202311871865.0 and invention name "A Communication Method and Device", the entire contents of which are incorporated by reference into this application. Technical Field

[0003] The present application relates to the field of communication technology, and in particular to a communication method and device. Background Art

[0004] Many use cases for machine learning (ML) models to assist in network analysis have been defined in current wireless networks, such as load analysis, user experience analysis, network performance analysis, user equipment (UE) mobility and communication characteristics analysis, and user data congestion analysis. To obtain ML models, network elements such as the access and mobility management function (AMF) or policy control function (PCF) can act as consumers and request one or more ML models that meet their needs from the network data analytics function (NWDAF).

[0005] Since each machine learning model requires a certain amount of energy consumption when running inference, when the energy consumption generated by the machine learning model obtained by consumers during inference is high, it will greatly increase the energy consumption of the network. Summary of the Invention

[0006] The present application provides a communication method and apparatus for saving network energy consumption.

[0007] In a first aspect, a communication method is provided, including: a first network element receives a first message from a second network element; the first message is used to request a machine learning model; the first message includes energy consumption indication information; the first network element sends model information of at least one machine learning model to the second network element, and the at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model.

[0008] Through the above method, when the second network element requests a machine learning model from the first network element, it indicates the energy consumption indication information that the machine learning model must meet, so that the first network element determines at least one machine learning model based on the energy consumption indication information. In this way, the machine learning model requested by the second network element is related to the energy consumption indication information. When running the machine learning model, the second network element can consider the energy consumption of each machine learning model, thereby avoiding high energy consumption during the operation of the machine learning model and achieving energy conservation in the network.

[0009] In one possible implementation, the method also includes: the first network element determines the at least one machine learning model based on the energy consumption indication information and the energy consumption used to run each of the at least one machine learning model.

[0010] In one possible implementation, the energy consumption indication information indicates that the second network element is in an energy-saving state; the first message also includes an analysis identifier, which is used to identify the analysis task; in the machine learning model used to implement the analysis task, the energy consumption used to run the at least one machine learning model is minimized.

[0011] Through the energy consumption indication information, the machine learning model with the minimum energy consumption is obtained, so that the analysis results of the analysis task are obtained with the minimum energy consumption, thereby saving the energy consumption of the network.

[0012] In one possible implementation, the energy consumption indication information includes an energy consumption threshold; the energy consumption used to run each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

[0013] By using the energy consumption threshold, a machine learning model is obtained whose energy consumption for operation is less than or equal to the energy consumption threshold, thereby avoiding high energy consumption of the running machine learning model and saving network energy consumption.

[0014] In one possible implementation, the method further includes: the first network element determines the energy consumption for running the at least one machine learning model based on the configuration information of the third network element; wherein the third network element is used to run the machine learning model in the at least one machine learning model.

[0015] In a possible implementation, the first message also includes configuration information of the third network element.

[0016] In one possible implementation, the first message also includes the identifier of the third network element; the method also includes: the first network element sends the identifier of the third network element to the fourth network element; the first network element receives the configuration information of the third network element from the fourth network element.

[0017] In one possible implementation, the model information includes first energy consumption information, and the first energy consumption information indicates the energy consumption used to run the at least one machine learning model.

[0018] By indicating the energy consumption used to run each machine learning model through the first energy consumption information, the energy consumption of each machine learning model can be considered when running the machine learning model, thereby saving network energy consumption.

[0019] In one possible implementation, the method further includes: the first network element sends a registration request message to the fifth network element, the registration request message includes capability information, and the capability information indicates that the first network element has the ability to determine the energy consumption used to run the machine learning model.

[0020] According to a second aspect, a communication method is provided, including: a second network element sends a first message to a first network element; the first message is used to request a machine learning model; the first message includes energy consumption indication information; the second network element receives model information of at least one machine learning model from the first network element, and the at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model.

[0021] In one possible implementation, before sending the first message, the method further includes: the second network element receives an analysis request message from the sixth network element, the analysis request message includes energy consumption request information, and the energy consumption request information is used to indicate a request to determine energy consumption of the analysis result.

[0022] In a possible implementation manner, the energy consumption indication information is determined by the second network element according to the energy consumption request information.

[0023] In one possible implementation, the model information includes first energy consumption information, and the first energy consumption information indicates the energy consumption used to run the at least one machine learning model.

[0024] In a possible implementation, the method further includes: the second network element determining a first machine learning model from the at least one machine learning model according to the first energy consumption information; and the second network element determining the analysis result according to the first machine learning model;

[0025] The second network element sends the analysis result and second energy consumption information to the sixth network element, where the second energy consumption information indicates energy consumption for determining the analysis result.

[0026] In one possible implementation, the second energy consumption information is determined based on the first energy consumption consumed for running the first machine learning model to obtain the analysis result; or, the second energy consumption information is determined based on the second energy consumption, and the second energy consumption is the energy consumption indicated by the first energy consumption information for running the first machine learning model.

[0027] In one possible implementation, the energy consumption indication information indicates that the second network element is in an energy-saving state; the first message also includes an analysis identifier, which is used to identify the analysis task; in the machine learning model used to implement the analysis task, the energy consumption used to run the at least one machine learning model is minimized.

[0028] In one possible implementation, the energy consumption indication information includes an energy consumption threshold; the energy consumption used to run each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

[0029] In one possible implementation, the first message also includes at least one of the following: configuration information of a third network element, or an identifier of the third network element; the third network element is used to run a machine learning model in the at least one machine learning model.

[0030] In one possible implementation, the method also includes: the second network element sends a second message to the fifth network element, the second message being used to request a network element capable of determining the energy consumption for running a machine learning model; the second network element receives the address information of the first network element from the fifth network element; and the second network element sends the first message to the first network element based on the address information of the first network element.

[0031] According to a third aspect, a communication method is provided, including: a second network element sends a first message to a first network element; the first message is used to request a machine learning model; the first message includes energy consumption indication information; the first network element receives the first message from the second network element; the first network element sends model information of at least one machine learning model to the second network element, and the at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model; the second network element receives model information of at least one machine learning model from the first network element.

[0032] In one possible implementation, the method also includes: the first network element determines the at least one machine learning model based on the energy consumption indication information and the energy consumption used to run each of the at least one machine learning model.

[0033] In one possible implementation, the energy consumption indication information indicates that the second network element is in an energy-saving state; the first message also includes an analysis identifier, which is used to identify the analysis task; in the machine learning model used to implement the analysis task, the energy consumption used to run the at least one machine learning model is minimized.

[0034] In one possible implementation, the energy consumption indication information includes an energy consumption threshold; the energy consumption used to run each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

[0035] In one possible implementation, the method further includes: the first network element determines the energy consumption for running the at least one machine learning model based on the configuration information of the third network element; wherein the third network element is used to run the machine learning model in the at least one machine learning model.

[0036] In a possible implementation, the first message also includes configuration information of the third network element.

[0037] In one possible implementation, the first message also includes an identifier of the third network element; the method also includes: the first network element sends the identifier of the third network element to the fourth network element; and the first network element receives configuration information of the third network element from the fourth network element.

[0038] In one possible implementation, the model information includes first energy consumption information, and the first energy consumption information indicates the energy consumption used to run the at least one machine learning model.

[0039] In one possible implementation, the method further includes: the first network element sends a registration request message to the fifth network element, the registration request message includes capability information, and the capability information indicates that the first network element has the ability to determine the energy consumption used to run the machine learning model.

[0040] In one possible implementation, the method also includes: the second network element sends a second message to the fifth network element, the second message being used to request a network element capable of determining energy consumption for running a machine learning model; the second network element receives the address information of the first network element from the fifth network element; and the second network element sends the first message to the first network element based on the address information of the first network element.

[0041] In a fourth aspect, the present application further provides a communication device capable of implementing any of the methods provided in any of the first to second aspects above. The communication device can be implemented via hardware, or by hardware executing corresponding software implementations. The hardware or software includes one or more units or modules corresponding to the above functions.

[0042] In one possible implementation, the communication device includes a processor configured to support the communication device in executing the corresponding functions of the first network element or the second network element in the method described above. The communication device may also include a memory, which may be coupled to the processor and stores program instructions and data necessary for the communication device. Optionally, the communication device also includes an interface circuit configured to support communication between the communication device and other devices.

[0043] In one possible implementation, the communication device includes corresponding functional modules for implementing the steps in the above method. The functions can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions.

[0044] In one possible implementation, the structure of the communication device includes a processing unit and a communication unit, which can perform the corresponding functions in the above method examples. For details, please refer to the description of the method provided in any one of the first aspect to the second aspect, which will not be repeated here.

[0045] In a fifth aspect, a communication device is provided, comprising a processor and an interface circuit, wherein the interface circuit is configured to receive signals from a communication device other than the communication device and transmit them to the processor, or to transmit signals from the processor to a communication device other than the communication device, wherein the processor implements the functional modules of the method in any possible implementation of any of the first and second aspects through logic circuits or by executing computer programs or instructions. Optionally, the communication device further includes a memory configured to store the computer program or instructions.

[0046] In a sixth aspect, a computer program product storing instructions is provided, which, when read and executed by a computer, implements the method in any possible implementation of any one of the first to second aspects.

[0047] In a seventh aspect, a circuit is provided, which is used to execute the method in any possible implementation of any one of the first to second aspects above, and the circuit may include a chip circuit. Optionally, the circuit may also be coupled to a memory.

[0048] In an eighth aspect, a chip is provided, comprising a processor. When the processor executes a computer program or instruction, the processor is configured to implement the method of any possible implementation of any of the first and second aspects. Optionally, the chip may further include a memory. The chip may be composed of a single chip or may include a chip and other discrete devices.

[0049] In a ninth aspect, a communication device is provided, comprising a processor, which implements the method in any possible implementation of any one of the first to second aspects through a logic circuit or by executing a computer program or instruction.

[0050] In a tenth aspect, a communication device is provided, comprising a unit or module for executing the method in any possible implementation of any one of the first to second aspects above.

[0051] In the eleventh aspect, a computer-readable storage medium is provided, which stores a computer program or instruction. When the computer program or instruction is executed by a processor, the method in any possible implementation of any one of the first to second aspects is implemented.

[0052] In a twelfth aspect, an embodiment of the present application further provides a communication system. The communication system includes: a first network element for implementing the method in the aforementioned first aspect and any possible implementation of the first aspect; and a second network element for implementing the method in the aforementioned second aspect and any possible implementation of the second aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] FIG1 is a schematic diagram of a network architecture applicable to an embodiment of the present application;

[0054] FIG2 is a schematic diagram of a model subscription process provided in an embodiment of the present application;

[0055] FIG3 is a flow chart of a communication method provided in an embodiment of the present application;

[0056] FIG4 is a flow chart of a communication method provided in an embodiment of the present application;

[0057] FIG5 is a flow chart of a communication method provided in an embodiment of the present application;

[0058] FIG6 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application;

[0059] FIG7 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application;

[0060] FIG8 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Obviously, the embodiments described are only a part of the embodiments of the present application, not all of the embodiments. The terms "first", "second" and corresponding terminology labels in the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances. This is merely a way of distinguishing objects with the same properties when describing the embodiments of the present application. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, so that a process, method, system, product or device that includes a series of units is not necessarily limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or devices. The methods and devices provided in the embodiments of the present application are based on the same or similar technical concepts. Since the principles of solving problems by the methods and devices are similar, the implementation of the devices and methods can refer to each other, and the repetitions will not be repeated.

[0062] The method provided in the embodiment of the present application can be applied to various mobile communication systems, for example, the Internet of Things (IoT), narrowband Internet of Things (NB-IoT), a fourth generation (4G) communication system (such as long term evolution (LTE)), a fifth generation (5G) communication system (such as 5G new radio (NR)), a hybrid architecture of LTE and NR, 6G or new communication systems emerging in future communication developments, etc. The communication system may also include a machine to machine (M2M) network, a machine type communication (MTC) or other networks.

[0063] Figure 1 is a schematic diagram of a 5G network architecture based on a service-oriented architecture. The 5G network architecture shown in Figure 1 includes terminal devices, access network devices, and core network devices. Terminal devices access the data network (DN) through the access network and core network devices. Among them, the core network equipment includes a variety of network functions (NF) or network elements, for example, including some or all of the following network elements: operations administration management (OAM) network element, unified data management (UDM) network element, unified data repository (UDR) network element, application function (AF) network element, policy control function (PCF) network element, access and mobility management function (AMF) network element, session management function (SMF) network element, user plane function (UPF) network element, network data analysis function (NWDAF) network element, network exposure function (NEF) network element, binding support function (BSF) network element, network repository function (NRF) network element (not shown in the figure), etc.

[0064] The following is a brief introduction to some core network equipment:

[0065] The AMF network element, referred to as AMF, performs functions such as mobility management and access authentication / authorization. It is also responsible for transmitting user policies between terminal devices and the PCF.

[0066] The SMF network element, referred to as SMF, includes functions such as session management, execution of control policies issued by PCF, selection of UPF, and allocation of Internet Protocol (IP) addresses for terminal devices.

[0067] The UPF network element, referred to as UPF, serves as the interface with the data network and includes functions such as user plane data forwarding, session / flow-level billing statistics, and bandwidth limitation.

[0068] UDM network element, referred to as UDM, includes functions such as executing and managing contract data and user access authorization.

[0069] The UDR network element, referred to as UDR, includes the access functions of executing contract data, policy data, application data and other types of data.

[0070] NEF network element, referred to as NEF, is used to support the opening of capabilities and events.

[0071] The AF network element, abbreviated as AF, conveys the application side's requirements to the network side, such as quality of service (QoS) requirements or user status event subscriptions. The AF can be a third-party functional entity or an application server deployed by the operator.

[0072] The PCF network element, referred to as PCF, includes policy control functions such as session and service flow level billing, QoS bandwidth guarantee and mobility management, and terminal device policy decision-making.

[0073] NRF network elements, or NRF for short, can be used to provide network element discovery capabilities. Based on requests from other network elements, they provide network element information corresponding to the network element type. NRF network elements also provide network element management services, such as network element registration, update, and deregistration, as well as network element status subscription and push notification.

[0074] NWDAF network element, referred to as NWDAF, is mainly used to collect data (including one or more of terminal device data, access network device data, core network network element data and third-party application device data), wherein these data can be the terminal device, access network device, core network network element or third-party application device data itself, or the terminal device on the access network device, the core network network element or the third-party application device. Then, data analysis is performed based on the collected data, and the data analysis results are output for use by the network, network management equipment and application execution policy decision-making. NWDAF can use machine learning models for data analysis. In an embodiment of the present application, an NWDAF can be a separate network element, or it can be set up together with other network elements, for example, NWDAF is set up in a PCF network element or an AMF network element.

[0075] In Release 17 of the 3rd Generation Partnership Project (3GPP), the training and inference functions of NWDAF are split. An NWDAF can support only the model training function, only the data inference function, or both the model training and data inference functions.

[0076] In this application, NWDAF that supports model training function, namely NWDAF (MTLF), may also be referred to as training NWDAF, or NWDAF that supports model training logical function (MTLF), referred to as MTLF. For example, MTLF can perform model training based on the acquired data to obtain a trained model.

[0077] NWDAF that supports data reasoning function, namely NWDAF (AnLF), can also be called reasoning NWDAF, or NWDAF that supports analytics logical function (AnLF), abbreviated as AnLF.

[0078] It is understood that MTLF can be understood as NWDAF that at least supports model training. As a possible implementation method, MTLF can also support data reasoning. AnLF can be understood as NWDAF that at least supports data reasoning. As a possible implementation method, AnLF can also support model training.

[0079] It can be understood that the above network elements are examples of one implementation method, and this application does not exclude the existence of network elements or devices with the above network element functions in 6G or newer wireless communication systems that have other names or other forms.

[0080] It is understood that the above-mentioned network element or function can be a network element in a hardware device, a software function running on dedicated hardware, or a virtualized function instantiated on a platform (e.g., a cloud platform). As a possible implementation method, the above-mentioned network element or function can be implemented by a single device, or can be implemented by multiple devices together, or can be a functional module within a single device, which is not specifically limited in the embodiments of the present application.

[0081] In Figure 1, Nudr, Npcf, Namf, Nudm, Nsmf, Naf, and Nnwdaf are the service interfaces provided by the above-mentioned UDR, PCF, AMF, UDM, SMF, AF, and NWDAF, respectively, which are used to call corresponding service operations; N2 is the service interface between RAN and AMF; N3, N4, and N6 are the service interfaces between RAN, SMF, DN, and UPF, respectively.

[0082] Below, some of the terms used in the embodiments of the present application are first explained to facilitate understanding by those skilled in the art. The explanations of these terms are only examples and do not represent limitations on such terms.

[0083] In the embodiment of the present application, the access network device may be a device in a wireless network, and the access network device may also be referred to as an access network apparatus or a wireless access network device. For example, the access network device may be a radio access network (RAN) node that connects a terminal device to a wireless network. Access network devices include, but are not limited to, base stations, evolved NodeBs (eNodeBs), transmission reception points (TRPs), next-generation NodeBs (gNBs) in fifth-generation (5G) mobile communication systems, access network devices in open radio access networks (O-RANs), next-generation base stations in sixth-generation (6G) mobile communication systems, base stations in future mobile communication systems, or access nodes in wireless fidelity (WiFi) systems; or may be modules or units that perform some functions of a base station, for example, a centralized unit (CU), a distributed unit (DU), a centralized unit control plane (CU-CP) module, or a centralized unit user plane (CU-UP) module. The access network device may be a macro base station, a micro base station, an indoor station, a relay node, a donor node, etc. The present application does not limit the specific technology and specific device form used by the access network device.

[0084] In some implementations, access network equipment may include a centralized unit (CU) and a distributed unit (DU). RAN equipment, including CU and DU nodes, splits the protocol layers of the gNB in ​​the NR system. Some protocol layer functions are centrally controlled by the CU, while some or all of the remaining protocol layer functions are distributed in the DU, which is centrally controlled by the CU. Furthermore, the CU can be divided into a control plane (CU-CP) and a user plane (CU-UP). The CU-CP is responsible for control plane functions, primarily including radio resource control (RRC) and the control plane's corresponding packet data convergence protocol (PDCP) (i.e., PDCP-C). PDCP-C is primarily responsible for encryption and decryption, integrity protection, and data transmission of control plane data. The CU-UP is responsible for user plane functions, primarily including the service data adaptation protocol (SDAP) and the user plane's corresponding PDCP (i.e., PDCP-U). SDAP is primarily responsible for processing core network data and mapping flows to bearers. The PDCP-U is primarily responsible for data plane encryption and decryption, integrity protection, header compression, sequence number maintenance, and data transmission. The CU-CP and CU-UP are connected via the E1 interface. The CU-CP represents the gNB's connection to the core network via the NG interface and to the DU via the F1 interface control plane (i.e., F1-C). The CU-UP connects to the DU via the F1 interface user plane (i.e., F1-U). Alternatively, the PDCP-C may also reside in the CU-UP.

[0085] It is understandable that in different systems, CU (including CU-CP or CU-UP) or DU may have different names, but those skilled in the art can understand their meanings. For example, in an open radio access network (O-RAN) system, CU may also be called O-CU (open CU), DU may also be called O-DU, CU-CP may also be called O-CU-CP, and CU-UP may also be called O-CU-UP. For convenience of description, this application uses CU, CU-CP, CU-UP and DU as examples for description. The access network equipment may also include an active antenna unit (AAU). The CU implements some functions of the gNB, and the DU implements some functions of the gNB. For example, the CU is responsible for processing non-real-time protocols and services and implementing the functions of the RRC layer. The DU is responsible for processing physical layer protocols and real-time services and implementing the functions of the radio link control (RLC) layer, the media access control (MAC) layer and the physical (PHY) layer. In some deployments, the CU can be further divided into a Centralized Unit Control Plane (CU-CP) node and a Centralized Unit User Plane (CU-UP) node, where the CU-CP is responsible for control plane functions and the CU-UP is responsible for user plane functions.

[0086] The terminal device involved in the embodiments of the present application may be a wireless terminal device capable of receiving scheduling and instruction information from a network device. The terminal device may be referred to as a terminal device, and may also be referred to as user equipment (UE), terminal, mobile station (MS), mobile terminal (MT), etc. The terminal device may be a device that includes wireless communication capabilities (providing voice / data connectivity to the user). For example, a handheld device with wireless connection capabilities, or an in-vehicle device, in-vehicle module, etc. Currently, some examples of terminal devices include: mobile phones, tablet computers, laptop computers, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in the Internet of Vehicles, wireless terminals in self-driving, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, or wireless terminals in smart homes, device-to-device (D2D) communication terminal devices, vehicle-to-everything (V2X) communication terminal devices, smart vehicles, telematics boxes (T-boxes), machine-to-machine / machine-type communications (M2M / MTC) terminal devices, Internet of Things (IoT) The IoT (Internet of Things) terminal devices, etc. For example, the terminal device can be an onboard device, complete vehicle equipment, an onboard module, a vehicle, an onboard unit (OBU), a roadside unit (RSU), a T-box, a chip, or a system on chip (SOC), etc. The above chip or SOC can be installed in the vehicle, OBU, RSU, or T-box. Wireless terminals in industrial control can be cameras, robots, etc. Wireless terminals in smart homes can be TVs, air conditioners, vacuum cleaners, speakers, set-top boxes, etc.The terminal device can also be a V2X device, such as a smart car (or intelligent car), a digital car, an unmanned car (or driverless car or pilotless car or automobile), a self-driving car or autonomous car, a pure electric vehicle (or Battery EV), a hybrid electric vehicle (HEV), a range extended EV (REEV), a plug-in hybrid electric vehicle (PHEV), a new energy vehicle (new energy vehicle), and a roadside unit (RSU). The terminal device can also be a device in device-to-device (D2D) communication, such as an electricity meter, a water meter, etc. In addition, in an embodiment of the present application, the terminal device can also be a terminal device in an IoT system. IoT is an important part of the future development of information technology. Its main technical feature is to connect objects to the network through communication technology, thereby realizing an intelligent network of human-machine interconnection and object-to-object interconnection.

[0087] In this application, predefined content generally refers to information that is defined by standards and does not require additional device configuration. It is pre-recorded / written in the hardware and / or software of the terminal device itself, or it can be understood as not being modifiable by the network device or other terminal devices. Pre-configured content generally refers to information that is pre-recorded / written in the hardware and / or software of the terminal device itself, determined by the equipment manufacturer, and can be modified through software or hardware.

[0088] Analysis ID: This ID can be used to identify an analysis task, also known as an analysis business or analysis service. This analysis task is associated with a machine learning model, meaning that the machine learning model can be used to perform the analysis task.

[0089] Or it can also be understood that MTLF is related to the analysis identifier, that is, the machine learning model support provided by MTLF is used to execute the analysis task corresponding to the analysis identifier. Exemplarily, MTLF can be related to one or more analysis identifiers. It can be understood that the MTLF can provide a machine learning model for the analysis task corresponding to each analysis identifier in one or more analysis identifiers. For example, MTLF1 is related to analysis identifier 1 and analysis identifier 2, that is, MTLF1 can provide a machine learning model for the analysis task corresponding to analysis identifier 1, and a machine learning model for the service corresponding to analysis identifier 2.

[0090] Model Subscription:

[0091] NWDAF (MTLF) trains the machine learning model based on its relevant data. NWDAF (AnLF) can request the machine learning model from NWDAF (MTLF) via a model subscription message. NWDAF (AnLF) can input the input data into the machine learning model obtained from NWDAF (MTLF) to obtain the analysis results output by the machine learning model.

[0092] For example, as shown in FIG2 , a schematic diagram of a model subscription process is provided in an embodiment of the present application.

[0093] Step 201: NWDAF (AnLF) sends a model subscription request message to NWDAF (MTLF).

[0094] The model subscription request message includes information such as an analysis identifier.

[0095] The model subscription request message may further include information such as accuracy level requirement (Accuracy level(s) of Interest) information and ML model filter information (ML Model Filter Information).

[0096] Step 202: NWDAF (MTLF) sends a model subscription response message to NWDAF (AnLF).

[0097] The model subscription response message includes model information of the machine learning model used to implement the analysis task corresponding to the analysis identifier.

[0098] If the model subscription message includes accuracy level requirement information, the accuracy of the machine learning model indicated by the model subscription response message also meets the accuracy indicated by the accuracy level requirement information.

[0099] If the model subscription message includes ML model filtering information, the machine learning model indicated by the model subscription response message is the model filtered according to the ML model filtering information.

[0100] If NWDAF (AnLF) needs to determine the analysis results through a machine learning model, when the energy consumption generated by the model provided by NWDAF (MTLF) during operation is high, it will greatly increase the energy consumption of the network, which is not conducive to network energy saving.

[0101] To this end, this application provides a method that can make the model provided by NWDAF (MTLF) meet the energy consumption requirements of NWDAF (AnLF), reduce energy consumption, and save network energy consumption, which will be described in detail below.

[0102] In this application, the names of each message and each information are just examples. With the evolution of network architecture and the emergence of new business scenarios, the names of messages or information may change accordingly, but the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0103] As shown in FIG3 , it is a flow chart of a communication method provided in an embodiment of the present application.

[0104] When this method is applied to the system shown in Figure 1, the first network element may be a network element such as NWDAF (MTLF), that is, a NWDAF that supports model training function; the second network element may be a network element such as NWDAF (AnLF) or UE; the third network element may be used to run a machine learning model, for example, it may be a network element such as NWDAF (AnLF) or UE; the fourth network element may be a network element such as OAM; the fifth network element may be a network element such as NRF; the sixth network element may be a network element such as AF, AMF, PCF, SMF, or UE.

[0105] In the present application, the first network element has the ability to determine the energy consumption used to run the machine learning model. For example, a model energy consumption prediction model is pre-configured in the first network element, and the first network element can determine the energy consumption used to run each machine learning model based on the model energy consumption prediction model. Optionally, the first network element also has a model training function. For example, the first network element can train the machine learning model based on training data to obtain a trained machine learning model.

[0106] Step 301: The second network element sends a first message to the first network element.

[0107] Correspondingly, the first network element receives the first message from the second network element.

[0108] The first message is used to request a machine learning model. This application does not limit the name of the first message. For example, the first message can also be called a NnwdafML model subscription (Nnwdaf_MLModelProvison_Subscribe) message.

[0109] This application does not limit the circumstances under which the second network element sends the first message. For example, the second network element receives a request message from the sixth network element, where the request message is used to request the analysis result of an analysis task. The second network element sends the first message to the first network element to request the machine learning model corresponding to the analysis task. The second network element obtains the analysis result by running the machine learning model based on the machine learning model received from the first network element and the collected data as the input data of the model, and finally sends the analysis result to the sixth network element. For another example, the second network element receives a request message from the sixth network element, where the request message is used to request the machine learning model. The second network element can then request the machine learning model from the first network element that stores the machine learning model.

[0110] The first message may include an analysis identifier, and the analysis identifier is used to identify the analysis task. In this case, the machine learning model requested by the first message is used to implement the analysis task corresponding to the analysis identifier. It can be understood that the machine learning model requested by the first message is used to determine the analysis result of the analysis task corresponding to the analysis identifier. The specific type of analysis task is not limited. For example, the analysis task includes but is not limited to any of the following: NF load analysis (NF load analytics), network performance analysis (Network Performance Analytics), UE mobility analytics (UE mobility analytics), UE communication analysis (UE communication analytics), or user data congestion analysis (User Data Congestion Analytics).

[0111] The first message includes energy consumption indication information, which can be used to indicate the energy consumption conditions that the machine learning model needs to meet. For example, in implementation method one, the energy consumption indication information includes an energy consumption threshold or an energy consumption level, which can be understood as the energy consumption conditions including: the energy consumption used to run (or infer) the machine learning model is less than or equal to the energy consumption threshold, or the energy consumption used to run the machine learning model complies with the energy consumption level, such as the energy consumption level includes high, medium, and low levels, where high, medium, and low respectively correspond to an energy consumption range, and the energy consumption range is preset, or the energy consumption range is a range agreed upon by each network element when calculating energy consumption.

[0112] Implementation method two: the energy consumption indication information indicates that the second network element is in an energy-saving state. At this time, the energy consumption indication information implicitly indicates a request for a machine learning model with the minimum energy consumption. It can be understood that the energy consumption conditions include: among the machine learning models used to implement the analysis task, at least one machine learning model with the minimum energy consumption used for operation.

[0113] Implementation method three, the energy consumption indication information directly indicates at least one machine learning model with the lowest energy consumption for operation among the machine learning models used to implement the analysis task.

[0114] The above are just examples. The specific implementation method of the energy consumption indication information is not limited in this application and will not be explained one by one here.

[0115] Optionally, the first message may further include at least one of model quantity information, accuracy level(s) of Interest information, or ML model filter information. The model quantity information indicates the number N of requested machine learning models, where N is an integer greater than 0; the accuracy level requirement information indicates the accuracy requirement for the analysis results output by the machine learning model; and the ML model filter information includes at least one filter parameter, where the ML model filter information indicates whether the machine learning model satisfies the filter parameters included in the ML model filter information. For example, the filter parameters include but are not limited to an area of ​​interest (Area of ​​Interest), a QoS requirement (QoS requirement), and the like.

[0116] Optionally, the first message may further include at least one of configuration information of the third network element, an identifier of the third network element, or platform type information. The configuration information indicates the hardware configuration and / or software configuration of the third network element. For example, the hardware configuration indicated by the configuration information includes the central processing unit (CPU) type and / or graphics processing unit (GPU) type of the third network element, and the software configuration indicated by the configuration information includes information such as the operating system type and the operating system version. The platform type information indicates the type of the platform of the third network element, and the platform types include general hardware and cloud-based virtual network elements.

[0117] The third network element is used to run the machine learning model. The third network element can be the same network element as the second network element. For example, the second network element is an NWDAF (AnLF), and the third network element is an NWDAF (AnLF). In this case, the second network element itself runs the machine learning model requested from the first network element. Alternatively, the third network element and the second network element are different network elements. The second network element is an NWDAF (AnLF), an AMF, or an LMF, and the third network element is a UE. In this case, the second network element forwards the model information of the machine learning model returned by the first network element to the third network element, and the third network element runs the machine learning model.

[0118] In one implementation, before sending the first message, the second network element may send a third message to the fourth network element. The third message includes an identifier of the third network element and is used to request configuration information of the third network element. The fourth network element may then send the configuration information of the third network element to the second network element. The third network element may be different from the second network element. Alternatively, the third network element may be the same as the second network element. When the third network element is the same as the second network element, the identifier carried in the third message request is the identifier of the second network element, and the fourth network element returns the configuration information of the second network element.

[0119] The first message may also include other information, which is not limited in this application and will not be illustrated one by one here.

[0120] Step 302: The first network element sends model information of at least one machine learning model to the second network element.

[0121] Accordingly, the second network element receives model information of at least one machine learning model from the first network element. The model information may indicate each of the at least one machine learning model. For example, the model information may indicate at least one of the model size, model algorithm, model depth, model architecture, or number of model parameters and model parameter values ​​of each of the at least one machine learning model. The at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model.

[0122] The first network element may determine, based on the analysis identifier in the first message, all machine learning models used to implement the analysis task corresponding to the analysis identifier. If the first message also includes ML model filtering information, the first network element may further select a machine learning model that satisfies the ML model filtering information from all machine learning models of the analysis task. Similarly, if the first message also includes accuracy level requirement information, the first network element may further select a machine learning model that satisfies the accuracy level requirement information from all machine learning models of the analysis task.

[0123] Furthermore, the first network element can determine the energy consumption of running each machine learning model that meets the accuracy level requirement information and / or ML model filtering information among all machine learning models used to implement the analysis task based on the configuration information of the third network element. Among them, if the first message does not include configuration information, the first network element needs to obtain configuration information from the fourth network element. If the first message also includes the identifier of the third network element, it means that the third network element and the second network element are different network elements. What the first network element obtains from the fourth network element is the configuration information of the third network element. For example, the first network element sends the identifier of the third network element to the fourth network element to request the configuration information of the third network element. If the first message does not include the identifier of the third network element, it means that the third network element and the second network element are the same network element. What the first network element obtains from the fourth network element is the configuration information of the second (third) network element.

[0124] This application does not limit how the first network element specifically determines the energy consumption for running each machine learning model based on the configuration information of the third network element. For example, in a first implementation method, the first network element can obtain an energy consumption information table, which indicates the energy consumption required for running a machine learning model to process unit data in unit time under different hardware configurations, different software configurations, and different model frameworks. The first network element can query the energy consumption of each machine learning model from the energy consumption information table based on the configuration information. Among them, this application does not limit how the first network element obtains the energy consumption information table. For example, the first network element obtains the energy consumption information table from the OAM network element, or it can obtain the energy consumption information table from a newly defined network element. The newly defined network element can be called an energy consumption information table storage network element or other names.

[0125] For example, the energy consumption information table may be as shown in Table 1.

[0126] Table 1

[0127] The above are just examples. The energy consumption information table may also include other parameters, such as the depth of the model, the algorithm used to train the model, and other information.

[0128] Assume that the configuration information indicates that the hardware configuration of the third network element is CPU type a1 and GPU type b1; the software configuration is Android operating system version V3.5. The first network element determines that the analysis task corresponds to two machine learning models, Model 1 and Model 2. The model framework of Model 1 is TensorFlow, and the model framework of Model 2 is pyTorch. Assuming that the data to be processed is 20 units of data, combined with the energy consumption information table shown in Table 1, it can be determined that the energy consumption for running Model 1 is 20*13=260 joules, and the energy consumption for running Model 2 is 20*15=300 joules. Alternatively, assuming that 260 joules corresponds to a low energy consumption level and 300 joules corresponds to a medium energy consumption level, then based on the configuration information, the energy consumption level for running Model 1 is determined to be low, and the energy consumption level for running Model 2 is determined to be medium.

[0129] For another example, in the second implementation method, the first network element includes a model energy consumption prediction model, which is a trained machine learning model. The model energy consumption prediction model is used to predict the energy consumption used to run the machine learning model. For each machine learning model among all machine learning models used to implement the analysis task, the first network element inputs the configuration information of the third network element running the machine learning model, the model information of the machine learning model (including at least one of the size of the model, the algorithm of the model, the depth of the model, the number of parameters of the model, and the length of time the model runs on the training function) and other information into the model energy consumption prediction model. The model energy consumption prediction model can then output the specific energy consumption or energy consumption level used to run the machine learning model.

[0130] The above are just examples. There may be other implementations for how the first network element determines the energy consumption for running the machine learning model, and we will not give examples one by one here.

[0131] In the present application, if the first message also includes platform type information, the first network element can also determine whether to correct the energy consumption used to run the machine learning model based on the platform type information. For example, if the type of the platform indicates that the type of the third network element is general hardware, then the energy consumption used to run the machine learning model may not be corrected. If the type of the platform indicates that the type of the third network element is a cloud-based virtual network element, then the energy consumption used to run the machine learning model is corrected, for example, each energy consumption used to run the machine learning model is multiplied by a correction coefficient, and the correction coefficient is a number greater than 1. The correction coefficient of each machine learning model can be the same, and the correction coefficient can be preset.

[0132] Furthermore, the first network element may determine at least one machine learning model based on the energy consumption indication information and the energy consumption used to run each machine learning model.

[0133] For example, in implementation method one, the energy consumption indication information includes an energy consumption threshold, then the energy consumption determined by the first network element for running each machine learning model in at least one machine learning model is less than or equal to the energy consumption threshold.

[0134] Optionally, if the first message also includes model quantity information, the model quantity information indicates the model number N. When the number of machine learning models whose energy consumption is less than or equal to the energy consumption threshold among all machine learning models used to implement the analysis task is greater than N, the first network element may select N machine learning models as at least one machine learning model from multiple machine learning models whose energy consumption is less than or equal to the energy consumption threshold, for example, select N machine learning models with the lowest energy consumption for running machine learning models from multiple machine learning models whose energy consumption is less than or equal to the energy consumption threshold.

[0135] Among them, if the first message also includes accuracy level requirement information and / or ML model filtering information, then at least one machine learning model determined by the first network element meets the accuracy level requirement information and / or ML model filtering information.

[0136] In a second implementation method, the energy consumption indication information includes an energy consumption level. In this case, the energy consumption used to run each machine learning model in the at least one machine learning model determined by the first network element complies with the energy consumption level, for example, the energy consumption level is high, medium, or low. If the first message also includes model quantity information, the model quantity information indicates the number of models N. The first network element then selects one or more models that comply with the energy consumption level from all machine learning models used to implement the analysis task. The at least one machine learning model determined by the first network element is the N machine learning models that comply with the energy consumption level among all machine learning models used to implement the analysis task.

[0137] Implementation method three: the energy consumption indication information indicates that the second network element is in an energy-saving state or the network is in an energy-saving state, then the first network element selects a machine learning model with the lowest energy consumption from all machine learning models used to implement the analysis task.

[0138] Optionally, if the first message further includes model quantity information, the model quantity information indicates the model number N. Then the at least one machine learning model determined by the first network element is the N machine learning models with the lowest energy consumption among all machine learning models used to implement the analysis task.

[0139] Similarly, if the first message also includes accuracy level requirement information and / or ML model filtering information, then at least one machine learning model determined by the first network element meets the accuracy level requirement information and / or ML model filtering information.

[0140] The above is just an example. The first network element can also use other methods to determine at least one machine learning model, and this application does not limit this.

[0141] In the present application, the model information may also include first energy consumption information, and the first energy consumption information indicates the energy consumption used to run at least one machine learning model. Specifically, the first energy consumption information may indicate the energy consumption used to run each machine learning model in at least one machine learning model. For example, the first energy consumption information includes the energy consumption or energy consumption range used to run each machine learning model in at least one machine learning model. Alternatively, the first energy consumption information includes the energy consumption level used to run each machine learning model in at least one machine learning model (for example, the energy consumption level can be identified as high, medium, low or level 1 (level1), level 2 (level2), level 3 (level3)), each energy consumption level corresponds to an energy consumption or energy consumption range; the correspondence between the energy consumption level and the energy consumption or energy consumption range is preset or preconfigured. Among them, the first energy consumption information may also be located outside the model information, and the model information is two independent pieces of information.

[0142] After obtaining the model information of the at least one machine learning model and the first energy consumption information, the second network element may select a machine learning model from the at least one machine learning model based on the first energy consumption information, and determine an analysis result of the analysis task based on the selected machine learning model. For example, the second network element may select a machine learning model with the lowest energy consumption among the at least one machine learning model, input the data for the analysis task into the machine learning model, run the machine learning model, and obtain an analysis result of the analysis task.

[0143] Optionally, before step 301, the first network element may also register with the fifth network element. When the second network element needs a machine learning model, it may request the address information of the first network element from the fifth network element. For details, please refer to the following process:

[0144] When the first network element is registering, the first network element may send a registration request message to the fifth network element.

[0145] The registration request message includes capability information, and the capability information indicates that the first network element has the ability to determine the energy consumption used to run the machine learning model.

[0146] The registration request message may also include an analysis identifier, indicating that the first network element has the ability to determine the energy consumption of the machine learning model used to run the analysis task corresponding to the analysis identifier. The number of analysis identifiers included in the registration request message is not limited, and may include one analysis identifier or multiple analysis identifiers.

[0147] This application does not limit the name of the registration request message. For example, the registration request message may also be called Nnrf_NF management NF registration request (Nnrf_NFmanagement_NFregister request) message or the like.

[0148] When the second network element needs a machine learning model, the second network element can send a discovery request message to the fifth network element, where the discovery request message is used to request a network element that has the ability to determine the energy consumption used to run the machine learning model.

[0149] The discovery request message may include an analysis identifier, which indicates a network element that is requested to have the ability to determine the energy consumption of the machine learning model used to run the analysis task corresponding to the analysis identifier.

[0150] Correspondingly, the fifth network element sends a discovery response message to the second network element.

[0151] The discovery response message includes address information of the first network element, and the address information of the first network element includes information such as an Internet Protocol (IP) address or a fully qualified domain name (FQDN) of the first network element.

[0152] Through the above process, after the second network element obtains the address information of the first network element, it can request the machine learning model from the first network element according to the address information of the first network element when the machine learning model is needed.

[0153] Through the above method, when the second network element requests a machine learning model from the first network element, it indicates the energy consumption indication information that the machine learning model must meet, so that the first network element determines at least one machine learning model based on the energy consumption indication information. In this way, when the second network element runs the machine learning model, it can consider the energy consumption of each machine learning model, avoiding high energy consumption during the operation of the machine learning model and achieving network energy saving.

[0154] In combination with the previous description, the following is an example in which the first network element is NWDAF (MTLF), the second network element and the third network element are the same network element, the second network element is NWDAF (AnLF), the fourth network element is OAM, the fifth network element is NRF, and the sixth network element is AF. The above network elements can also be replaced with other network elements, and this application is not limited. In this process, AF requests analysis results from NWDAF (AnLF), NWDAF (AnLF) requests a machine learning model from NWDAF (MTLF), runs the machine learning model, obtains analysis results, and sends the analysis results to AF.

[0155] As shown in FIG4 , it is a flow chart of a communication method provided in an embodiment of the present application.

[0156] Step 401: NWDAF (MTLF) sends a registration request message to NRF.

[0157] The registration request message includes capability information, which indicates that the NWDAF (MTLF) has the ability to determine the energy consumption used to run the machine learning model.

[0158] The registration request message may also include an analysis identifier, indicating that the NWDAF (MTLF) has the ability to determine the energy consumption of the machine learning model running for the analysis task corresponding to the analysis identifier. The registration request message is not limited to the number of analysis identifiers and may include one analysis identifier or multiple analysis identifiers.

[0159] This application does not limit the name of the registration request message. For example, the registration request message may also be called Nnrf_NF management NF registration request (Nnrf_NFmanagement_NFregister request) message or the like.

[0160] Step 402: The AF sends an analysis request message to the NWDAF (AnLF).

[0161] The analysis request message includes an analysis identifier, and is used to request an analysis result of an analysis task corresponding to the analysis identifier.

[0162] The analysis request message may also include energy consumption request information, which is used to request the determination of the energy consumption of the analysis result, or to indicate the expected energy consumed for the requested analysis result, such as an energy consumption range or energy consumption level, indicating that the AF expects to use the energy consumption information indicated in the energy consumption request to obtain the analysis result.

[0163] This application does not limit the name of the analysis request message. For example, the analysis subscription message can also be called an analysis subscription message, or an Nnwdaf analysis subscription (Nnwdaf_analyticsSubscription_Subscribe) message, etc. The analysis request message can also include other information, which will not be illustrated one by one here.

[0164] Optionally, step 403: the NWDAF (AnLF) sends a configuration request message to the OAM.

[0165] The configuration request message includes the second identifier of the NWDAF (AnLF), and the configuration request message is used to request configuration information of the NWDAF (AnLF).

[0166] Step 404: OAM sends a configuration response message of NWDAF (AnLF) to NWDAF (AnLF).

[0167] The configuration response message includes the configuration information of the NWDAF (AnLF).

[0168] If the NWDAF (AnLF) is able to determine its own configuration information, then step 403 and step 404 may not be performed.

[0169] Optionally, NWDAF (AnLF) determines that the analysis result needs to be generated through a machine learning model, and may request at least one machine learning model from NWDAF (MTLF), and then execute step 405.

[0170] Step 405: NWDAF (AnLF) sends a discovery request message to NRF, where the discovery request message is used to request a network element capable of determining the energy consumption for running the machine learning model.

[0171] The discovery request message may include an analysis identifier, which indicates a network element that is requested to have the ability to determine the energy consumption of a machine learning model that can run the analysis task corresponding to the analysis identifier.

[0172] Step 406: The NRF sends a discovery response message to the NWDAF (AnLF).

[0173] The discovery response message includes address information of at least one NWDAF (MTLF), for example, an Internet Protocol (IP) address or a fully qualified domain name (FQDN) of each NWDAF (MTLF) in the at least one NWDAF (MTLF).

[0174] The NWDAF (AnLF) selects one NWDAF (MTLF) from at least one NWDAF (MTLF) and executes step 407. The present application does not limit how the NWDAF (AnLF) selects the NWDAF (MTLF).

[0175] Step 407: NWDAF (AnLF) sends a model subscription message to NWDAF (MTLF).

[0176] Model subscription messages are used to request machine learning models.

[0177] The model subscription message includes an analysis identifier and energy consumption indication information. Optionally, the energy consumption indication information is determined based on the energy consumption request information. It can be understood that when the NWDAF (AnLF) obtains the energy consumption request information, the energy consumption indication information is carried in the model subscription message requesting the machine learning model.

[0178] Optionally, the model subscription message may further include at least one of the following:

[0179] Model quantity information, accuracy level requirements, ML model filtering information, NWDAF (AnLF) configuration information, or platform type information.

[0180] Optionally, step 408: if the model subscription message does not include the configuration information of NWDAF (AnLF), NWDAF (MTLF) sends a subscription (subscribe input) input message to OAM to subscribe to the configuration information of NWDAF (AnLF), and the subscription message includes the identification information of NWDAF (AnLF).

[0181] Step 409: OAM sends a subscribe output message to NWDAF (MTLF).

[0182] The subscription output message includes the configuration information of NWDAF (AnLF).

[0183] Optionally, the subscription output message also includes information such as an energy consumption information table.

[0184] Optionally, if the subscription output message does not include the energy consumption information table, the NWDAF (MTLF) may obtain the energy consumption information table from other network elements, which is not limited in this application.

[0185] Step 410: NWDAF (MTLF) determines at least one machine learning model based on the energy consumption indication information.

[0186] Among them, NWDAF (MTLF) obtains the energy consumption information table and the configuration information of NWDAF (AnLF). NWDAF (MTLF) can determine the energy consumption of each machine learning model among all machine learning models used to run the analysis task corresponding to the analysis identifier based on the energy consumption information table and the configuration information of NWDAF (AnLF). This application does not limit how NWDAF (MTLF) specifically determines the energy consumption used to run each machine learning model. Please refer to the description in step 302 and will not repeat it here.

[0187] If NWDAF (MTLF) obtains platform type information, it can also correct the energy consumption used to run each machine learning model. The specific process can be referred to the description in step 302 and will not be repeated here.

[0188] Furthermore, NWDAF (MTLF) determines at least one machine learning model based on the energy consumption indication information and the energy consumption used to run each machine learning model. The specific process can be referred to the description in step 302 and will not be repeated here.

[0189] Wherein, if the model subscription message includes accuracy level requirements and / or ML model filtering information, the at least one machine learning model meets the accuracy level requirements and / or ML model filtering information.

[0190] Step 411: NWDAF (MTLF) sends a model provision notification message to NWDAF (AnLF).

[0191] The model-provided notification message includes model information of at least one machine learning model and first energy consumption information, where the first energy consumption information indicates the energy consumption used to run each machine learning model in at least one machine learning model.

[0192] Step 412: NWDAF (AnLF) determines a first machine learning model from at least one machine learning model based on the first energy consumption information, and runs the first machine learning model to obtain an analysis result of the analysis task.

[0193] This application does not limit how NWDAF (AnLF) determines the first machine learning model from at least one machine learning model. For example, the machine learning model with the lowest energy consumption can be used as the first machine learning model.

[0194] This application does not limit how NWDAF (AnLF) obtains the analysis result based on the first machine learning model. For example, NWDAF (AnLF) can input the data corresponding to the analysis task into the first machine learning model to obtain the analysis result.

[0195] Step 413: NWDAF (AnLF) sends a subscription notification message to AF.

[0196] The subscription notification message includes the analysis result and second energy consumption information, where the second energy consumption information indicates the energy consumption for determining the analysis result, that is, the energy consumption for obtaining the analysis result by the NWDAF (AnLF).

[0197] In one implementation, the second energy consumption information is determined based on the first energy consumption consumed by running the first machine learning model to obtain the analysis result. In this case, the energy consumption indicated by the second energy consumption information may be the energy consumption actually consumed by the NWDAF (AnLF) when running the first machine learning model to obtain the analysis result. Optionally, in this implementation, the second energy consumption information is a specific value of the first energy consumption, or the second energy consumption information is an energy consumption level corresponding to the first energy consumption.

[0198] In another implementation, the second energy consumption information is determined based on the second energy consumption, and the second energy consumption is the energy consumption indicated by the first energy consumption information for running the first machine learning model. In this case, the energy consumption indicated by the second energy consumption information may be the energy consumption determined based on the first energy consumption information for running the first machine learning model to obtain the analysis result (i.e., NWDAF (MTLF) inference or estimated energy consumption). Optionally, in this implementation, the second energy consumption information is a specific value of the second energy consumption, or the second energy consumption information is an energy consumption level corresponding to the second energy consumption.

[0199] The second energy consumption information may include a specific energy consumption value or energy consumption range, or may include an energy consumption level, for example, the energy consumption level includes one of high, medium, and low, or the energy consumption level includes one of level 1, level 2, and level 3.

[0200] Among them, if NWDAF (AnLF) does not need to request NWDAF (MTLF) to analyze the machine learning model of the analysis task corresponding to the analysis identifier, steps 403-411 can be omitted. NWDAF (AnLF) can directly generate analysis results, and NWDAF (AnLF) collects energy consumption information when generating analysis results. For example, NWDAF (ANLF) collects energy consumption before running the analysis results, then runs the analysis, and then collects energy consumption after running the analysis results, and then calculates the difference between the two energy consumptions to obtain the energy consumption of NWDAF (AnLF) to generate the analysis result. NWDAF (AnLF) uses the energy consumption of generating the analysis result as the second energy consumption information, or converts the energy consumption into the corresponding energy consumption level as the second energy consumption information and sends it to AF.

[0201] If NWDAF (AnLF) needs to request the machine learning model of the analysis task corresponding to the analysis identifier from NWDAF (MTLF), but does not need to request the inference energy consumption information corresponding to the machine learning model. The network element discovery and model subscription message requesting the machine learning model in steps 403-411 may not include energy consumption indication information, and the model information provided in step 411 may also not include the first energy consumption information. NWDAF (AnLF) selects a machine learning model based on at least one machine learning model returned by NWDAF (MTLF), and generates an analysis result based on the machine learning model. When generating the analysis result, the energy consumption when NWDAF (AnLF) runs the machine learning model to generate the analysis result is collected. The specific energy consumption or the energy consumption level corresponding to the energy consumption is returned to AF as the second energy consumption information.

[0202] The above three methods of generating the second energy consumption information are all methods of NWDAF (AnLF) for generating real-time energy consumption of analysis results.

[0203] Alternatively, the NWDAF (ANLF) uses historical energy consumption information corresponding to the corresponding analysis identifiers to mathematically calculate the energy consumption baseline corresponding to each analysis identifier. When returning analysis results, the NWDAF (AnLF) includes the energy consumption baseline information corresponding to each analysis result. Specifically, the specific energy consumption value, energy consumption range, or energy consumption level corresponding to the energy consumption baseline is returned to the AF as the second energy consumption information.

[0204] From the above process, we can see that when determining the machine learning model, NWDAF (MTLF) not only considers information such as accuracy level requirements and ML model filtering information, but also considers the energy consumption used to run the machine learning model. This allows NWDAF (AnLF) to determine the machine learning model with the lowest energy consumption or a machine learning model with energy consumption less than or equal to the energy consumption threshold. This allows NWDAF (AnLF) to consume less energy when running the machine learning model, saving energy and avoiding energy waste.

[0205] Combined with the previous description, the following is an example in which the first network element is NWDAF (MTLF), the second network element is AMF (AMF can also be replaced by AF or other network elements), the third network element is UE (UE can also be replaced by other network elements capable of running machine learning models), the fourth network element is OAM, and the fifth network element is NRF. In this process, the UE requests the machine learning model from the NWDAF (MTLF) through the AMF and runs the obtained machine learning model.

[0206] As shown in FIG5 , it is a flow chart of a communication method provided in an embodiment of the present application.

[0207] Step 501: NWDAF (MTLF) sends a registration request message to NRF.

[0208] The registration request message includes capability information, which indicates that the NWDAF (MTLF) has the ability to determine the energy consumption used to run the machine learning model.

[0209] The registration request message may also include an analysis identifier, indicating that the NWDAF (MTLF) has the ability to determine the energy consumption of the machine learning model used to run the analysis task corresponding to the analysis identifier. The registration request message is not limited to the number of analysis identifiers included, and may include one analysis identifier or multiple analysis identifiers.

[0210] Step 502: The UE sends a request message to the AMF.

[0211] The subscription message includes an analysis identifier, and the subscription message is used to request the machine learning model corresponding to the analysis identifier.

[0212] The subscription message may also include UE configuration information, energy consumption indication information, UE identification and other information.

[0213] Step 503: AMF sends a discovery request message to NRF, where the discovery request message is used to request a network element capable of determining the energy consumption for running the machine learning model.

[0214] The discovery request message may include an analysis identifier, which indicates a network element that is requested to have the ability to determine the energy consumption of a machine learning model that can run the analysis task corresponding to the analysis identifier.

[0215] Step 504: The NRF sends a discovery response message to the AMF.

[0216] The discovery response message includes information of at least one NWDAF (MTLF), for example, an Internet Protocol (IP) address or a fully qualified domain name (FQDN) of each NWDAF (MTLF) in the at least one NWDAF (MTLF).

[0217] The AMF selects an NWDAF (MTLF) from at least one NWDAF (MTLF) and executes step 505. The present application does not limit how the AMF selects the NWDAF (MTLF).

[0218] Step 505: AMF sends a model subscription message to NWDAF (MTLF).

[0219] Model subscription messages are used to request machine learning models.

[0220] The model subscription message includes analysis identification and energy consumption indication information.

[0221] Optionally, the model subscription message may further include at least one of the following:

[0222] Model quantity information, accuracy level requirements, ML model filtering information, UE configuration information, or platform type information. The UE configuration information comes from the UE request message.

[0223] Optionally, if the NWDAF (MTLF) does not include the energy consumption information table, the energy consumption information table may be requested from the OAM.

[0224] Step 506: NWDAF (MTLF) sends a subscription input message to OAM to subscribe to the energy consumption information table.

[0225] Step 507: OAM sends a subscription output message to NWDAF (MTLF).

[0226] Subscription output messages include energy consumption information table and other information.

[0227] Among them, step 506 and step 507 are optional steps. If NWDAF (MTLF) includes a model energy consumption prediction model, NWDAF (MTLF) can determine the energy consumption used to run the machine learning model based on the model energy consumption prediction model without obtaining the energy consumption information table from OAM.

[0228] Step 508: NWDAF (MTLF) determines at least one machine learning model based on the energy consumption indication information.

[0229] Among them, this application does not limit how NWDAF (MTLF) specifically determines the energy consumption for running each machine learning model. Please refer to the description in step 302 and will not repeat it here.

[0230] Step 509: NWDAF (MTLF) sends a model provision notification message to AMF.

[0231] The model providing notification message includes model information of at least one machine learning model and first energy consumption information, where the first energy consumption information indicates the energy consumption used to run each of the at least one machine learning model.

[0232] Step 510: The AMF sends a response message to the UE.

[0233] The response message includes model information of at least one machine learning model and first energy consumption information.

[0234] Step 511: The UE determines a first machine learning model from at least one machine learning model based on the first energy consumption information, and runs the first machine learning model to obtain an analysis result of the analysis task.

[0235] This application does not limit how the UE determines the first machine learning model from at least one machine learning model. For example, the machine learning model with the lowest energy consumption can be used as the first machine learning model.

[0236] This application does not limit how the UE obtains the analysis result based on the first machine learning model. For example, the UE can input the data corresponding to the analysis task into the first machine learning model to obtain the analysis result.

[0237] It is understood that, in order to implement the functions in the above embodiments, the first network element or the second network element includes hardware structures and / or software modules corresponding to the respective functions. Those skilled in the art should readily appreciate that, in combination with the units and method steps of the various examples described in the embodiments disclosed in this application, this application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or in a hardware-driven manner by computer software depends on the specific application scenario and design constraints of the technical solution.

[0238] The following is a schematic diagram of the structure of possible communication devices provided in the embodiments of the present application. These communication devices can be used to implement the functions of the first network element or the second network element in the above method embodiments, and thus can also achieve the beneficial effects of the above method embodiments.

[0239] As shown in Figure 6, a communication device 600 includes a processing unit 610 and a communication unit 620. The communication device 600 is used to implement the functions of the first network element or the second network element in each of the above-mentioned method embodiments.

[0240] When the communication device 600 is used to implement the function of the first network element:

[0241] The processing unit is configured to receive a first message from a second network element through the communication unit; the first message is used to request a machine learning model; and the first message includes energy consumption indication information;

[0242] The processing unit is used to send model information of at least one machine learning model to the second network element through the communication unit, and the at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model.

[0243] When the communication device 600 is used to implement the function of the second network element:

[0244] The processing unit is configured to send a first message to the first network element through the communication unit; the first message is used to request a machine learning model; and the first message includes energy consumption indication information;

[0245] The processing unit is used to receive model information of at least one machine learning model from the first network element through the communication unit, and the at least one machine learning model is determined based on the energy consumption indication information and the energy consumption used to run the at least one machine learning model.

[0246] A more detailed description of the processing unit 610 and the communication unit 620 can be directly obtained by referring to the relevant descriptions in the above-mentioned method embodiments, and will not be repeated here.

[0247] It should be understood that the division of units in the above device is merely a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or physically separated. Moreover, the units in the device can all be implemented in the form of software called through processing elements; or all be implemented in the form of hardware; or some units can be implemented in the form of software called through processing elements, and some units can be implemented in the form of hardware. For example, each unit can be a separately established processing element, or it can be integrated into a certain chip of the device. In addition, it can also be stored in the form of a program in a memory, called by a certain processing element of the device and execute the function of the unit. In addition, all or part of these units can be integrated together, or they can be implemented independently. The processing element here can also be a processor, which can be an integrated circuit with signal processing capabilities. In the implementation process, each operation of the above method or each unit above can be implemented by the integrated logic circuit of the hardware in the processor element or by software called through the processing element.

[0248] In one example, the unit in any of the above devices may be one or more integrated circuits configured to implement the above method, such as one or more application specific integrated circuits (ASICs), one or more digital singnal processors (DSPs), one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms. For another example, when the unit in the device can be implemented in the form of a processing element scheduler, the processing element can be a processor, such as a general-purpose central processing unit (CPU), or other processor that can call a program. For another example, these units can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0249] The above-mentioned receiving unit is an interface circuit of the device, which is used to receive signals from other devices. For example, when the device is implemented as a chip, the receiving unit is the interface circuit of the chip used to receive signals from other chips or devices. The above-mentioned sending unit is an interface circuit of the device, which is used to send signals to other devices. For example, when the device is implemented as a chip, the sending unit is the interface circuit of the chip used to send signals to other chips or devices.

[0250] As another possible product form, the first network element or the second network element of the embodiment of the present application can be implemented by a general bus architecture. For ease of explanation, refer to Figure 7, which is a structural diagram of a communication device 700 provided in an embodiment of the present application, and the communication device 700 includes a processor 701 and a transceiver 702. The communication device 700 can be a terminal device, or a chip or chip system therein; or, the communication device 700 can be a network device, or a chip or module therein. Figure 7 only shows the main components of the communication device 700. In addition to the processor 701 and the transceiver 702, the communication device 700 can further include a memory 703, and an input and output device (not shown in the figure).

[0251] Optionally, processor 701 is primarily used to process communication protocols and communication data, as well as control the entire communication device, execute software programs, and process software program data. Memory 703 is primarily used to store software programs and data. Transceiver 702 may include a radio frequency circuit and an antenna. The radio frequency circuit is primarily used to convert baseband signals into radio frequency signals and process radio frequency signals. The antenna is primarily used to transmit and receive radio frequency signals in the form of electromagnetic waves. Input and output devices, such as a touch screen, display, and keyboard, are primarily used to receive user input and output data to the user.

[0252] Optionally, the processor 701 , the transceiver 702 , and the memory 703 may be connected via a communication bus.

[0253] When the communication device is powered on, the processor 701 can read the software program in the memory 703, interpret and execute the instructions of the software program, and process the data of the software program. When data needs to be sent wirelessly, the processor 701 performs baseband processing on the data to be sent and outputs the baseband signal to the radio frequency circuit. The radio frequency circuit performs radio frequency processing on the baseband signal and then transmits the radio frequency signal to the outside in the form of electromagnetic waves through the antenna. When data is sent to the communication device, the radio frequency circuit receives the radio frequency signal through the antenna, converts the radio frequency signal into a baseband signal, and outputs the baseband signal to the processor 701. The processor 701 converts the baseband signal into data and processes the data.

[0254] In another implementation, the RF circuit and antenna can be set independently of the processor performing baseband processing. For example, in a distributed scenario, the RF circuit and antenna can be arranged remotely from the communication device.

[0255] In some embodiments, in terms of hardware implementation, those skilled in the art may conceive that the above-mentioned communication device 600 may take the form of the communication device 700 shown in FIG. 7 .

[0256] As an example, the functions / implementation process of the processing unit 610 in FIG6 can be implemented by the processor 701 in the communication device 700 shown in FIG7 calling the computer-executable instructions stored in the memory 703. The functions / implementation process of the communication unit 620 in FIG6 can be implemented by the transceiver 702 in the communication device 700 shown in FIG7.

[0257] As another possible product form, the first network element or the second network element in the present application may adopt the structure shown in Figure 8, or include the components shown in Figure 8. Figure 8 is a schematic diagram of the structure of a communication device 800 provided in the present application.

[0258] As shown in FIG8 , a communication device 800 includes at least one processor 801. Optionally, the communication device further includes a communication interface 802.

[0259] When the program instructions are executed in the at least one processor 801, the communication device 800 can implement the method provided in any of the aforementioned embodiments and any possible designs thereof. Alternatively, the processor 801 implements the method provided in any of the aforementioned embodiments and any possible designs thereof through logic circuits or by executing code instructions.

[0260] The communication interface 802 can be used to receive program instructions and transmit them to the processor. Alternatively, the communication interface 802 can be used for the communication device 800 to communicate with other communication devices, such as exchanging control signaling and / or service data. Exemplarily, the communication interface 802 can be used to receive signals from devices other than the communication device 800 and transmit them to the processor 801, or to send signals from the processor 801 to other communication devices other than the communication device 800.

[0261] Optionally, the communication interface 802 may be a code and / or data read and write interface circuit, or the communication interface 802 may be a signal transmission interface circuit between a communication processor and a transceiver, or a pin of a chip.

[0262] Optionally, the communication device 800 may further include at least one memory 803, which may be used to store required program instructions and / or data. It should be noted that the memory 803 may exist independently of the processor 801 or may be integrated with the processor 801. The memory 803 may be located within the communication device 800 or outside the communication device 800, without limitation.

[0263] Optionally, the communication device 800 may further include a power supply circuit 804, which may be used to supply power to the processor 801. The power supply circuit 804 may be located in the same chip as the processor 801, or in another chip other than the chip where the processor 801 is located.

[0264] Optionally, the communication device 800 may further include a bus, and various parts of the communication device 800 may be interconnected via the bus.

[0265] In some embodiments, in terms of hardware implementation, those skilled in the art may conceive that the communication device 600 shown in FIG. 6 may take the form of the communication device 800 shown in FIG. 8 .

[0266] As an example, the functions / implementation process of the processing unit 610 in FIG6 can be implemented by the processor 801 in the communication device 800 shown in FIG8 calling the computer-executable instructions stored in the memory 803. The functions / implementation process of the communication unit 620 in FIG6 can be implemented by the communication interface 802 in the communication device 800 shown in FIG8.

[0267] It should be noted that the structure shown in FIG8 does not constitute a specific limitation on the first network element or the second network element. For example, in other embodiments of the present application, the first network element or the second network element may include more or fewer components than shown in the figure, or some components may be combined or separated, or the components may be arranged differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0268] It is understood that the processor in the embodiments of the present application may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0269] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, a register, a hard disk, a mobile hard disk, a CD-ROM or any other form of storage medium well known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an ASIC. In addition, the ASIC can be located in a base station or a terminal. Of course, the processor and the storage medium can also exist in a base station or a terminal as discrete components.

[0270] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are performed in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, or other programmable device. The computer program or instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions may be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video disk; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or nonvolatile storage medium, or may include both volatile and nonvolatile types of storage media.

[0271] In the various embodiments of the present application, unless otherwise specified or there is a logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0272] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, optical storage, etc.) that contain computer-usable program code.

[0273] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each flow and / or box in the flow chart and / or block diagram, as well as the combination of the flow chart and / or box in the flow chart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one or more flow charts and / or one or more boxes in the block diagram.

[0274] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device that implements the functions specified in one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0275] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include these modifications and variations.

Claims

1. A communication method, characterized in that, including: A first network element receives a first message from a second network element; The first message is used to request a machine learning model; The first message includes energy consumption indication information; The first network element sends model information of at least one machine learning model to the second network element, and the at least one machine learning model is determined according to the energy consumption indication information and the energy consumption for running the at least one machine learning model.

2. The method according to claim 1, wherein The method further includes: The first network element determines the at least one machine learning model according to the energy consumption indication information and the energy consumption for running each machine learning model in the at least one machine learning model.

3. The method according to claim 1 or 2, characterized in that, The energy consumption indication information indicates that the second network element is in an energy-saving state; The first message further includes an analysis identifier, and the analysis identifier is used to identify an analysis task; Among the machine learning models for implementing the analysis task, the energy consumption for running the at least one machine learning model is the smallest.

4. The method according to claim 1 or 2, characterized in that The energy consumption indication information includes an energy consumption threshold; The energy consumption for running each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: The first network element determines the energy consumption for running the at least one machine learning model according to the configuration information of a third network element; wherein, the third network element is used to run the machine learning model in the at least one machine learning model.

6. The method according to claim 5, characterized in that, The first message further includes the configuration information of the third network element.

7. The method according to claim 5, wherein The first message further includes an identifier of the third network element; the method further includes: The first network element sends the identifier of the third network element to a fourth network element; The first network element receives the configuration information of the third network element from the fourth network element.

8. The method according to any one of claims 1 to 7, characterized in that The model information contains first energy consumption information, and the first energy consumption information indicates the energy consumption for running the at least one machine learning model.

9. The method according to any one of claims 1 to 8, characterized in that, The method further includes: The first network element sends a registration request message to a fifth network element, and the registration request message includes capability information, and the capability information indicates that the first network element has the capability to determine the energy consumption for running a machine learning model.

10. A communication method, characterized in that, including: A second network element sends a first message to a first network element; The first message is used to request a machine learning model; The first message includes energy consumption indication information; The second network element receives model information of at least one machine learning model from the first network element, and the at least one machine learning model is determined according to the energy consumption indication information and the energy consumption for running the at least one machine learning model.

11. The method according to claim 10, wherein Before sending the first message, the method further includes: The second network element receives an analysis request message from a sixth network element, and the analysis request message contains energy consumption request information, and the energy consumption request information is used to indicate the energy consumption for requesting to determine an analysis result.

12. The method according to claim 11, wherein The energy consumption indication information is determined according to the energy consumption request information.

13. The method according to any one of claims 10 to 12, characterized in that, The model information contains first energy consumption information, and the first energy consumption information indicates the energy consumption for running the at least one machine learning model.

14. The method according to claim 13, wherein The method further includes: The second network element determines a first machine learning model from the at least one machine learning model according to the first energy consumption information; The second network element determines the analysis result according to the first machine learning model; The second network element sends the analysis result and second energy consumption information to a sixth network element, where the second energy consumption information indicates the energy consumption for determining the analysis result.

15. The method according to claim 14, wherein The second energy consumption information is determined according to the first energy consumption consumed to obtain the analysis result by running the first machine learning model; Alternatively, the second energy consumption information is determined according to a second energy consumption, where the second energy consumption is the energy consumption indicated by the first energy consumption information for running the first machine learning model.

16. The method according to any one of claims 10 to 15, characterized in that, The energy consumption indication information indicates that the second network element is in an energy-saving state; The first message further includes an analysis identifier, where the analysis identifier is used to identify an analysis task; Among the machine learning models for implementing the analysis task, the energy consumption for running the at least one machine learning model is the smallest.

17. The method according to any one of claims 10 to 15, characterized in that, The energy consumption indication information includes an energy consumption threshold; The energy consumption for running each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

18. The method according to any one of claims 10 to 17, characterized in that The first message further includes at least one of the following: Configuration information of a third network element, or an identifier of the third network element; the third network element is used to run the machine learning model in the at least one machine learning model.

19. The method according to any one of claims 10 to 18, characterized in that, The method further includes: The second network element sends a second message to a fifth network element, where the second message is used to request a network element with the ability to determine the energy consumption for running a machine learning model; The second network element receives the address information of the first network element from the fifth network element; The second network element sends the first message to the first network element according to the address information of the first network element.

20. A communication method, characterized in that, Includes: The second network element sends a first message to the first network element; The first message is used to request a machine learning model; The first message includes energy consumption indication information; The first network element receives the first message from the second network element; The first network element sends model information of at least one machine learning model to the second network element, where the at least one machine learning model is determined according to the energy consumption indication information and the energy consumption for running the at least one machine learning model; The second network element receives the model information of at least one machine learning model from the first network element.

21. The method according to claim 20, wherein The energy consumption indication information indicates that the second network element is in an energy-saving state; The first message further includes an analysis identifier, where the analysis identifier is used to identify an analysis task; Among the machine learning models for implementing the analysis task, the energy consumption for running the at least one machine learning model is the smallest.

22. The method according to claim 20, characterized in that, The energy consumption indication information includes an energy consumption threshold; The energy consumption for running each machine learning model in the at least one machine learning model is less than or equal to the energy consumption threshold.

23. The method according to any one of claims 20 to 22, characterized in that The method further includes: The first network element determines the energy consumption for running the at least one machine learning model according to the configuration information of a third network element; Wherein, the third network element is used to run the machine learning model in the at least one machine learning model.

24. The method according to claim 23, wherein The first message further includes the configuration information of the third network element.

25. The method according to any one of claims 20 to 24, characterized in that, The model information includes first energy consumption information, and the first energy consumption information indicates the energy consumption for running the at least one machine learning model.

26. A communication system, characterized in that, Comprising: A first network element and a second network element; The first network element is configured to execute the method according to any one of claims 1 to 9, and the second network element is configured to execute the method according to any one of claims 10 to 25.

27. A communication device, characterized in that, Comprising: A processing unit and a communication unit; The processing unit is configured to receive a first message from the second network element through the communication unit; The first message is used to request a machine learning model; the first message includes energy consumption indication information; The processing unit is configured to send model information of at least one machine learning model to the second network element through the communication unit, and the at least one machine learning model is determined according to the energy consumption indication information and the energy consumption for running the at least one machine learning model.

28. A communication device, characterized in that, Comprising: A processing unit and a communication unit; The processing unit is configured to send a first message to the first network element through the communication unit; The first message is used to request a machine learning model; the first message includes energy consumption indication information; The processing unit is configured to receive model information of at least one machine learning model from the first network element through the communication unit, and the at least one machine learning model is determined according to the energy consumption indication information and the energy consumption for running the at least one machine learning model.

29. A communication device, characterized in that, Comprising a processor; The processor is configured to execute a computer program or instruction stored in a memory, so that the communication device implements the method according to any one of claims 1 to 25.

30. A computer-readable storage medium, characterized in that, Stored with a computer program or instruction, when the computer program or instruction runs on a computer, the computer implements the method according to any one of claims 1 to 25.

31. A computer program product storing instructions, characterized in that, When a computer reads and executes the computer program product, the computer implements the method according to any one of claims 1 to 25.

Images