WO2024012326A1 - 一种通信方法、装置及系统 - Google Patents

一种通信方法、装置及系统 Download PDF

Info

Publication number
WO2024012326A1
WO2024012326A1 PCT/CN2023/105896 CN2023105896W WO2024012326A1 WO 2024012326 A1 WO2024012326 A1 WO 2024012326A1 CN 2023105896 W CN2023105896 W CN 2023105896W WO 2024012326 A1 WO2024012326 A1 WO 2024012326A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
terminal device
data
distribution information
server
Prior art date
Application number
PCT/CN2023/105896
Other languages
English (en)
French (fr)
Inventor
王四海
秦城
杨锐
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from CN202210977688.3A external-priority patent/CN117459961A/zh
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2024012326A1 publication Critical patent/WO2024012326A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition

Definitions

  • the present application relates to the field of communication technology, and in particular, to a communication method, device and system.
  • the two devices are a network device and a terminal device respectively.
  • the terminal device can use an artificial intelligence (artificial intelligence, AI) model (such as an AI encoder) to encode the information to be transmitted (such as the above-mentioned measurement information, status information), and
  • AI artificial intelligence
  • the encoded information is fed back to the network device, and the network device obtains the information reported by the terminal device (such as the above-mentioned measurement information and status information) through decoding by the AI decoder corresponding to the AI encoder.
  • the AI model used by the terminal device in a specific area is generated by training by the server associated with the terminal device.
  • the training data of the AI model includes data collected by multiple terminal devices associated with the server in the specific area.
  • the specific area includes not only multiple terminal devices associated with the server, but also other terminal devices associated with other servers; that is to say, the training data of the AI model is data collected from some terminal devices in the specific area. Therefore, the feature distribution of the training data of the AI model may be different from the feature distribution of the total training data in a specific area, which will affect the performance of the AI model. For example, the information to be transmitted may be distorted after AI encoding and AI decoding.
  • This application provides a communication method, device and system to solve the problem that the characteristic distribution of the training data of the AI model is different from the characteristic distribution of the total training data in a specific area, thus affecting the performance of the AI model.
  • inventions of the present application provide a communication system.
  • the communication system includes a network device, a first terminal device and a first server, and the first terminal device is associated with the first server.
  • the network device is used to send first characteristic distribution information to the first terminal device, and the first characteristic distribution information is used to indicate the characteristic distribution of data collected by the terminal device located in the first area in the first time period;
  • the first terminal The device is also configured to receive the first feature distribution information and send the first feature distribution information to the first server;
  • the first server is configured to receive the first feature distribution information and generate artificial intelligence according to the first feature distribution information and training data.
  • an intelligent AI model and, sending the configuration information of the AI model to the first terminal device or the second terminal device associated with the first server.
  • the network device sends the first feature distribution information to the first terminal device, and the first terminal device sends the first feature distribution information to the first server associated with the first terminal device.
  • the first training data used by the first server to generate the AI model is data collected by the terminal device associated with the first server in the first area
  • the feature distribution of the first training data may be different from the total training in the first area.
  • the first server can obtain the first characteristic distribution information (used to indicate the characteristic distribution of the total training data in the first area), the first server can use the first characteristic distribution information as the generated AI
  • the auxiliary information of the model makes the AI model generated based on the first training data close to the AI model generated based on the total training data in the first area, which can effectively improve the performance of the AI model.
  • the first terminal device is configured to send first information to the network device.
  • the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data, P1 group data is the data collected by the first terminal device located in the first area and within the first time period, and P1 is a positive integer; the terminal device includes the first terminal device; the network device is also used to determine the first feature distribution information based on the first information .
  • the network device is further configured to send first request information.
  • the first request information is used to request the terminal device to collect data within a first time period and report characteristic information and/or features of the collected data. Distributing information; before sending the first information to the network device, the first terminal device is also used to receive the first request information.
  • the first request information includes at least one of the following: first indication information, the first indication information indicates the request terminal
  • the information reported by the terminal device is the characteristic information and/or characteristic distribution information of the data collected by the terminal device
  • the second indication information indicates the first mode and/or the second mode, and the first mode is used for Determine the characteristic information of the data collected by the terminal device.
  • the second method is used to determine the characteristic distribution information of the data collected by the terminal device; terminal configuration information.
  • the terminal configuration information indicates that the first terminal device should collect the P1 group of data.
  • trigger information includes trigger mode and/or trigger configuration
  • the trigger mode includes at least one of the following: event triggering, cycle trigger
  • the trigger configuration includes at least one of the following: trigger event, trigger cycle
  • quantity information The quantity information is used to indicate the value of P1
  • the area information is used for the first area
  • time period information is used to indicate the first time period.
  • the first terminal device is further configured to send second request information to the network device, and the second request information is used to request to send the first feature distribution information to the first terminal device; the network device sends the first feature distribution information to the first terminal device. Before sending the first feature distribution information, the terminal device is also used to receive the second request information.
  • the first server is further configured to send third request information to the first terminal device, and the third request information is used to request to send the first feature distribution information to the first server; Before sending the first feature distribution information, the first server is also configured to receive the third request information.
  • the first server before generating the AI model based on the first feature distribution information and the training data, the first server is also used to determine that the difference between the first feature distribution information and the second feature distribution information satisfies the first condition.
  • the second feature distribution information is obtained in advance by the first server.
  • the first server can determine whether the AI model needs to be updated based on the first feature distribution information and the second feature distribution information, so that it can more easily and quickly identify whether the AI model needs to be updated, so as to update the AI model more reasonably. , to avoid frequent updates to the AI model that result in a heavy processing burden on the first server, or failure to update the AI model for a long time, resulting in poor performance of the AI model.
  • the first server is also configured to receive the P1 group data from the first terminal device.
  • the training data includes a P1 group of data.
  • the P1 group of data is data collected by the first terminal device located in the first area and during the first time period. P1 is a positive integer.
  • the first server is also used to receive the P2 group data from the second terminal device.
  • the training data includes a P2 group of data.
  • the P2 group of data is data collected by the second terminal device located in the first area and during the first time period. P2 is a positive integer.
  • the training data includes data obtained in advance by the first server.
  • embodiments of the present application provide a communication method.
  • This method can be applied to the first terminal device.
  • the first terminal device receives first feature distribution information from the network device, and the first feature distribution information is used to indicate the feature distribution of data collected by the terminal device located in the first area in the first time period; to the first A server sends first feature distribution information, and the first server is associated with the first terminal device.
  • first information is sent to the network device, and the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data, and the P1 group data is that the first terminal device is located at the For data collected in an area and within the first time period, P1 is a positive integer; wherein the terminal device includes the first terminal device, and the first feature distribution information is determined based on the first information.
  • the method before sending the first information to the network device, the method further includes: receiving first request information from the network device, where the first request information is used to request the terminal device to collect data within a first time period. , and report the characteristic information and/or characteristic distribution information of the collected data.
  • the first request information includes at least one of the following: first indication information, the first indication information indicates that the information requested to be reported by the terminal device is the characteristic information and/or the characteristic distribution information of the data collected by the terminal device; Second indication information, the second indication information indicates a first way and/or a second way, the first way is used to determine the characteristic information of the data collected by the terminal device, and the second way is used to determine the characteristic distribution information of the data collected by the terminal device.
  • Terminal configuration information indicates the configuration conditions that should be met when the first terminal device collects P1 group data
  • Trigger information includes the trigger method and/or trigger configuration, the trigger method includes at least one of the following: event trigger, cycle Trigger, the trigger configuration includes at least one of the following: trigger event, trigger cycle; quantity information, quantity information is used to indicate the value of P1; area information, area information is used to indicate the first area; time period information, time period information is used to indicate first time period.
  • the method before receiving the first feature distribution information from the network device, the method further includes: sending second request information to the network device, where the second request information is used to request to send the first feature distribution information to the first terminal device.
  • Feature distribution information is used to request to send the first feature distribution information to the first terminal device.
  • the method before sending the second request information to the network device, the method further includes: receiving third request information from the first server, where the third request information is used to request to send the first feature to the first server. Distribution information.
  • embodiments of the present application provide a communication method.
  • This method can be applied to network devices.
  • the network device Determine first feature distribution information and send the first feature distribution information to the first terminal device, where the first feature distribution information is used to indicate the feature distribution of data collected by the terminal device located in the first area within the first time period.
  • the method further includes: receiving first information from the first terminal device, where the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data, The data is data collected by the first terminal device located in the first area and within the first time period, and P1 is a positive integer; the terminal device includes the first terminal device; determining the first feature distribution information includes: determining the first characteristic distribution information based on the first information. 1. Feature distribution information.
  • the method before receiving the first information from the first terminal device, the method further includes: sending first request information to the terminal device, where the first request information is used to request the terminal device to send information within a first time period. Collect data and report the characteristic information and/or characteristic distribution information of the collected data.
  • the first request information includes at least one of the following: first indication information, the first indication information indicates that the information requested to be reported by the terminal device is the characteristic information and/or the characteristic distribution information of the data collected by the terminal device; Second indication information, the second indication information indicates a first way and/or a second way, the first way is used to determine the characteristic information of the data collected by the terminal device, and the second way is used to determine the characteristic distribution information of the data collected by the terminal device.
  • Terminal configuration information indicates the configuration conditions that should be met when the first terminal device collects P1 group data
  • Trigger information includes the trigger method and/or trigger configuration, the trigger method includes at least one of the following: event trigger, cycle Trigger, the trigger configuration includes at least one of the following: trigger event, trigger cycle; quantity information, quantity information is used to indicate the value of P1; area information, area information is used to indicate the first area; time period information, time period information is used to indicate first time period.
  • the method before sending the first feature distribution information to the first terminal device, the method further includes: receiving second request information from the first terminal device, the second request information being used to request to send the first feature distribution information to the first terminal device.
  • the device sends first feature distribution information.
  • embodiments of the present application provide a communication method.
  • This method can be applied to the first server.
  • the first server receives first feature distribution information from the first terminal device.
  • the first feature distribution information is used to indicate the feature distribution of data collected by the terminal device located in the first area in the first time period.
  • a server is associated with the first terminal device; generates an AI model according to the first feature distribution information; and sends configuration information of the AI model to the first terminal device or a second terminal device associated with the first server.
  • the method further includes: receiving a P1 group of data from the first terminal device.
  • the P1 group of data is data collected by the first terminal device in the first area and within the first time period.
  • P1 is positive. Integer; among them, the training data includes P1 group data.
  • the method further includes: receiving a P2 set of data from the second terminal device.
  • the P2 set of data is data collected by the second terminal device in the first area and within the first time period.
  • P2 is positive. Integer; among them, the training data includes P2 group data.
  • the training data includes data obtained in advance by the first server.
  • the method before generating the AI model based on the first feature distribution information, the method further includes: determining that the difference between the first feature distribution information and the second feature distribution information satisfies the first condition, and the second feature distribution information Prefetched for the first server.
  • the method before receiving the first feature distribution information from the first terminal device, the method further includes: sending third request information to the first terminal device, the third request information being used to request to the first server Send first feature distribution information.
  • inventions of the present application provide a communication system.
  • the communication system includes network equipment and a first terminal equipment.
  • the first terminal device is used to send first information to the network device.
  • the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data, and the P1 group data is that the first terminal device is located at the For data collected in an area and within the first time period, P1 is a positive integer.
  • the network device is configured to receive the first information.
  • the terminal device can report the characteristic information and/or characteristic distribution information of the P1 group data to the network device, so that the network device can promptly learn the distribution of the total training data in the first area, which is convenient for use on the terminal device side.
  • AI models are managed.
  • the network device is also configured to send deactivation information to the first terminal device, where the deactivation information is used to indicate deactivation of the AI model or AI mode.
  • the terminal device is also used to receive deactivation information.
  • the network device is further configured to send switching information to the first terminal device, where the switching information is used to instruct the used AI model to be switched to the target AI model.
  • the terminal device is also used to receive switching information.
  • the network device is also used to generate an AI model based on the training data and send the configuration information of the AI model to the first terminal device.
  • the terminal device is also used to receive the configuration information of the AI model and update the used model to the AI model.
  • the network device is further configured to determine that the difference between the first feature distribution information and the second feature distribution information satisfies the third requirement.
  • One condition is that the first characteristic distribution information is used to indicate the characteristic distribution of data collected by the terminal device located in the first area and within the first time period, and the first characteristic distribution information is determined based on the first information.
  • the terminal device includes a first terminal device, and the second feature distribution information is obtained in advance by the network device.
  • the network device can determine the feature distribution of the currently used AI model based on the first feature distribution information and the second feature distribution information. performance, making it easier and faster to identify whether the performance of the AI model is poor.
  • the network device is further configured to send first request information to the terminal device.
  • the first request information is used to request the terminal device to collect data within a first time period and report the characteristic information and characteristics of the collected data. /or feature distribution information.
  • the first terminal device is also configured to receive the first request information.
  • the first request information includes at least one of the following: first indication information, the first indication information indicates that the information requested to be reported by the terminal device is the characteristic information and/or the characteristic distribution information of the data collected by the terminal device; Second indication information, the second indication information indicates a first way and/or a second way, the first way is used to determine the characteristic information of the data collected by the terminal device, and the second way is used to determine the characteristic distribution information of the data collected by the terminal device. ; Terminal configuration information, the terminal configuration information indicates the configuration conditions that should be met when the first terminal device collects the P1 group data; Trigger information, the trigger information includes the trigger method and/or trigger configuration.
  • the triggering method includes at least one of the following: event triggering, periodic triggering; the triggering configuration includes at least one of the following: triggering event, triggering period; quantity information, the quantity information is used to indicate the value of P1; area information, the area information is used to indicate the first Area; time period information, the time period information is used to indicate the first time period.
  • the first terminal device is also used to send the P1 group of data to the network device; the network device is also used to receive the P1 group of data, and the training data includes the P1 group of data.
  • the network device is further configured to send fifth request information to the first terminal device, where the fifth request information is used to request the first terminal device to report data located in the first area and collected within the first time period. data; before sending the P1 group of data to the network device, the first terminal device is also used to receive the fifth request information.
  • the network device determines that it needs to generate an AI model, it sends the fifth request information to the first terminal device, and then the first terminal device reports the P1 set of data to the network device based on the fifth request. That is to say, when the network device does not need to generate an AI model, the first terminal device does not need to send the P1 group of data to the network device, thereby effectively saving transmission resources.
  • embodiments of the present application provide a communication method.
  • This method can be applied to network devices.
  • the network device receives first information from the first terminal device.
  • the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data, and the P1 group data is the first terminal device.
  • P1 is a positive integer.
  • the method further includes: sending deactivation information to the first terminal device, where the deactivation information is used to indicate deactivation of the AI model or AI mode.
  • the method further includes: sending switching information to the first terminal device, where the switching information is used to indicate switching the used AI model to the target AI model.
  • the switching information includes at least one of the following: identification information of the target AI model and configuration information of the target AI model.
  • the method further includes: generating an AI model based on the training data, and sending configuration information of the AI model to the first terminal device; the configuration information of the AI model is used to instruct the used model to be updated to the AI model.
  • the method further includes: determining that the difference between the first characteristic distribution information and the second characteristic distribution information satisfies the first condition, the first characteristic distribution information being used to indicate the terminal device located in the first area and in Characteristic distribution of data collected in the first time period, the first characteristic distribution information is determined based on the first information, the terminal device includes the first terminal device, and the second characteristic distribution information is the network pre-acquired by the device.
  • the method further includes: sending first request information to the terminal device, the first request information being used to request the terminal device to collect data within the first time period, and Report the characteristic information and/or characteristic distribution information of the collected data.
  • the first request information includes at least one of the following: first indication information indicating that the information requested to be reported by the terminal device is a characteristic of the data collected by the terminal device. information and/or feature distribution information; second indication information, the second indication information indicates the first way and/or the second way, the first way is used to determine the feature information of the data collected by the terminal device, so The second method is used to determine the characteristic distribution information of the data collected by the terminal device; terminal configuration information, the terminal configuration information indicates the configuration conditions that should be met when the first terminal device collects the P1 group of data; trigger information, the trigger information includes a trigger method And/or triggering configuration, the triggering method includes at least one of the following: event triggering, periodic triggering; the triggering configuration includes at least one of the following: triggering Event, trigger cycle; quantity information, quantity information is used to indicate the value of P1; area information, area information is used to indicate the first area; time period information, time period information is used to indicate the first time period.
  • the method before sending the configuration information of the AI model to the first terminal device, the method further includes: receiving the P1 set of data from the first terminal device, and the training data includes the P1 set of data.
  • the method further includes: sending fifth request information to the first terminal device, where the fifth request information is used to request the first terminal device to report data located in the first area and collected within the first time period. .
  • embodiments of the present application provide a communication method.
  • This method can be applied to the first terminal device.
  • the first terminal device sends first information to the network device.
  • the first information includes at least one of the following: characteristic information of group P1 data, characteristic distribution information of group P1 data, group P1 data is data collected by the first terminal device located in the first area and within the first time period, and P1 is positive integer.
  • the method further includes: receiving deactivation information from the network device, where the deactivation information is used to indicate deactivation of the AI model or AI mode.
  • the method further includes: receiving switching information from the network device, where the switching information is used to indicate switching the used AI model to the target AI model.
  • the switching information includes at least one of the following: identification information of the target AI model and configuration information of the target AI model.
  • the method further includes: receiving configuration information of the AI model from the network device; updating the used model to the AI model.
  • the method before sending the first information to the network device, the method further includes: receiving first request information from the network device, where the first request information is used to request the terminal device to collect data within a first time period. , and report the characteristic information and/or characteristic distribution information of the collected data.
  • the first request information includes at least one of the following: first indication information, the first indication information indicates that the information requested to be reported by the terminal device is the characteristic information and/or the characteristic distribution information of the data collected by the terminal device; Second indication information, the second indication information indicates a first way and/or a second way, the first way is used to determine the characteristic information of the data collected by the terminal device, and the second way is used to determine the characteristic distribution information of the data collected by the terminal device.
  • Terminal configuration information indicates the configuration conditions that should be met when the first terminal device collects P1 group data
  • Trigger information includes the trigger method and/or trigger configuration, the trigger method includes at least one of the following: event trigger, cycle Trigger; trigger configuration includes at least one of the following: trigger event, trigger cycle; quantity information, quantity information is used to indicate the value of P1; area information, area information is used to indicate the first area; time period information, time period information is used to indicate first time period.
  • the method also includes: sending the P1 group of data to the network device, and the training data of the AI model includes the P1 group of data.
  • the method further includes: receiving fifth request information from the network device, where the fifth request information is used to request the first terminal device to report data located in the first area and collected within the first time period.
  • the present application provides a communication device.
  • the communication device is provided with the functions related to the above-mentioned second to fourth aspects, sixth and seventh aspects.
  • the communication device includes modules or units or means corresponding to the operations related to the second to fourth aspects and the sixth and seventh aspects.
  • the functions, units or means can be implemented by software, or by hardware. Implementation can also be implemented by hardware executing corresponding software.
  • the communications device includes a processor, which may be coupled to a memory.
  • the memory may store necessary computer programs or instructions to implement the functions involved in the above-mentioned second to fourth aspects and the sixth and seventh aspects.
  • the processor can execute the computer program or instructions stored in the memory.
  • the communication device implements any possible design or implementation of the above-mentioned second to fourth aspects as well as the sixth and seventh aspects. method.
  • the communication device includes a processor and an interface circuit.
  • the processor communicates with other devices through the interface circuit, and executes the method in any possible design or implementation manner of the above-mentioned second to fourth aspects as well as the sixth and seventh aspects.
  • the processor can be implemented by hardware or software.
  • the processor may be a logic circuit, an integrated circuit, or the like.
  • the processor may be a general-purpose processor implemented by reading software code stored in memory.
  • the above processors may be one or more, and the memories may be one or more.
  • the memory can be integrated with the processor, or the memory can be provided separately from the processor. In the specific implementation process, the memory can It can be integrated with the processor on the same chip, or can also be installed on different chips.
  • the embodiment of the present application does not limit the type of memory and the arrangement method of the memory and the processor.
  • the present application provides a computer-readable storage medium.
  • Computer readable instructions are stored in the computer storage medium.
  • the computer reads and executes the computer readable instructions, the computer is caused to perform the method in any possible design of the above-mentioned first aspect to the fourth aspect.
  • the present application provides a computer program product, which when a computer reads and executes the computer program product, causes the computer to execute the method in any possible design of the above-mentioned first to fourth aspects.
  • the present application provides a chip, including a processor, which is coupled to a memory and configured to read and execute software programs stored in the memory to implement the second to fourth aspects and the sixth and third aspects. Seven possible approaches to any design.
  • Figure 1 is a schematic diagram of a communication system applicable to the embodiment of the present application.
  • Figure 2 is a schematic diagram of a histogram statistics provided by an embodiment of the present application.
  • Figure 3 is a schematic diagram of terminal equipment from different manufacturers in area 1 provided by the embodiment of the present application.
  • Figure 4 is a schematic diagram of different feature distribution provided by the embodiment of the present application.
  • Figure 5 is a schematic flow chart corresponding to the communication method provided in Embodiment 1 of the present application.
  • Figure 6 is a schematic flow chart corresponding to the communication method provided in Embodiment 2 of the present application.
  • Figure 7 is a schematic flow chart corresponding to the communication method provided in Embodiment 3 of the present application.
  • Figure 8 is a schematic flow chart corresponding to the communication method provided in Embodiment 4 of the present application.
  • Figure 9 is a possible exemplary block diagram of the device involved in the embodiment of the present application.
  • Figure 10 is a schematic structural diagram of a terminal device provided by an embodiment of the present application.
  • Figure 11 is a schematic structural diagram of a network device provided by an embodiment of the present application.
  • Figure 12 is a schematic structural diagram of a server provided by an embodiment of the present application.
  • the technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
  • the technical solutions in the embodiments of the present application can be applied to various communication systems, such as universal mobile telecommunications system (UMTS), wireless local area network (WLAN), wireless fidelity (wireless fidelity, Wi- Fi) system, global interoperability for microwave access (WiMAX) communication system, vehicle to everything (V2X) communication system, device-to-device (D2D) communication system, vehicle Networked communication systems, 4th generation (4G) mobile communication systems, such as long term evolution (LTE) systems, fifth generation (5th generation, 5G) mobile communication systems, such as new radio, NR) system, as well as future evolved communication systems, such as the sixth generation (6th generation, 6G) mobile communication system.
  • UMTS universal mobile telecommunications system
  • WLAN wireless local area network
  • Wi- Fi wireless fidelity
  • WiMAX global interoperability for microwave access
  • V2X vehicle to everything
  • the communication system includes multiple terminal devices (such as terminal device 101, terminal device 102 and terminal device 103) and one or more network devices (such as network device 110).
  • the communication system may also include a or multiple servers (such as server 121, server 122).
  • the terminal device may be a terminal that is connected to the above communication system and has a wireless transceiver function, or a chip or chip system that can be installed in the terminal.
  • the terminal equipment may also be called user equipment (UE), user device, access terminal, user unit, user station, mobile station, mobile station (MS), remote station, remote terminal, mobile device, User terminal, terminal, terminal unit, terminal station, terminal device, wireless communication device, user agent or user device.
  • the terminal device in the embodiment of the present application may be a mobile phone (mobile phone), a wireless data card, a personal digital assistant (personal digital assistant, PDA) computer, a laptop computer (laptop computer), a tablet computer (Pad), Drones, computers with wireless transceiver functions, machine type communication (MTC) terminals, virtual reality (VR) terminal equipment, augmented reality (AR) terminal equipment, Internet of Things (internet of things, IoT) terminal equipment, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical, wireless terminals in smart grid Terminals, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes (such as game consoles, smart TVs, smart speakers, smart refrigerators and fitness equipment etc.), vehicle-mounted terminals, RSUs with terminal functions.
  • MTC machine type communication
  • VR virtual reality
  • AR augmented reality
  • IoT Internet of Things
  • wireless terminals in industrial control wireless terminals in self-driving
  • wireless terminals in remote medical wireless
  • the access terminal can be a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA) , handheld devices (handsets) with wireless communication capabilities, computing devices or other processing devices connected to wireless modems, wearable devices, etc.
  • SIP session initiation protocol
  • WLL wireless local loop
  • PDA personal digital assistant
  • the terminal device in the embodiment of the present application may be an express terminal in smart logistics (such as a device that can monitor the location of cargo vehicles, a device that can monitor the temperature and humidity of cargo, etc.), a wireless terminal in smart agriculture (such as a device that can collect poultry, etc.) wearable devices with animal-related data), wireless terminals in smart buildings (such as smart elevators, fire monitoring equipment, and smart meters, etc.), wireless terminals in smart medical care (such as wireless terminals that can monitor the physiological status of people or animals) Wearable devices), wireless terminals in smart transportation (such as smart buses, smart vehicles, shared bicycles, charging pile monitoring equipment, smart traffic lights, smart monitoring and smart parking equipment, etc.), wireless terminals in smart retail (such as automatic vending machines) Cargo aircraft, self-service checkout machines, and unmanned convenience stores, etc.).
  • smart logistics such as a device that can monitor the location of cargo vehicles, a device that can monitor the temperature and humidity of cargo, etc.
  • a wireless terminal in smart agriculture such as a device that can collect poultry,
  • the terminal device of this application may be a vehicle-mounted module, vehicle-mounted module, vehicle-mounted component, vehicle-mounted chip or vehicle-mounted unit built into the vehicle as one or more components or units.
  • the vehicle uses the built-in vehicle-mounted module, vehicle-mounted module
  • the group, vehicle-mounted component, vehicle-mounted chip or vehicle-mounted unit can implement the method provided by this application.
  • the network device may be one of an access network device and a core network element, or the network device may be an integrated device of one or more devices in the core network element and the access network device.
  • the above-mentioned access network device is a device located on the network side of the above-mentioned communication system and has a wireless transceiver function, or a chip or chip system that can be installed on the device.
  • the access network equipment includes but is not limited to: access points (APs) in wireless fidelity (Wi-Fi) systems, such as home gateways, routers, servers, switches, bridges, etc., evolved Node B (evolved Node B, eNB), radio network controller (radio network controller, RNC), node B (Node B, NB), base station controller (base station controller, BSC), base transceiver station (base transceiver station, BTS), home base station (for example, home evolved NodeB, or home Node B, HNB), baseband unit (BBU), wireless relay node, wireless backhaul node, transmission point (transmission and reception point, TRP or transmission point, TP), etc., can also be 5G, such as gNB in the new radio (NR) system
  • the above-mentioned core network elements may include but are not limited to one or more of the following: user plane network elements, mobility management network elements, session management network elements, policy control network elements, and storage function network elements.
  • the user plane network element As the interface with the data network, it completes functions such as user plane data forwarding, session/flow level-based billing statistics, bandwidth limitation, etc. That is, packet routing and forwarding and quality of service (QoS) processing of user plane data, etc.
  • the user plane network element may be a user plane function (UPF) network element.
  • UPF user plane function
  • Mobility management network elements are mainly used for mobility management and access management.
  • the access management network element may be an access and mobility management function (AMF) network element, which mainly performs functions such as mobility management and access authentication/authorization.
  • AMF access and mobility management function
  • the mobility management network element is also responsible for the terminal and policy control functions (policy control function (PCF) to transmit user policies between network elements.
  • PCF policy control function
  • Session management network element Mainly used for session management (such as creation, deletion, etc.), maintenance of session context and user plane forwarding pipeline information, network interconnection protocol (IP) address allocation and management of user equipment, and selection of manageable users Termination points of plane functions, policy control and charging function interfaces, and downlink data notifications, etc.
  • the session management network element can be a session management function (SMF) network element, which completes terminal IP address allocation, UPF selection, accounting and QoS policy control, etc.
  • SMF session management function
  • Policy control network element including user subscription data management function, policy control function, billing policy control function, quality of service (QoS) control, etc. It is a unified policy framework used to guide network behavior and is a control plane functional network element (such as AMF, SMF network elements, etc.) to provide policy rule information, etc.
  • the policy control network element may be the PCF.
  • Storage function network element Provides storage and selection functions for network function entity information for other core network elements.
  • the network element may be a network function repository function (NRF) network element.
  • NRF network function repository function
  • the server can also be called an over-the-top (OTT) server.
  • the server can be associated with multiple terminal devices; when the terminal device is associated with the server, the terminal device can communicate with the server. As shown in FIG. 1 , the terminal device 101 and the terminal device 102 are associated with the server 121 , and the terminal device 103 is associated with the server 122 .
  • the terminal device 101 can communicate with the server 121; for example, the terminal device 101 can send data collected by the terminal device 101 to the server 121, and the data collected by the terminal device 101 can be used by the server 121 to generate an AI model, and the server 121
  • the configuration information of the AI model can be sent to the terminal device 101, and the terminal device 101 can use the AI model according to the configuration information of the AI model.
  • the communication system shown in Figure 1 may include multiple terminal devices. These multiple terminal devices may belong to different terminal device manufacturers, or the chips used by these multiple terminal devices may belong to different Chip manufacturers.
  • one or more terminal equipment manufacturers can form a group (or interest body), or one or more chip manufacturers can form a group, or one or more terminal equipment manufacturers and one or more chip manufacturers can form a group.
  • a group can maintain a server or a server cluster (a server cluster can include multiple servers). In the embodiment of this application, "a group maintains a server" will be used as an example for description. Servers maintained by a group are associated with the group's terminal devices.
  • terminal equipment 101 and terminal equipment 102 both belong to terminal equipment manufacturer 1, and server 121 is a server maintained by terminal equipment manufacturer 1. Therefore, terminal equipment 101 and terminal equipment 102 are associated with server 121; terminal equipment 103 belongs to terminal equipment Manufacturer 2 and server 122 are servers maintained by terminal device manufacturer 2. Therefore, terminal device 103 is associated with server 122.
  • the communication method provided by the embodiment of the present application can be applied between the server, terminal device and network device shown in Figure 1, or can also be applied between the terminal device and the network device.
  • the communication system may also include at least one of other network devices, other terminal devices, and other servers, which are not shown in FIG. 1 .
  • a group of data may also be called one piece of data.
  • a group of data is the first data
  • the first data may be determined based on the downlink channel state information of the terminal device; or the first data may be data related to the positioning of the terminal device.
  • the embodiments of this application do not limit the first data.
  • the first data may be different.
  • the characteristic information of the first data is used to indicate the characteristics of the first data.
  • the characteristics of the first data may be obtained by processing the first data in the first manner.
  • the first method may include a first processing process.
  • the first processing process may include histogram statistics.
  • the first manner may also include a second processing process and/or a third processing process, where the second processing process is a processing process before the first processing process, and the third processing process is a processing process after the first processing process.
  • the characteristics of the first data may be obtained according to the first processing process, or the characteristics of the first data may be obtained according to the second processing process and the first processing process, or the characteristics of the first data may be According to the first treatment process and the or the characteristics of the first data may be obtained according to the first processing process, the second processing process and the third processing process.
  • the first process may include histogram statistics.
  • Figure 2 is a schematic diagram of histogram statistics provided by an embodiment of the present application.
  • the first data is three-dimensional data, for example, the dimensions are W ⁇ H ⁇ D, W and H can be information related to the first data, D can be 2, and D represents the real part and the imaginary part.
  • the traversal process can have overlap or no overlap, and for each cell
  • Performing histogram statistics on the grid P ⁇ Q ⁇ T can obtain histogram data, and all the obtained histogram data constitute the characteristics (matrix or vector) of the first data.
  • the histogram data may be based on the value of each element in the cell P ⁇ Q ⁇ T, or may be based on the direction and intensity of each element in the cell P ⁇ Q ⁇ T. It can be understood that this application does not limit the dimensions of the first data, and Figure 2 is only an example provided by this application.
  • the first data is determined based on downlink channel status information.
  • the first data includes downlink channel status information.
  • Downlink channel state information can be represented by the first dimension, the second dimension and the third dimension.
  • the first dimension corresponds to the number of carriers or the number of subbands
  • the second dimension corresponds to the number of antenna ports or the number of radio frequency chains of the network device
  • the third dimension represents the real part and the imaginary part.
  • the first dimension is W
  • the second dimension is H
  • the third dimension is D
  • the downlink channel status information can be W ⁇ H ⁇ D.
  • the downlink channel status is traversed in the unit of cell P ⁇ Q ⁇ T. information, and conduct histogram statistics for each cell P ⁇ Q ⁇ T to obtain histogram data. All the histogram data obtained constitute the characteristics (matrix or vector) of the downlink channel state information.
  • the first processing process may include wireless channel feature extraction, where the wireless channel features may include but are not limited to one or more of the following features: power delay spectrum profile, PDP), time-varying Doppler spectrum, angular power spectrum (angular power spectrum, APS), PDP sparsity, Doppler spectrum sparsity, APS sparsity.
  • the wireless channel features may include but are not limited to one or more of the following features: power delay spectrum profile, PDP), time-varying Doppler spectrum, angular power spectrum (angular power spectrum, APS), PDP sparsity, Doppler spectrum sparsity, APS sparsity.
  • the second processing process may include but is not limited to one or more of the following: Fourier transform, fast fourier transform (FFT), discrete fourier transform (DFT) , data truncation, coordinate conversion, amplitude taking, noise reduction, normalization, grayscale, data correction (such as gamma correction), quantization, singular value decomposition (SVD), translation, data expansion, gradient finding , feature template filtering, part selection, and combination of multiple processing results.
  • FFT fast fourier transform
  • DFT discrete fourier transform
  • data truncation coordinate conversion
  • amplitude taking amplitude taking
  • noise reduction normalization
  • grayscale grayscale
  • data correction such as gamma correction
  • quantization singular value decomposition
  • translation data expansion
  • gradient finding feature template filtering
  • part selection part selection, and combination of multiple processing results.
  • finding the gradient may include binarizing the gradient in the local binary mode algorithm.
  • Feature template filtering can include performing Sobel operators on data (such as features), calculating to obtain Haar-like features, and/or obtaining edges in the data through canny edge detection algorithms, etc.
  • Part of the selection may include: extracting the region of interest (ROI) using methods such as scale-invariant feature transformation.
  • Combining multiple processing results may include combining data calculated by multiple different Sobel operators.
  • this application does not include Fourier transform, FFT, DFT, data truncation, coordinate conversion, amplitude taking, noise reduction, normalization, grayscale, and data correction (for example, gamma correction), quantization, SVD, translation, data expansion, gradient finding, feature template filtering, part selection, and the execution order of combining multiple processing results are limited.
  • the third processing process may include, but is not limited to, one or more of the following: normalization, feature combination, dimensionality reduction, quantization, and encoding.
  • Feature combinations may include: histogram feature combinations obtained through multiple scales (for example, preprocessing interpolation or dimensionality reduction to generate multiple images at different scales, and then extracting features separately), or different configurations (for example, cells of different sizes).
  • Reducing the dimensionality may include but is not limited to one or more of the following: combining and summing the histogram data (data obtained after performing histogram statistics), and compressing the dimensions of the histogram data by reducing the feature dimension.
  • a combined summation of the histogram data can result in fewer feature dimensions.
  • the combined summation of the histogram data may include separately adding rows and columns of the histogram data matrix in the deformable part model algorithm.
  • methods for reducing feature dimensions may include but are not limited to: principal component analysis (PCA) and linear discriminant analysis (LDA).
  • the first method may be any of the following: histogram of oriented gradient (HOG), local binary mode (local binary patterns, LBP), edge orientation histogram (EOH), edge histogram descriptor (edge histogram descriptor, EHD), edge direction histogram (EDH), deformable part model ( deformable part model (DPM), scale-invariant feature transform (SIFT) or speed up robust features (SURF).
  • HOG histogram of oriented gradient
  • LBP local binary mode
  • EOH edge orientation histogram
  • EHD edge histogram descriptor
  • EHD edge direction histogram
  • EH edge direction histogram
  • DPM deformable part model
  • SIFT scale-invariant feature transform
  • SURF speed up robust features
  • the HOG method includes not only the first processing process, but also the second processing process such as grayscale and gradient calculation, and the third processing process such as normalization.
  • the local binary pattern LBP method includes a first processing process, a second processing process and a third processing process. This application does not elaborate on each of them.
  • the second processing process before performing histogram statistics on the first data, the second processing process is used to process the first data, and/or, after performing histogram statistics on the first data, a third processing process is used to perform histogram statistics.
  • the results are processed to further increase the accuracy of the features of the first data.
  • the first data can be input into the AI model; as an implementation method, the AI model can be the encoding part of the autoencoder (Autoencoder) or the backbone network part of the AI classifier. Furthermore, the AI model can output the characteristics of the first data.
  • the first data can be preprocessed, and the specific preprocessing method is not limited.
  • the multiple sets of data may be determined based on multiple sets of downlink channel status information.
  • the multiple sets of data include multiple sets of downlink channel status information; or the multiple sets of data are data related to the positioning of the terminal device.
  • the multiple sets of data may include the above-mentioned first data, or the above-mentioned first data is one set of data among the multiple sets of data.
  • the feature distribution information of multiple sets of data is used to indicate the feature distribution of multiple sets of data.
  • the characteristic distribution of multiple sets of data can represent the number or proportion of data belonging to each category in multiple categories.
  • the feature distribution of multiple sets of data can be obtained by processing the multiple sets of data in the second manner. The second method is described below in conjunction with Example 1 and Example 2.
  • the second manner may include the first manner and the third manner.
  • using the second method to process multiple groups of data to obtain the characteristic distribution of the multiple groups of data may mean: using the first method to process each group of data in the multiple groups of data to obtain each group of data in the multiple groups of data. Characteristics of multiple groups of data; further, a third method is used to process the characteristics of multiple groups of data to obtain characteristic distribution of multiple groups of data.
  • the third way may include but is not limited to at least one of the following: taking an average value for each element in the features of multiple sets of data in the dimension of the group, taking an average value in the dimension of the group for each element of the features of multiple sets of data.
  • the quantile can be obtained by statistics in the dimension. For each element in the characteristics of multiple groups of data, Gaussian fitting can be performed on the group dimension to obtain the mean and variance, or other distribution fitting can obtain its parameter value.
  • group A downlink channel status information may include group A W ⁇ H ⁇ D data, where A is an integer greater than 1.
  • group A W perform histogram statistics on each group of data in group A to obtain the characteristics of group A data.
  • the second way may include the first way, or the second way may be the same as the first way.
  • using the second method to process multiple sets of data to obtain feature distributions of multiple sets of data may mean: using the first way to process multiple sets of data to obtain feature distributions of multiple sets of data.
  • the above downlink channel state information is represented by the first dimension, the second dimension and the third dimension, and may include: the downlink channel state information is represented by the first dimension, The second, third and fourth dimensions are represented. Assuming that the first dimension is represented by W, the second dimension is represented by H, the third dimension is represented by D, and the fourth dimension is represented by A, then the example of group A downlink channel state information can refer to the following method a or method b.
  • the downlink channel status information of group A can be W ⁇ H ⁇ (D ⁇ A), or W ⁇ (H ⁇ A) ⁇ D, or (W ⁇ A) ⁇ H ⁇ D data.
  • the first dimension can be , any one of the second and third dimensions is expanded to A times.
  • the downlink channel status information of Group A can be W ⁇ H ⁇ D ⁇ A data, that is, the downlink channel status information of Group A can be represented by four-dimensional data.
  • group A When group A downlink channel status information is W ⁇ H ⁇ (D ⁇ A), or W ⁇ (H ⁇ A) ⁇ D, or (W ⁇ A) ⁇ H ⁇ D data, group A is obtained according to the first method
  • the implementation method of the characteristic distribution of the downlink channel state information may refer to the corresponding implementation method when the downlink channel state information is represented by the first dimension, the second dimension and the third dimension.
  • the traversal process can have overlap or no overlap, and perform histogram statistics for each cell P ⁇ Q ⁇ T to obtain histogram data. All the histogram data obtained constitute group A. Characteristic distribution of data.
  • the characteristic distribution of group A downlink channel state information can be obtained according to the first method.
  • the downlink channel state information can be obtained through the first dimension, the second dimension and The third dimension represents the corresponding implementation method.
  • the downlink channel status information is traversed in units of cells P ⁇ Q ⁇ T ⁇ R, 1 ⁇ P ⁇ W, 1 ⁇ Q ⁇ H, 1 ⁇ T ⁇ D, 1 ⁇ R ⁇ A.
  • the traversal process may overlap. There can also be no overlap, and histogram statistics can be performed for each cell P ⁇ Q ⁇ T ⁇ R to obtain histogram data. All the histogram data obtained constitute the characteristic distribution of group A data.
  • the downlink channel status information of Group A may include the downlink channel status information of Group A collected by the terminal equipment at different time points within a time period; for example, the downlink channel status information of Group A collected by the terminal equipment at time point 1 One or more sets of downlink channel status information. One or more sets of downlink channel status information are collected at time point 2, and so on.
  • the terminal equipment collects a total of group A of downlink channel status information within a time period.
  • the terminal device can send the data (which can be called training data) collected by the terminal device in a specific area (such as area 1) to the server.
  • the terminal device reports the collected data to a server associated with the terminal device.
  • the training data received by the server is reported by the terminal device associated with the server.
  • the server generates an AI model based on the training data, and sends the configuration information of the AI model to the terminal device, so that the terminal device can use the AI model in area 1 based on the configuration information of the AI model.
  • two terminal equipment manufacturers are taken as an example.
  • the two terminal equipment manufacturers are manufacturer 1 and manufacturer 2 respectively.
  • Manufacturer 1 manages server 1, and manufacturer 2 manages server 2.
  • the terminal equipment of manufacturer 1 includes m1 terminal equipment
  • the terminal equipment of manufacturer 2 includes m2 terminal equipment.
  • m1 and m2 are both positive integers, and m1 and m2 may be the same or different.
  • m1 terminal devices respectively send the training data collected by m1 terminal devices in area 1 to server 1, and then server 1 can generate AI model 1 based on the received training data; similarly, m2 terminal devices send m2 training data to server 2 respectively.
  • the terminal device collects training data in area 1, and server 2 can generate AI model 2 based on the received training data.
  • the feature distribution of the training data uploaded to the respective servers will also be different.
  • the characteristic distribution of the training data of AI model 1 is shown in (a) in Figure 4
  • the characteristic distribution of the training data of AI model 2 is shown in (b) of Figure 4.
  • area 1 including m1 terminal devices and m2 terminal devices is the total training data in area 1 (including m1 terminal equipment collected in area 1 Schematic diagram of feature distribution of training data and training data collected by m2 terminal devices in area 1).
  • the characteristic distribution of the training data of AI model 1 is different from the characteristic distribution of the total training data, which will cause the performance of AI model 1 to be worse when the terminal equipment of manufacturer 1 uses AI model 1 in area 1.
  • Difference. For example, the terminal equipment of Manufacturer 1 uses AI model 1 to encode the downlink channel status information in Area 1 and sends it to the network device. The downlink channel status information decoded by the network equipment is distorted; for another example, the terminal equipment of Manufacturer 1 The positioning results obtained using the AI model within Area 1 are inaccurate.
  • the feature distribution of the training data of AI model 2 is also different from the feature distribution of the total training data, which will lead to poor performance of AI model 2 when the terminal equipment of manufacturer 2 uses AI model 2 in area 1.
  • embodiments of the present application provide a communication method to solve the problem that the characteristic distribution of the training data of the AI model is different from the characteristic distribution of the total training data in a specific area, thereby affecting the performance of the AI model.
  • Figure 5 is a schematic flowchart corresponding to the communication method provided in Embodiment 1 of the present application.
  • a terminal device, a network device, and a server are used as the execution subjects of the interaction gesture as an example to illustrate the method.
  • this application does not limit the execution subjects of the interaction gesture.
  • the terminal device in Figure 5 can also be a chip, chip system, or processor that supports the terminal device to implement the method;
  • the network device in Figure 5 can also be a chip, chip that supports the access network device to implement the method
  • the system or processor can also be a logic module or software that can realize all or part of the functions of the access network equipment;
  • the server in Figure 5 can also be a chip, chip system, or processor that supports the server to implement the method.
  • the method includes the following steps:
  • the first terminal device sends first information to the network device.
  • the first information includes at least one of the following: characteristic information of the P1 group data and characteristic distribution information of the P1 group data.
  • the P1 group data indicates that the first terminal device is located in the first area. And for the data collected in the first time period, P1 is a positive integer.
  • the first area may include multiple cells. These multiple cells may belong to the same base station, or may belong to different base stations. The description below will take "these multiple cells belong to the same base station, and the network equipment is the base station to which the multiple cells belong" as an example.
  • the first terminal device may always be located in the first area during the first time period, or the first terminal device may move out of the first area before the end of the first time period. . If the first terminal device is always located in the first area during the first time period, the first terminal device can obtain the characteristics of the P1 group of data based on the P1 group of data collected during the first time period after the first time period. information and/or feature distribution information, and send the first information to the network device. Or, if the first terminal device moves out of the first area before the end of the first time period, for example, the starting time of the first time period is time point t1, the end time is time point t2, and the first terminal device moves out of the first area at time point t3.
  • time point t3 is before time point t2, then before moving out of the first area, the first terminal device can use the P1 group of data collected in the first time period (that is, from time point t1 to time point t3). P1 group of data collected within a period of time), obtain the characteristic information and/or characteristic distribution information of the P1 group of data, and send the first information to the network device.
  • the implementation of the first terminal device obtaining the feature information and/or feature distribution information of the P1 group of data may refer to the foregoing description and will not be described again.
  • the above method may also include: the network device may send first request information to the terminal device located in the first area, where the first request information is used to request the terminal device located in the first area to perform the processing in the first time period. Collect data within the system and report the characteristic information and/or characteristic distribution information of the collected data.
  • the terminal device located in the first area includes the first terminal device, and the first terminal device can receive the first request information.
  • the network device may send the first request information in multiple ways; for example, the network device may send the first request information to the terminal device located in the first area by broadcasting.
  • the first request information may include at least one of the following:
  • the first indication information indicates that the information requested to be reported by the terminal device is the characteristic information and/or the characteristic distribution information of the data collected by the terminal device.
  • the second indication information indicates the first method and/or the second method.
  • the first method is used to determine the characteristic information of the data collected by the terminal device
  • the second method is used to determine the characteristic distribution of the data collected by the terminal device. information.
  • the terminal configuration information indicates the configuration conditions that the first terminal device should meet when collecting P1 group data.
  • the terminal configuration information includes: the identifier of the first frequency band, and/or the number of the first antenna; in this case, the configuration conditions that the first terminal device should satisfy when collecting the P1 group of data refer to: the first terminal device
  • the frequency band used is the first frequency band, and/or the number of antennas of the first terminal device is equal to the number of first antennas.
  • Trigger information includes triggering mode and/or triggering configuration.
  • Triggering mode includes at least one of the following: event triggering and periodic triggering.
  • Triggering configuration includes at least one of the following: triggering event and triggering period. For example, the triggering event is that the communication system performance meets certain conditions.
  • the communication system performance may include at least one of the following: throughput rate, bit error rate (BER), block error rate (block error rate, BLER), acknowledgment (ACK)/negative acknowledgment (NACK) ratio, retransmission rate, etc.; for another example, the trigger event can satisfy certain conditions for the terminal equipment measurement quantity, and the measurement quantity may include at least one of the following : Signal to Interference plus Noise Ratio (SINR), Reference Signal Receiving Power (RSRP), Received Signal Strength Indication (RSSI), Reference Signal Receiving Quality (Reference Signal Receiving Quality (RSRQ), Received Signal Code Power (RSCP).
  • SINR Signal to Interference plus Noise Ratio
  • RSRP Reference Signal Receiving Power
  • RSSI Received Signal Strength Indication
  • RSSQ Reference Signal Receiving Quality
  • RSSCP Received Signal Code Power
  • the terminal device receives the first request information and can collect data in the first area and within the first time period according to the trigger information.
  • quantity information is used to indicate the value of P1.
  • the area information may include the identifiers of the multiple cells. Taking cell 1 among the multiple cells as an example, the identifier of cell 1 may be the physical cell identifier of cell 1. PCI).
  • the time period information is used to indicate the first time period.
  • the time period information may include the start time and end time of the first time period, or include the start time (or end time) of the first time period and the duration of the first time period.
  • the starting time of the first time period is X1 year Take value.
  • the third indication information indicates a condition that needs to be met by at least one feature of the data collected by the terminal device.
  • the conditions that the feature needs to meet may refer to: when the distance between the feature and a certain vector is greater than or equal to a threshold, the definition of a certain vector, threshold, and distance should include In the third instruction message or as agreed in advance.
  • a certain vector can be a feature of the training set of the current AI model, and the distance can be Euclidean distance, Manhattan distance, Chebyshev distance, Minkovsky distance, cosine similarity, Mahalanobis distance, Hamming distance One of the definitions of equal distance.
  • the terminal device when the terminal device collects a set of data, if at least one feature of the set of data does not meet the above conditions, the terminal device can determine that the set of data is invalid data, and the above-mentioned P1 set of data is all valid data, excluding invalid data. data.
  • the request information does not include the first indication information
  • whether the terminal device reports feature information or feature distribution information may be predefined by the protocol or preconfigured.
  • the request information does not include the second indication information
  • the method of determining the characteristic information and/or the characteristic distribution information may be predefined by the protocol or preconfigured.
  • the request information does not include area information
  • the scope of the first area may be predefined by the protocol or preconfigured.
  • the request information does not include the first time period, for example, the starting time of the first time period can be determined based on the reception time of the request information, and the length of the first time period can be predefined by the protocol or preconfigured.
  • the terminal device may consider that the multiple sets of data collected by the terminal device in the first area and within the first time period are all valid data. Other items of information can be understood by reference and will not be repeated one by one.
  • the terminal device may also send indication information to the network device.
  • the indication information indicates at least one of the terminal configuration information, trigger information, quantity information, area and time period corresponding to the P1 group data, such as an indication.
  • the information includes the above-mentioned area information and time period information.
  • the network device sends the first feature distribution information to the first terminal device, and accordingly, the first terminal device receives the first feature distribution information.
  • the network device can receive the first information reported by the first terminal device.
  • the network device can also receive the information reported by other terminal devices, such as the second information reported by the second terminal device and the third information reported by the third terminal device.
  • Three messages The second information includes at least one of the following: characteristic information of the P2 group data, characteristic distribution information of the P2 group data, the P2 group data is data collected by the second terminal device located in the first area and within the first time period, and P2 is positive integer.
  • the third information includes at least one of the following: characteristic information of the P3 group data and characteristic distribution information of the P3 group data.
  • the P3 group data is the data collected by the third terminal device located in the first area and within the first time period. P3 is positive integer.
  • the network device After the network device receives the information reported by multiple terminal devices, it can aggregate and obtain the first feature distribution information, that is, the first feature distribution information is based on the first information (optionally, as well as the second information and the third information) Sure.
  • the first characteristic distribution information is used to indicate the characteristic distribution of data collected by the terminal device located in the first area within the first time period.
  • the terminal devices located in the first area may include terminal devices associated with different servers.
  • the terminal devices located in the first area include terminal devices associated with the first server (such as the first terminal device and the second terminal device), and the second server Associated terminal equipment (such as a third terminal equipment). That is to say, the first feature distribution information is used to indicate the feature distribution of the total training data in the first area.
  • the embodiments of this application do not limit the specific implementation of "network devices summarize and obtain the first feature distribution information".
  • the first information includes the characteristic information of the P1 group of data
  • the second information includes the characteristic information of the P2 group of data.
  • the network device can
  • the characteristics of group data and the characteristics of group P2 data are processed (for example, the third method described above is used to process the characteristics of group P1 data and the characteristics of group P2 data) to obtain the first feature distribution information.
  • the network device can actively send the first feature distribution information to the terminal device located in the first area; for example, the network device sends the first feature distribution information to the terminal device located in the first area by broadcasting.
  • the network device may send the first feature distribution information to the first terminal device based on the request of the first terminal device. For example, before S502, the first terminal device may send second request information to the network device. The second request information is used to request to send the first feature distribution information to the first terminal device; accordingly, the network device receives the second request information. Afterwards, the first feature distribution information may be sent to the first terminal device.
  • the first terminal device may receive third request information from the first server, where the third request information is used to request to send the first request information to the first server.
  • Feature distribution information is used to request to send the first request information to the first server.
  • the first server is associated with the first terminal device. That is to say, the first terminal device may request the first feature distribution information from the network device based on the request of the first server.
  • the third request information may include area information and/or time period information, and the content included in the second request information may be determined based on the content included in the third request information, such as the content included in the second request information and the third request information.
  • the contents included in the three request messages are the same.
  • the first terminal device may send second request information to the network device, and the second request information is used to request to send the first feature distribution information to the first terminal device; accordingly, after the network device receives the second request information, , it can be learned that the first feature distribution information needs to be sent to the first terminal device.
  • the first terminal device may send the second request information to the network device before the network device sends the first request information (see the description in S501); that is, after the network device receives the second request information, it may The second request information sends the first request information.
  • the first terminal device may receive third request information from the first server, where the third request information is used to request to send the first feature to the first server. Distribution information.
  • the third request information may include at least one of the following: first indication information, second indication information, terminal configuration information, trigger information, quantity information, area information, time period information, and third indication information.
  • the content included in the second request information may be determined based on the content included in the third request information.
  • the content included in the second request information is the same as the content included in the third request information.
  • the content included in the first request information may be determined based on the content included in the second request information.
  • the content included in the first request information and the content included in the second request information are the same.
  • Implementation Mode 2 the network device independently determines the first feature distribution information, and sends the first feature distribution information to the first terminal device based on the request of the first terminal device.
  • Feature distribution information in implementation mode 3, the network device determines the first feature distribution information based on the request of the first terminal device, and sends the first feature distribution information to the first terminal device.
  • the network device may also send indication information to the first terminal device, the indication information indicating the terminal configuration information, trigger information, quantity information, area information and time period information corresponding to the first characteristic distribution information.
  • the indication information includes the above-mentioned area information and time period information.
  • the first information in the above S501 may also include the P1 group data, but not the characteristic information of the P1 group data and the characteristic distribution information of the P1 group data.
  • the network device can determine the first feature distribution information based on the P1 group data (and optionally, the P2 group data and the P3 group data).
  • the above S501 is an optional step.
  • the first terminal device may not be in the first area, and the first terminal device may not send the first information to the network device.
  • the terminal device that reports feature information and/or feature distribution information to the network device may overlap with the terminal device that receives the first feature distribution information. That is to say, a certain terminal device may report feature information and/or feature distribution information to the network device, but does not receive the first feature distribution information from the network device; or, a certain terminal device may not report features to the network device. information and/or feature distribution information, but received the first feature distribution information from the network device; or, a terminal device reported feature information and/or feature distribution information to the network device, but also received the third feature distribution information from the network device. 1. Feature distribution information.
  • the first terminal device sends the first feature distribution information to the first server.
  • the first server receives the first feature distribution information from the first terminal device.
  • the first terminal device may also send indication information to the first server, the indication information indicating at least one of terminal configuration information, trigger information, quantity information, area information and time period information corresponding to the first characteristic distribution information,
  • the indication information includes the above-mentioned area information and time period information.
  • the first server generates the first AI model based on the first feature distribution information and the first training data.
  • the first server uses the first feature distribution information as auxiliary information for generating the first AI model.
  • the first server can The training process is adjusted based on the first feature distribution information, so that the first AI model generated based on the first training data is close to the AI model generated based on the total training data in the first area.
  • the first training data please refer to the description below.
  • the embodiments of this application do not limit the specific implementation of "the first server generates the first AI model based on the first feature distribution information and the first training data".
  • the first server can generate the first AI model based on the first feature distribution information and the first training data.
  • the first server may first determine whether the difference between the first feature distribution information and the second feature distribution information satisfies the first condition. If yes, the first AI model can be generated based on the first training data and the first feature distribution information; if not, there is no need to generate the first AI model.
  • the second feature distribution information is obtained in advance by the first server.
  • the first server before S504, the first server generates a second AI model based on the second training data and the second feature distribution information.
  • the first server receives the first feature distribution information
  • the first server determines that the difference between the first feature distribution information and the second feature distribution information meets the first condition, it means that the second AI model is not suitable for the current data.
  • the second AI model needs to be updated; further, the first server can generate the first AI model based on the first training data and the first feature distribution information. If the first server determines that the difference between the first feature distribution information and the second feature distribution information does not meet the first condition, it means that the second model can still be applied to the current data, and there is no need to update the second AI model.
  • the embodiment of the present application provides a judgment basis for updating the AI model, that is, the first server can judge whether the AI model needs to be updated based on the first feature distribution information and the second feature distribution information, so that it can more easily and quickly identify Whether the AI model needs to be updated so that the AI model can be updated more reasonably to avoid frequent updates to the AI model, which will cause a heavy processing load on the server, or the AI model has not been updated for a long time, resulting in poor performance of the AI model.
  • the quantitative index is distance
  • the difference between the first feature distribution information and the second feature distribution information satisfies the first condition, which may mean: the distance between the first feature distribution information and the second feature distribution information is greater than or equal to the distance threshold.
  • the distance threshold can be pre-configured.
  • the distance between the first feature distribution information and the second feature distribution information may include but is not limited to one or more of the following: cosine similarity, Euclidean distance, Manhattan distance, standardized Euclidean distance Euclidean distance), squared fuclidean distance, canberra distance, chebyshev distance, correlation distance, mahalanobis distance and Minkowski distance distance(minkowski distance).
  • the quantitative index is a deviation measure between two distributions
  • the difference between the first feature distribution information and the second feature distribution information satisfies the first condition, which may refer to: the first feature distribution information and the second feature distribution
  • the distance between the information is greater than or equal to the deviation measure threshold between the two distributions.
  • the deviation metric threshold can be pre-configured.
  • the deviation measure between the first feature distribution information and the second feature distribution information may include but is not limited to one or more of the following: KL divergence, Kolmogorov-Smirnov test (Kolmogorov-Smirnov test) test, K-S test), model stability index (population stability index, PSI), Hellinger distance (Hellinger distance), chi-squared test (chi-squared test), cross entropy, feature cardinality or frequency, etc.
  • the first server can send fourth request information to the terminal device associated with the first server.
  • the fourth request information is used to request the terminal device to report that the terminal device is located in the first area and in the first time period.
  • the data collected within the first server that is, the first server can request the terminal device to report the data collected previously by the terminal device.
  • the first terminal device can send the P1 group of data to the first server.
  • the first training data includes the P1 group of data; for another example, the first server If the associated terminal device includes a second terminal device, the second terminal device may send the P2 set of data to the first server. In this case, the first training data includes the P2 set of data.
  • the first server may send fourth request information to the terminal device associated with the first server.
  • the fourth request information is used to request the terminal device to report that the terminal device is located in the first area and at the second time. data collected within the segment.
  • the duration of the second time period is the same as that of the first time period, and the start time of the second time period is located after the end time of the first time period. That is, the first server may request the terminal device to collect data again within the second time period and report the collected data.
  • the time interval between the start time of the second time period and the end time of the first time period is small.
  • the terminal device is located in the first area and The data collected during the first time period have the same characteristic distribution as the data collected during the second time period when the terminal device is located in the first area. That is to say, it can be considered that the first characteristic distribution information is also used to indicate the characteristic distribution of data collected by the terminal device located in the first area in the second time period.
  • the terminal device associated with the first server includes the first terminal device.
  • the first terminal device collects the Q1 set of data in the first area and within the second time period and reports it to the first server.
  • the first training The data includes the Q1 group of data; for another example, the terminal device associated with the first server includes the second terminal device, and the second terminal device collects the Q2 group of data in the first area and within the second time period, and reports it to the first server.
  • the first training data includes the Q2 group of data.
  • Q1 and Q2 are both positive integers.
  • the first server determines that the AI model needs to be updated, it can send the fourth request information to the first terminal device, and then the first terminal device reports the collected data to the first server based on the fourth request; that is, when the first When the server does not need to update the AI model, the first terminal device does not need to report the collected data to the first server, thereby effectively saving transmission resources.
  • the terminal device that sends the training data to the first server may overlap with the terminal device that sends the first feature distribution information to the first server.
  • the first training data may also be data pre-acquired by the first server, such as data pre-stored in the first server.
  • the first server sends the configuration information of the first AI model to the first terminal device.
  • the first server can actively send the configuration information of the first AI model to one or some terminal devices associated with the first server; or, the first server can also send the configuration information of the first AI model based on one or some terminal devices associated with the first server.
  • the configuration information of the first AI model is sent to the terminal device. That is to say, the terminal device that sends the first feature distribution information to the first server and the terminal device that receives the configuration information of the first AI model may be the same terminal device, or they may be different.
  • the description here takes the first server sending the configuration information of the first AI model to the first terminal device as an example.
  • the configuration information of the first AI model is used to configure the first AI model.
  • the configuration information of the first AI model includes parameters of the first AI model, and the parameters of the first AI model include, for example, the structural parameters and/or weights of the first AI model; for another example, the configuration information of the first AI model includes the first AI model.
  • the obtained address may be, for example, an Internet protocol (IP) address, or may also be the identifier and/or address of the device.
  • IP Internet protocol
  • the embodiments of this application do not limit the application scenarios of the first AI model.
  • the AI model can be an algorithm model with different functions.
  • the application scenarios of the first AI model may include but are not limited to one or more of the following: downlink channel state information encoding on the terminal device side, beam management on the terminal device side, and positioning on the terminal device side.
  • the first server can determine whether the difference between the third feature distribution information and the first feature distribution information satisfies the first condition. If the first condition is met, , then a new AI model can be generated based on the third feature distribution information to update the AI model.
  • the network device sends the first feature distribution information to the first terminal device, and the first terminal device sends the first feature distribution information to the first server associated with the first terminal device.
  • the first training data used by the first server to generate the AI model is data collected by the terminal device associated with the first server in the first area
  • the feature distribution of the first training data may be different from the total training in the first area.
  • the first server can obtain the first characteristic distribution information (used to indicate the characteristic distribution of the total training data in the first area), the first server can use the first characteristic distribution information as the generated AI
  • the auxiliary information of the model makes the AI model generated based on the first training data close to the AI model generated based on the total training data in the first area, which can effectively improve the performance of the AI model.
  • Figure 6 is a schematic flow chart corresponding to the communication method provided in Embodiment 2 of the present application. The process shown in Figure 6 can correspond to implementation mode 1 in Embodiment 1.
  • the method includes the following steps:
  • the network device sends the first request information to the terminal device located in the first area; accordingly, M terminal devices located in the first area receive the first request information, where M is a positive integer.
  • the first request information is used to request the terminal device located in the first area to collect data within a first time period and report characteristic information and/or characteristic distribution information of the collected data.
  • the network device can send the first request information by broadcasting.
  • the network device periodically determines the characteristic distribution of the total training data within a set time period in the first area, and then the network device can periodically send request information to request the terminal device located in the first area to collect data within the set time period. data, and report the characteristic information and/or characteristic distribution information of the collected data.
  • S602 M terminal devices collect data in the first area and within the first time period, and report the characteristic information and/or characteristic distribution information of the collected data to the network device.
  • the M terminal devices are associated with different servers.
  • some of the M terminal devices are associated with servers maintained by vendor 1, and another part of the M terminal devices are associated with servers maintained by vendor 2.
  • the network device determines the first feature distribution information based on the feature information and/or feature distribution information reported by the M terminal devices.
  • the network device sends the first feature distribution information to the terminal devices located in the first area; accordingly, N terminal devices located in the first area receive the first feature distribution information, where N is a positive integer.
  • the network device may send the first feature distribution information by broadcasting.
  • M terminal devices and N terminal devices may overlap.
  • the N terminal devices may include M terminal devices.
  • N terminal devices respectively send the first feature distribution information to their respective associated servers.
  • the N terminal devices include a first terminal device, and the first terminal device is associated with the first server.
  • the first server receives the first feature distribution information and determines that the difference between the first feature distribution information and the second feature distribution information satisfies the first condition.
  • the first server sends the fourth request information to the K terminal devices.
  • the fourth request information is used to request the K terminal devices to collect data in the first area and within the second time period and report the collected data.
  • K is positive. integer.
  • K terminal devices and N terminal devices may overlap.
  • K terminal devices collect data in the first area and within the second time period, and report the collected data to the first server.
  • the first server generates the first AI model based on the first feature distribution information and the first training data.
  • the first training data includes data collected by K terminal devices in the first area and within the second time period.
  • the first server sends the configuration information of the first AI model to the K terminal devices.
  • Figure 7 is a schematic flowchart corresponding to the communication method provided in Embodiment 3 of the present application. The process shown in Figure 7 can correspond to implementation mode 3 in Embodiment 1.
  • the method includes the following steps:
  • the first server sends third request information to K terminal devices associated with the first server.
  • the third request information is used to request the K terminal devices to send first feature distribution information to the first server.
  • the third request information may include at least one of the following: first indication information, second indication information, terminal configuration information, trigger information, quantity information, area information, time period information, and third indication information.
  • the K terminal devices are terminal devices located in the first area.
  • K terminal devices send second request information to the network device.
  • the K terminal devices include a first terminal device, and the second request information sent by the first terminal device is used to request to send the first feature distribution information to the first terminal device.
  • the network device After receiving the second request information from the K terminal devices, the network device sends the first request information to the terminal devices located in the first area; accordingly, the M terminal devices located in the first area receive the request information.
  • the network device sends the first request information by broadcasting.
  • the M terminal devices include K terminal devices, and may also include terminal devices associated with the second server or other servers.
  • S704 M terminal devices collect data in the first area and within the first time period, and report the characteristic information and/or characteristic distribution information of the collected data to the network device.
  • the network device determines the first feature distribution information based on the feature information and/or feature distribution information reported by the M terminal devices.
  • the network device sends the first feature distribution information to the K terminal devices; accordingly, the K terminal devices receive the first feature distribution information. Distribute information.
  • the network device may send the first feature distribution information to the K terminal devices.
  • S707 The K terminal devices send the first feature distribution information to the first server.
  • the first server After receiving the first feature distribution information, the first server determines that the difference between the first feature distribution information and the second feature distribution information satisfies the first condition.
  • the first server sends fourth request information to the K terminal devices.
  • the fourth request information is used to request the K terminal devices to collect data in the first area and within the second time period, and report the collected data.
  • K terminal devices collect data in the first area and within the second time period, and report the collected data to the first server.
  • the first server generates the first AI model based on the first feature distribution information and the first training data.
  • the first training data includes data collected by K terminal devices in the first area and within the second time period.
  • the first server sends the configuration information of the first AI model to the K terminal devices.
  • the terminal devices that send the first feature distribution information to the first server and the terminal devices that send data to the first server are K terminal devices, and the specific implementation is not limited thereto.
  • Figure 8 is a schematic flowchart corresponding to the communication method provided in Embodiment 4 of the present application.
  • the method is illustrated by taking the terminal device and the network device as the execution subjects of the interaction gesture as an example, but this application does not limit the execution subjects of the interaction gesture.
  • the terminal device in Figure 8 can also be a chip, chip system, or processor that supports the terminal device to implement the method;
  • the network device in Figure 8 can also be a chip, chip that supports the access network device to implement the method
  • the system or processor can also be a logical module or software that can realize all or part of the functions of the access network equipment.
  • the method includes the following steps:
  • the first terminal device sends first information to the network device.
  • the first information includes at least one of the following: characteristic information of the P1 group data and characteristic distribution information of the P1 group data.
  • the P1 group data indicates that the first terminal device is located in the first area. And the data collected during the first time period.
  • the above method may also include: the network device may send the first request information to the terminal device located in the first area, and the first request information is used to request the terminal device located in the first area to perform the operation in the first time period. Collect data within the system and report the characteristic information and/or characteristic distribution information of the collected data.
  • the terminal device located in the first area includes the first terminal device, and the first terminal device can receive the first request information.
  • the terminal device can also send indication information to the network device, and the indication information indicates at least one of the terminal configuration information, trigger information, quantity information, area information and time period information corresponding to the P1 group data,
  • the indication information includes the above-mentioned area information and time period information.
  • the network device determines the first feature distribution information based on the first information.
  • the network device can receive the first information reported by the first terminal device.
  • the network device can also receive the information reported by other terminal devices, such as the second information reported by the second terminal device and the third information reported by the third terminal device.
  • Three messages The second information includes at least one of the following: characteristic information of the P2 group data and characteristic distribution information of the P2 group data.
  • the P2 group data is data collected by the second terminal device in the first area and within the first time period.
  • the third information includes at least one of the following: characteristic information of the P3 group data and characteristic distribution information of the P3 group data.
  • the P3 group data is the data collected by the third terminal device located in the first area and within the first time period. P3 is positive integer.
  • the network device after the network device receives the information reported by multiple terminal devices, it can aggregate and obtain the first feature distribution information.
  • the first characteristic distribution information is used to indicate the characteristic distribution of data collected by the terminal device located in the first area within the first time period. For example, "data collected by the terminal device located in the first area during the first time period" includes group P1 data, group P2 data and group P3 data, then the first feature distribution information is used to indicate group P1 data, group P2 data and The characteristic distribution of the P3 group of data, or in other words, the first characteristic distribution information is used to indicate the characteristic distribution of the third training data. For the third training data, see below.
  • the network device determines that the difference between the first feature distribution information and the second feature distribution information satisfies the first condition.
  • the second feature distribution information is obtained in advance by the network device.
  • the second feature distribution information is used to indicate the feature distribution of the training data of the fourth AI model, which is the AI model currently used by the terminal device located in the first area.
  • the network device determines that the difference between the first feature distribution information and the second feature distribution information meets the first condition, it means that the fourth AI model is not suitable for the current data, that is, the performance of the current fourth AI model is poor. If the network device determines that the difference between the first feature distribution information and the second feature distribution information does not meet the first condition, it means that the fourth AI model can still be applied to the current data.
  • the network device may execute S804-a (ie, scenario a), or execute S804-b (ie, scenario b), or execute S804-c and S805-c (i.e. case c).
  • the network device sends deactivation information to the terminal device (such as the first terminal device) located in the first area.
  • the deactivation information is used to indicate deactivation of the AI model or AI mode; accordingly, the terminal device located in the first area Receive deactivation information.
  • the deactivation information is used to indicate deactivation of the AI model or AI mode, which can be understood as: no longer using the AI model or exiting the AI mode.
  • the first terminal device uses the second AI model to encode the downlink channel status information; then after receiving the deactivation information, the first terminal device can use traditional methods to encode the downlink channel status information. Encoding is performed instead of using the AI model to encode downlink channel state information.
  • the network device sends switching information to the terminal device (such as the first terminal device) located in the first area.
  • the switching information is used to instruct to switch the used AI model to the target AI model; accordingly, the terminal device in the first area Receive switching instructions.
  • the switching information may include at least one of the following: identification information of the target AI model and configuration information of the target AI model.
  • the configuration information of the target AI model may refer to the description of the configuration information of the first AI model above.
  • the network device has sent the configuration information of multiple AI models (such as AI model 1, AI model 2, and AI model 3) to the first terminal device in advance.
  • feature distribution information 1 is used to indicate the feature distribution of the training data of AI model 1
  • feature distribution information 2 is used to indicate the feature distribution of the training data of AI model 2
  • feature distribution information 3 is used to indicate the feature distribution of the training data of AI model 3.
  • the AI model currently used by the first terminal device is AI model 1.
  • the network device determines that the difference between the first characteristic distribution information and the characteristic distribution information 1 satisfies the first condition, it can use the characteristic distribution information 2 and the characteristic distribution information 3 respectively.
  • the difference from the first feature distribution information selects the target AI model from AI model 2 and AI model 3.
  • AI model 2 can be selected as the target AI model.
  • the switching information may include the identification information of AI model 2.
  • the network device generates the third AI model based on the third training data.
  • the third training data is described here.
  • the network device can send the fifth request information to the terminal device located in the first area.
  • the information is used to request the terminal device to report data that the terminal device is located in the first area and collected within the first time period. That is, the network device can request the terminal device to report data collected previously by the terminal device.
  • the terminal device located in the first area includes a first terminal device, a second terminal device and a third terminal device
  • the first terminal device can send the P1 group of data to the network device
  • the second terminal device can send the P2 group of data to the network device.
  • the third terminal device can send the P3 group of data to the network device; in this case, the third training data includes the P1 group of data, the P2 group of data and the P3 group of data.
  • the P1 group data may also be included in the first information.
  • the network device does not need to send the fifth request information.
  • the terminal device located in the first area can send the fifth request information.
  • the information is used to request the terminal device to report data collected during the second time period when the terminal device is located in the first area.
  • the duration of the second time period is the same as that of the first time period
  • the start time of the second time period is located after the end time of the first time period. That is, the network device can request the terminal device to collect data again within the second time period and report the collected data.
  • the time interval between the start time of the second time period and the end time of the first time period is small.
  • the terminal device is located in the first area and the data collected in the first time period
  • the characteristic distribution is the same as the data collected during the second time period when the terminal device is located in the first area. That is to say, it can be considered that the first characteristic distribution information is also used to indicate the characteristic distribution of data collected by the terminal device located in the first area in the second time period.
  • the terminal equipment located in the first area includes a first terminal equipment, a second terminal equipment and a third terminal equipment; the first terminal equipment collects the Q1 group of data in the first area and within the second time period, and reports it to the network equipment; the second terminal equipment collects the Q2 group of data in the first area and within the second time period, and reports it to the network equipment; the third terminal equipment collects the Q3 group of data in the first area and within the second time period, and reports it to network equipment.
  • the third training data includes the Q1 group data, the Q2 group data and the Q3 group data.
  • Q1, Q2 and 3 are all positive integers.
  • the network device sends the configuration information of the third AI model to the terminal device (such as the first terminal device) located in the first area; accordingly, the terminal device located in the first area can receive the configuration information of the third AI model.
  • the configuration information of the third AI model may refer to the description of the configuration information of the first AI model in Embodiment 1.
  • the first terminal device after receiving the configuration information of the third AI model, the first terminal device can use the third AI model in the first area. AI model instead of using the fourth AI model.
  • the network device can determine the performance of the current AI model based on the first feature distribution information and the second feature distribution information, thereby being able to easily and quickly identify whether the performance of the AI model is poor. And, when the performance of the AI model is poor, the network device can take a series of solution measures; for example, when the performance of the AI model is poor, the network device can update the AI model.
  • the network device when the network device determines that the AI model needs to be updated, it sends the fifth request information to the first terminal device, and then the first terminal device reports the P1 group data to the network device based on the fifth request; also That is to say, when the network device does not need to update the AI model, the first terminal device does not need to send the P1 group of data to the network device, thereby effectively saving transmission resources.
  • Embodiments 1 to 4 can be referred to each other; in addition, in the same embodiment, , different implementations or different examples or different steps can also be referred to each other.
  • the base station can obtain the first feature distribution information based on the received feature information and/or feature distribution information.
  • the base station can obtain the first feature distribution information based on the received feature information and/or feature distribution information.
  • two methods of determining the first characteristic distribution information are provided here.
  • Method 1 Multiple base stations can respectively send the received feature information and/or feature distribution information to the core network device, and then the core network device can aggregate the first feature distribution information and send the first feature distribution information to the core network device. Multiple base stations.
  • Method 2 Multiple base stations include the first base station. Other base stations among the multiple base stations except the first base station can respectively send the received feature information and/or feature distribution information to the first base station, and then the first base station can summarize The first characteristic distribution information is obtained, and the first characteristic distribution information is sent to other base stations.
  • multiple cells included in the first cell belong to multiple base stations can be implemented with reference to the implementation of “multiple cells included in the first cell belong to the same base station", which will not be described again.
  • network device generates AI model
  • the network device can send the training data to the server corresponding to the network device, and then the network device can The server generates an AI model based on the training data and sends it to the network device.
  • feature distribution information of multiple sets of data can also be replaced by "distribution information of multiple sets of data”.
  • the characteristic distribution information of multiple groups of data is used to indicate the characteristic distribution of multiple groups of data
  • the distribution information of multiple groups of data is used to indicate the distribution of multiple groups of data.
  • the distribution of multiple sets of data can represent the amount or proportion of data that belongs to each of multiple categories. The difference between the feature distribution of multiple sets of data and the distribution of multiple sets of data is that the feature distribution of multiple sets of data determines the distribution of data based on extracted features, while the distribution of multiple sets of data determines the distribution of data based on the data itself. .
  • the data reported by the terminal device to the network device or the first server may be unpreprocessed data collected by the terminal device, or may also be preprocessed data collected by the terminal device.
  • the method of preprocessing is not limited.
  • "generating an AI model based on the training data and the first feature distribution information” may include: fine-tuning the existing AI model based on the training data and the first feature distribution information. New AI model. Similarities can be dealt with by reference.
  • each device may include a corresponding hardware structure and/or software module to perform each function.
  • each device may include a corresponding hardware structure and/or software module to perform each function.
  • the embodiments of the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is performed by hardware or computer software driving the hardware depends on the specific application and design constraints of the technical solution. Skilled artisans may implement the described functionality using different methods for each specific application, but such implementations should not be considered beyond the scope of this application.
  • Embodiments of the present application can divide network equipment, terminal equipment and servers into functional units according to the above method examples.
  • each functional unit can be divided corresponding to each function, or two or more functions can be integrated into one unit.
  • the above integrated units can be implemented in the form of hardware or software functional units.
  • Figure 9 shows a possible exemplary block diagram of the device involved in the embodiment of the present application.
  • the device 900 may include: a processing unit 902 and a communication unit 903.
  • the processing unit 902 is used to process the actions of the device 900 Control management.
  • the communication unit 903 is used to support communication between the device 900 and other devices.
  • the communication unit 903 is also called a transceiver unit and may include a receiving unit and/or a sending unit, respectively configured to perform receiving and sending operations.
  • the device 900 may also include a storage unit 901 for storing program codes and/or data of the device 900 .
  • the device 900 may be the first terminal device in the above embodiment.
  • the processing unit 902 can support the apparatus 900 to perform the actions of the first terminal device in the above method examples.
  • the processing unit 902 mainly performs internal actions of the first terminal device in the method example, and the communication unit 903 may support communication between the device 900 and other devices.
  • the communication unit 903 is configured to: send first information to the network device, where the first information includes at least one of the following: characteristic information of the P1 group data, characteristic distribution information of the P1 group data,
  • the P1 group of data is the data collected by the first terminal device in the first area and within the first time period, and P1 is a positive integer; and, receiving the first feature distribution information from the network device, the first characteristic distribution information is received.
  • a characteristic distribution information is used to indicate the characteristic distribution of data collected by a terminal device located in the first area in the first time period, the terminal device including the first terminal device; and, sending to the first server
  • the first feature distribution information is associated with the first server and the first terminal device.
  • the device 900 may be the network device in the above embodiment.
  • the processing unit 902 can support the apparatus 900 to perform the actions of the network device in each of the above method examples.
  • the processing unit 902 mainly performs internal actions of the network device in the method example, and the communication unit 903 may support communication between the apparatus 900 and other devices.
  • the communication unit 903 is configured to: receive first information from the first terminal device, where the first information includes at least one of the following: characteristic information of the P1 group data, characteristics of the P1 group data Distribution information, the P1 group of data is the data collected by the first terminal device in the first area and within the first time period, P1 is a positive integer; and, sending the first characteristic distribution information to the first terminal device , the first characteristic distribution information is used to indicate the characteristic distribution of data collected by a terminal device located in the first area in a first time period, and the terminal device includes the first terminal device.
  • the device 900 may be the first server in the above embodiment.
  • the processing unit 902 can support the device 900 to perform the actions of the first server in each of the above method examples.
  • the processing unit 902 mainly performs internal actions of the first server in the method example, and the communication unit 903 may support communication between the device 900 and other devices.
  • the communication unit 903 is configured to: receive first feature distribution information from a first terminal device, where the first feature distribution information is used to indicate that the terminal device located in the first area is within the first time period. Feature distribution of the collected data, the first server is associated with the first terminal device; the processing unit 902 is used to: generate an AI model according to the first feature distribution information; and the communication unit 903 is also used to: Send the configuration information of the AI model to the first terminal device.
  • each unit in the device can be a separate processing element, or it can be integrated and implemented in a certain chip of the device.
  • it can also be stored in the memory in the form of a program, and a certain processing element of the device can call and execute the unit. Function.
  • all or part of these units can be integrated together or implemented independently.
  • the processing element described here can also be a processor, which can be an integrated circuit with signal processing capabilities.
  • each operation of the above method or each unit above can be implemented by an integrated logic circuit of hardware in the processor element or implemented in the form of software calling through the processing element.
  • 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 (ASIC), or one or Multiple microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.
  • ASIC application specific integrated circuits
  • DSPs digital signal processors
  • FPGAs field programmable gate arrays
  • 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 central processing unit (CPU), or other processors that can call programs.
  • these units can be integrated together and implemented in the form of a system-on-a-chip (SOC).
  • SOC system-on-a-chip
  • the above receiving unit is an interface circuit of the device and is used to receive signals from other devices.
  • the receiving unit is an interface circuit used by the chip to receive signals from other chips or devices.
  • the above unit used for sending is an interface circuit of the device and is used to send signals to other devices.
  • the sending unit is an interface circuit used by the chip to send signals to other chips or devices.
  • the terminal device can be applied in the communication system shown in Figure 1 to implement the operations of the terminal device in the above embodiment.
  • the terminal device includes: an antenna 1010, a radio frequency part 1020, and a signal processing part 1030.
  • the antenna 1010 is connected to the radio frequency part 1020.
  • the radio frequency section 1020 The information sent by the access network device is received through the antenna 1010, and the information sent by the access network device is sent to the signal processing part 1030 for processing.
  • the signal processing part 1030 processes the information of the terminal device and sends it to the radio frequency part 1020.
  • the radio frequency part 1020 processes the information of the terminal device and sends it to the access network device through the antenna 1010.
  • the signal processing part 1030 may include a modulation and demodulation subsystem for processing each communication protocol layer of data; it may also include a central processing subsystem for processing the operating system and application layer of the terminal device; in addition, it may It includes other subsystems, such as multimedia subsystem, peripheral subsystem, etc.
  • the multimedia subsystem is used to control the camera, screen display, etc. of the terminal device, and the peripheral subsystem is used to realize the connection with other devices.
  • the modem subsystem can be a separately configured chip.
  • the modem subsystem may include one or more processing elements 1031, including, for example, a host CPU and other integrated circuits.
  • the modem subsystem may also include a storage element 1032 and an interface circuit 1033.
  • the storage element 1032 is used to store data and programs, but the program used to execute the method performed by the terminal device in the above method may not be stored in the storage element 1032, but is stored in a memory outside the modem subsystem.
  • the modem subsystem is loaded and used when used.
  • Interface circuit 1033 is used to communicate with other subsystems.
  • the modulation and demodulation subsystem can be implemented by a chip, which includes at least one processing element and an interface circuit, wherein the processing element is used to perform various steps of any method performed by the above terminal equipment, and the interface circuit is used to communicate with other devices.
  • the unit for the terminal device to implement each step in the above method can be implemented in the form of a processing element scheduler.
  • the device for the terminal device includes a processing element and a storage element, and the processing element calls a program stored in the storage element to Execute the method executed by the terminal device in the above method embodiment.
  • the storage element may be a storage element on the same chip as the processing element, that is, an on-chip storage element.
  • the program for executing the method performed by the terminal device in the above method may be in a storage element on a different chip from the processing element, that is, an off-chip storage element.
  • the processing element calls from the off-chip storage element or loads the program on the on-chip storage element to call and execute the method executed by the terminal device in the above method embodiment.
  • the unit of the terminal device that implements each step in the above method may be configured as one or more processing elements. These processing elements are provided on the modulation and demodulation subsystem.
  • the processing elements here may be integrated circuits. For example: one or more ASICs, or one or more DSPs, or one or more FPGAs, or a combination of these types of integrated circuits. These integrated circuits can be integrated together to form a chip.
  • the units of the terminal device that implement each step in the above method can be integrated together and implemented in the form of a SOC.
  • the SOC chip is used to implement the above method.
  • the chip can integrate at least one processing element and a storage element, and the processing element calls the stored program of the storage element to implement the above method executed by the terminal device; or, the chip can integrate at least one integrated circuit to implement the above terminal device.
  • the method of device execution; or, the above implementation methods can be combined, and the functions of some units are realized in the form of processing components calling programs, and the functions of some units are realized in the form of integrated circuits.
  • the above apparatus for a terminal device may include at least one processing element and an interface circuit, wherein at least one processing element is used to execute any method performed by the terminal device provided in the above method embodiments.
  • the processing element can execute part or all of the steps executed by the terminal device in the first way: that is, by calling the program stored in the storage element; or it can also use the second way: that is, by combining the instructions with the integrated logic circuit of the hardware in the processor element. method to perform part or all of the steps performed by the terminal device; of course, the first method and the second method may also be combined to perform part or all of the steps performed by the terminal device.
  • the processing elements here are the same as described above and can be implemented by a processor.
  • the functions of the processing elements can be the same as the functions of the processing unit described in FIG. 9 .
  • the processing element may be a general-purpose processor, such as a CPU, or one or more integrated circuits configured to implement the above method, such as: one or more ASICs, or one or more microprocessors DSP , or, one or more FPGAs, etc., or a combination of at least two of these integrated circuit forms.
  • the storage element can be implemented by a memory, and the function of the storage element can be the same as the function of the storage unit described in FIG. 9 .
  • the storage element can be one memory or a collective name for multiple memories.
  • the terminal device shown in Figure 10 can implement various processes related to the terminal device in the above method embodiment.
  • the operations and/or functions of each module in the terminal device shown in Figure 10 are respectively intended to implement the corresponding processes in the above method embodiment.
  • network device 110 may include one or more DUs 1101 and one or more CUs 1102.
  • the DU 1101 may include at least one antenna 11011, at least one radio frequency unit 11012, at least one processor 11013 and at least one memory 11014.
  • the DU 1101 part is mainly used for transmitting and receiving radio frequency signals, converting radio frequency signals and baseband signals, and performing partial baseband processing.
  • CU 1102 may include at least one processor 11022 and at least one memory 11021.
  • the CU 1102 part is mainly used for baseband processing, control of network equipment, etc.
  • the DU 1101 and the CU 1102 can be physically set together or physically separated, that is, a distributed base station.
  • the CU 1102 is the control center of the network device, which can also be called a processing unit, and is mainly used to complete the baseband processing function.
  • the CU 1102 can be used to control the network device to perform the operation process of the network device in the above method embodiment.
  • the network device 110 may include one or more radio frequency units, one or more DUs, and one or more CUs.
  • the DU may include at least one processor 11013 and at least one memory 11014
  • the radio frequency unit may include at least one antenna 11011 and at least one radio frequency unit 11012
  • the CU may include at least one processor 11022 and at least one memory 11021.
  • the CU1102 can be composed of one or more single boards. Multiple single boards can jointly support a wireless access network (such as a 5G network) with a single access indication, or can respectively support wireless access networks of different access standards. Access network (such as LTE network, 5G network or other networks).
  • the memory 11021 and processor 11022 may serve one or more single boards. In other words, the memory and processor can be set independently on each board. It is also possible for multiple boards to share the same memory and processor. In addition, necessary circuits can also be installed on each board.
  • the DU1101 can be composed of one or more single boards.
  • Multiple single boards can jointly support a wireless access network with a single access indication (such as a 5G network), or can respectively support wireless access networks of different access standards (such as a 5G network).
  • a single access indication such as a 5G network
  • the memory 11014 and processor 11013 may serve one or more single boards. In other words, the memory and processor can be set independently on each board. It is also possible for multiple boards to share the same memory and processor. In addition, necessary circuits can also be installed on each board.
  • the network device shown in Figure 11 can implement various processes involving the network device in the above method embodiment.
  • the operations and/or functions of each module in the network device shown in Figure 11 are respectively intended to implement the corresponding processes in the above method embodiment.
  • the server 1200 may include a processor 1201 , a memory 1202 and an interface circuit 1203 .
  • the processor 1201 can be used to process communication protocols and communication data, and control communication devices.
  • the memory 1202 can be used to store programs and data, and the processor 1201 can execute the method executed by the first server in the embodiment of the present application based on the program.
  • the interface circuit 1203 can be used for the server 1200 to communicate with other devices.
  • the communication can be wired communication or wireless communication.
  • the interface circuit can be, for example, a service-based communication interface.
  • the above memory 1202 may also be externally connected to the server 1200.
  • the server 1200 may include an interface circuit 1203 and a processor 1201.
  • the above interface circuit 1203 may also be externally connected to the server 1200.
  • the server 1200 may include a memory 1202 and a processor 1201.
  • the server 1200 may include a processor 1201.
  • the server shown in Figure 12 can implement each process involving the first server in the above method embodiment.
  • the operations and/or functions of each module in the server shown in Figure 12 are respectively intended to implement the corresponding processes in the above method embodiment.
  • system and “network” in the embodiments of this application may be used interchangeably.
  • “At least one” means one or more, and “plurality” means two or more.
  • “And/or” describes the relationship between associated objects, indicating that there can be three relationships, for example, A and/or B, which can mean: A alone exists, A and B exist simultaneously, and B alone exists, where A, B can be singular or plural.
  • the character “/” generally indicates that the related objects are in an “or” relationship.
  • “At least one of the following” or similar expressions thereof refers to any combination of these items, including any combination of a single item (items) or a plurality of items (items).
  • At least one of A, B, and C includes A, B, C, AB, AC, BC, or ABC.
  • the ordinal numbers such as “first” and “second” mentioned in the embodiments of this application are used to distinguish multiple objects and are not used to limit the order, timing, priority or importance of multiple objects. degree.
  • embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment that combines software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) having computer-usable program code embodied therein.
  • a computer-usable storage media including, but not limited to, disk storage, optical storage, etc.
  • These computer program instructions may also be stored in a computer-readable memory that causes a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction means, the instructions
  • the device implements the functions specified in a process or processes of the flowchart and/or a block or blocks of the block diagram.
  • These computer program instructions may also be loaded onto a computer or other programmable data processing device, causing a series of operating steps to be performed on the computer or other programmable device to produce computer-implemented processing, thereby executing on the computer or other programmable device.
  • Instructions provide steps for implementing the functions specified in a process or processes of a flowchart diagram and/or a block or blocks of a block diagram.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请涉及通信技术领域。具体公开了一种通信方法、装置及系统。该方法包括：第一终端设备从网络设备获取第一特征分布信息，并向第一终端设备关联的第一服务器发送第一特征分布信息。其中第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布。如此，第一服务器接收到第一特征分布信息后，能将第一特征分布信息作为生成AI模型的辅助信息，从而使得第一服务器根据训练数据生成的AI模型接近于根据第一区域内总的训练数据生成的AI模型，进而有效提高AI模型的性能。

Description

一种通信方法、装置及系统
相关申请的交叉引用
本申请要求在2022年07月12日提交中国专利局、申请号为202210820351.1、申请名称为“一种数据特征分布处理方法与装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中；本申请要求在2022年08月15日提交中国专利局、申请号为202210977688.3、申请名称为“一种通信方法、装置及系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术领域，尤其涉及一种通信方法、装置及系统。
背景技术
在无线通信系统中，两个设备之间往往会传输一些测量信息、状态信息等，以此实现后续的信号传输，这样可以提高通信系统性能。例如，两个设备分别为网络设备和终端设备，终端设备可以使用人工智能(artificial intelligence，AI)模型(比如AI编码器)对待传输的信息(例如上述测量信息、状态信息)进行编码，并将编码后的信息反馈给网络设备，网络设备通过和AI编码器对应的AI解码器解码获得终端设备上报的信息(例如上述测量信息、状态信息)。
其中，终端设备在特定区域内使用的AI模型是由终端设备关联的服务器训练生成的，AI模型的训练数据包括服务器关联的多个终端设备在特定区域内采集的数据。
发明内容
然而，由于特定区域内除了包括服务器关联的多个终端设备以外，还包括其它服务器关联的其它终端设备；也就是说，AI模型的训练数据为特定区域内的部分终端设备采集的数据。因此，AI模型的训练数据的特征分布可能与特定区域内总的训练数据的特征分布不同，从而会影响AI模型的性能，比如待传输的信息经过AI编码和AI解码之后可能会失真。
本申请提供了一种通信方法、装置及系统，用于解决AI模型的训练数据的特征分布与特定区域内总的训练数据的特征分布不同，从而影响AI模型的性能的问题。
第一方面，本申请实施例提供一种通信系统。该通信系统包括网络设备、第一终端设备和第一服务器，第一终端设备与第一服务器相关联。其中，网络设备用于，向第一终端设备发送第一特征分布信息，第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布；第一终端设备还用于，接收第一特征分布信息，向第一服务器发送第一特征分布信息；第一服务器用于，接收第一特征分布信息，根据所述第一特征分布信息和训练数据，生成人工智能AI模型；以及，向第一终端设备或与第一服务器关联的第二终端设备发送该AI模型的配置信息。
采用上述方法，由于网络设备将第一特征分布信息发送给第一终端设备，以及第一终端设备将第一特征分布信息发送给第一终端设备关联的第一服务器。如此，虽然第一服务器生成AI模型所使用的第一训练数据是第一服务器关联的终端设备在第一区域内采集的数据，第一训练数据的特征分布可能不同于第一区域内总的训练数据的特征分布，但由于第一服务器可以获取到第一特征分布信息(用于指示第一区域内总的训练数据的特征分布)，因此，第一服务器可以将第一特征分布信息作为生成AI模型的辅助信息，从而使得根据第一训练数据生成的AI模型接近于根据第一区域内总的训练数据生成的AI模型，能够有效提高AI模型的性能。
在一种可能的设计中，第一终端设备用于，向网络设备发送第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；终端设备包括第一终端设备；网络设备还用于，根据第一信息，确定第一特征分布信息。
在一种可能的设计中，网络设备还用于，发送第一请求信息，第一请求信息用于请求终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息；第一终端设备在向网络设备发送第一信息之前，还用于接收第一请求信息。
在一种可能的设计中，第一请求信息包括以下至少一项：第一指示信息，第一指示信息指示请求终 端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，所述第二指示信息指示第一方式和/或第二方式，所述第一方式用于确定所述终端设备采集的数据的特征信息，所述第二方式用于确定所述终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发，触发配置包括以下至少一项：触发事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于所述第一区域；时间段信息，时间段信息用于指示第一时间段。如此，第一终端设备可以网络设备的请求，向网络设备上报第一信息，从而便于网络设备对终端设备的灵活控制。
在一种可能的设计中，第一终端设备还用于，向网络设备发送第二请求信息，第二请求信息用于请求向第一终端设备发送第一特征分布信息；网络设备在向第一终端设备发送第一特征分布信息之前，还用于接收第二请求信息。
在一种可能的设计中，第一服务器还用于，向第一终端设备发送第三请求信息，第三请求信息用于请求向第一服务器发送第一特征分布信息；第一终端设备在向第一服务器发送第一特征分布信息之前，还用于接收第三请求信息。
在一种可能的设计中，第一服务器在根据第一特征分布信息和训练数据，生成AI模型之前，还用于确定第一特征分布信息和第二特征分布信息的差异满足第一条件，第二特征分布信息为第一服务器预先获取的。如此，第一服务器可以依据第一特征分布信息和第二特征分布信息来判断AI模型是否需要更新，从而能够较为简便快捷地识别出AI模型是否需要更新，以便于更合理地对AI模型进行更新，避免频繁更新AI模型导致第一服务器的处理负担较重，或者长时间未更新AI模型导致AI模型的性能较差。
在一种可能的设计中，第一服务器还用于接收来自第一终端设备的P1组数据。训练数据包括P1组数据，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数。
在一种可能的设计中，第一服务器还用于接收来自第二终端设备的P2组数据。训练数据包括P2组数据，P2组数据为第二终端设备位于第一区域且在第一时间段内采集的数据，P2为正整数。
在一种可能的设计中，训练数据包括第一服务器预先获取的数据。
第二方面，本申请实施例提供一种通信方法。该方法可以应用于第一终端设备。在该方法中，第一终端设备接收来自网络设备的第一特征分布信息，第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布；向第一服务器发送第一特征分布信息，第一服务器与第一终端设备相关联。
在一种可能的设计中，向网络设备发送第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；其中，终端设备包括第一终端设备，第一特征分布信息根据第一信息确定。
在一种可能的设计中，在向网络设备发送第一信息之前，该方法还包括：接收来自网络设备的第一请求信息，第一请求信息用于请求终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
在一种可能的设计中，第一请求信息包括以下至少一项：第一指示信息，第一指示信息指示请求终端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，第二指示信息指示第一方式和/或第二方式，第一方式用于确定终端设备采集的数据的特征信息，第二方式用于确定终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发，触发配置包括以下至少一项：触发事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于指示第一区域；时间段信息，时间段信息用于指示第一时间段。
在一种可能的设计中，在接收来自网络设备的第一特征分布信息前，该方法还包括：向网络设备发送第二请求信息，第二请求信息用于请求向第一终端设备发送第一特征分布信息。
在一种可能的设计中，在向网络设备发送第二请求信息前，该方法还包括：接收来自第一服务器的第三请求信息，第三请求信息用于请求向第一服务器发送第一特征分布信息。
第三方面，本申请实施例提供一种通信方法。该方法可以应用于网络设备。在该方法中，网络设备 确定第一特征分布信息，并向第一终端设备发送第一特征分布信息，第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布。
在一种可能的设计中，该方法还包括：接收来自第一终端设备的第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；终端设备包括第一终端设备；确定第一特征分布信息，包括：根据第一信息，确定第一特征分布信息。
在一种可能的设计中，在接收来自第一终端设备的第一信息之前，该方法还包括：向终端设备发送第一请求信息，第一请求信息用于请求终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
在一种可能的设计中，第一请求信息包括以下至少一项：第一指示信息，第一指示信息指示请求终端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，第二指示信息指示第一方式和/或第二方式，第一方式用于确定终端设备采集的数据的特征信息，第二方式用于确定终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发，触发配置包括以下至少一项：触发事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于指示第一区域；时间段信息，时间段信息用于指示第一时间段。
在一种可能的设计中，在向第一终端设备发送第一特征分布信息之前，该方法还包括：接收来自第一终端设备的第二请求信息，第二请求信息用于请求向第一终端设备发送第一特征分布信息。
第四方面，本申请实施例提供一种通信方法。该方法可以应用于第一服务器。在该方法中，第一服务器接收来自第一终端设备的第一特征分布信息，第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布，第一服务器与第一终端设备相关联；根据第一特征分布信息，生成AI模型；以及，向第一终端设备或与第一服务器关联的第二终端设备发送AI模型的配置信息。
在一种可能的设计中，该方法还包括：接收来自第一终端设备的P1组数据，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；其中，训练数据包括P1组数据。
在一种可能的设计中，该方法还包括：接收来自第二终端设备的P2组数据，P2组数据为第二终端设备位于第一区域且在第一时间段内采集的数据，P2为正整数；其中，训练数据包括P2组数据。
在一种可能的设计中，训练数据包括第一服务器预先获取的数据。
在一种可能的设计中，在根据第一特征分布信息，生成AI模型之前，该方法还包括：确定第一特征分布信息和第二特征分布信息的差异满足第一条件，第二特征分布信息为第一服务器预先获取的。
在一种可能的设计中，在接收来自第一终端设备的第一特征分布信息之前，该方法还包括：向第一终端设备发送第三请求信息，第三请求信息用于请求向第一服务器发送第一特征分布信息。
可以理解的是，上述第二方面至第四方面所请求保护的方法与第一方面所请求保护的通信系统相对应。因此，第二方面至第四方面中相关技术特征的有益效果可以参照第一方面中的描述，不再赘述。
第五方面，本申请实施例提供一种通信系统。通信系统包括网络设备和第一终端设备。其中第一终端设备用于，向网络设备发送第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数。网络设备用于，接收第一信息。
采用上述方法，终端设备可以向网络设备上报P1组数据的特征信息和/或特征分布信息，从而使得网络设备可以及时获知第一区域内总的训练数据的分布情况，便于对终端设备侧使用的AI模型进行管理。
在一种可能的设计中，网络设备还用于，向第一终端设备发送去激活信息，去激活信息用于指示去激活AI模型或AI模式。终端设备还用于，接收去激活信息。
在一种可能的设计中，网络设备还用于，向第一终端设备发送切换信息，切换信息用于指示将使用的AI模型切换为目标AI模型。终端设备还用于，接收切换信息。
在一种可能的设计中，网络设备还用于，根据训练数据生成AI模型，并向第一终端设备发送AI模型的配置信息。终端设备还用于，接收AI模型的配置信息，并将使用的模型更新为AI模型。
在一种可能的设计中，网络设备还用于，确定第一特征分布信息和第二特征分布信息的差异满足第 一条件，第一特征分布信息用于指示位于第一区域的终端设备且在第一时间段内采集的数据的特征分布，第一特征分布信息根据第一信息确定。终端设备包括第一终端设备，第二特征分布信息为网络设备预先获取的。
采用上述方法，比如第二特征分布信息用于指示当前使用的AI模型的训练数据的特征分布，则网络设备可以依据第一特征分布信息和第二特征分布信息，来判断当前使用的AI模型的性能，从而能够较为简便快捷地识别出AI模型的性能是否较差。
在一种可能的设计中，网络设备还用于，向终端设备发送第一请求信息，第一请求信息用于请求终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。第一终端设备在向网络设备发送第一信息之前，还用于接收第一请求信息。
在一种可能的设计中，第一请求信息包括以下至少一项：第一指示信息，第一指示信息指示请求终端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，第二指示信息指示第一方式和/或第二方式，第一方式用于确定终端设备采集的数据的特征信息，第二方式用于确定终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置。触发方式包括以下至少一项：事件触发、周期触发；触发配置包括以下至少一项：触发事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于指示第一区域；时间段信息，时间段信息用于指示第一时间段。
在一种可能的设计中，第一终端设备还用于，向网络设备发送P1组数据；网络设备还用于，接收P1组数据，训练数据包括P1组数据。
在一种可能的设计中，网络设备还用于，向第一终端设备发送第五请求信息，第五请求信息用于请求第一终端设备上报位于第一区域且在第一时间段内采集的数据；第一终端设备在向网络设备发送P1组数据之前，还用于接收第五请求信息。
采用上述方法，当网络设备确定需要生成AI模型时，向第一终端设备发送第五请求信息，进而第一终端设备基于第五请求向网络设备上报P1组数据。也就是说，当网络设备不需要生成AI模型时，第一终端设备可无需向网络设备发送P1组数据，从而有效节省传输资源。
第六方面，本申请实施例提供一种通信方法。该方法可以应用于网络设备。在该方法中，网络设备接收来自第一终端设备的第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数。
在一种可能的设计中，该方法还包括：向第一终端设备发送去激活信息，去激活信息用于指示去激活AI模型或AI模式。
在一种可能的设计中，该方法还包括：向第一终端设备发送切换信息，切换信息用于指示将使用的AI模型切换为目标AI模型。
在一种可能的设计中，该切换信息包括以下至少一项：目标AI模型的标识信息、目标AI模型的配置信息。
在一种可能的设计中，该方法还包括：根据训练数据生成AI模型，并向第一终端设备发送AI模型的配置信息；AI模型的配置信息用于指示将使用的模型更新为AI模型。
在一种可能的设计中，方法还包括：确定第一特征分布信息和第二特征分布信息的差异满足第一条件，所述第一特征分布信息用于指示位于第一区域的终端设备且在第一时间段内采集的数据的特征分布，所述第一特征分布信息根据所述第一信息确定，所述终端设备包括所述第一终端设备，所述第二特征分布信息为所述网络设备预先获取的。
在一种可能的设计中，所述方法还包括：向所述终端设备发送第一请求信息，所述第一请求信息用于请求所述终端设备在所述第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
在一种可能的设计中，所述第一请求信息包括以下至少一项：第一指示信息，所述第一指示信息指示请求所述终端设备上报的信息为所述终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，所述第二指示信息指示第一方式和/或第二方式，所述第一方式用于确定所述终端设备采集的数据的特征信息，所述第二方式用于确定所述终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发；触发配置包括以下至少一项：触发 事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于指示第一区域；时间段信息，时间段信息用于指示第一时间段。
在一种可能的设计中，在向第一终端设备发送AI模型的配置信息之前，该方法还包括：接收来自第一终端设备的P1组数据，训练数据包括P1组数据。
在一种可能的设计中，该方法还包括：向第一终端设备发送第五请求信息，第五请求信息用于请求第一终端设备上报位于第一区域且在第一时间段内采集的数据。
第七方面，本申请实施例提供一种通信方法。该方法可以应用于第一终端设备。在该方法中，第一终端设备向网络设备发送第一信息。第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数。
在一种可能的设计中，该方法还包括：接收来自网络设备的去激活信息，去激活信息用于指示去激活AI模型或AI模式。
在一种可能的设计中，该方法还包括：接收来自网络设备的切换信息，切换信息用于指示将使用的AI模型切换为目标AI模型。
在一种可能的设计中，切换信息包括以下至少一项：目标AI模型的标识信息、目标AI模型的配置信息。
在一种可能的设计中，该方法还包括：接收来自网络设备的AI模型的配置信息；将使用的模型更新为AI模型。
在一种可能的设计中，在向网络设备发送第一信息之前，该方法还包括：接收来自网络设备的第一请求信息，第一请求信息用于请求终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
在一种可能的设计中，第一请求信息包括以下至少一项：第一指示信息，第一指示信息指示请求终端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息；第二指示信息，第二指示信息指示第一方式和/或第二方式，第一方式用于确定终端设备采集的数据的特征信息，第二方式用于确定终端设备采集的数据的特征分布信息；终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件；触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发；触发配置包括以下至少一项：触发事件、触发周期；数量信息，数量信息用于指示P1的数值；区域信息，区域信息用于指示第一区域；时间段信息，时间段信息用于指示第一时间段。
在一种可能的设计中，该方法还包括：向网络设备发送P1组数据，AI模型的训练数据包括P1组数据。
在一种可能的设计中，该方法还包括：接收来自网络设备的第五请求信息，第五请求信息用于请求第一终端设备上报位于第一区域且在第一时间段内采集的数据。
可以理解的是，上述第六方面和第七方面所请求保护的方法与第五方面所请求保护的通信系统相对应。因此，第六方面和第七方面中相关技术特征的有益效果可以参照第五方面中的描述，不再赘述。
第八方面，本申请提供一种通信装置。通信装置具备实现上述第二方面至第四方面以及第六方面和第七方面涉及的功能。比如，所述通信装置包括执行上述第二方面至第四方面以及第六方面和第七方面涉及操作所对应的模块或单元或手段，所述功能或单元或手段可以通过软件实现，或者通过硬件实现，也可以通过硬件执行相应的软件实现。
在一种可能的设计中，通信装置包括处理器，处理器可以与存储器耦合。存储器可以保存实现上述第二方面至第四方面以及第六方面和第七方面涉及的功能的必要计算机程序或指令。处理器可执行存储器存储的计算机程序或指令，当计算机程序或指令被执行时，使得通信装置实现上述第二方面至第四方面以及第六方面和第七方面任意可能的设计或实现方式中的方法。
在一种可能的设计中，通信装置包括处理器和接口电路。其中，处理器通过接口电路与其它装置通信，并执行上述第二方面至第四方面以及第六方面和第七方面任意可能的设计或实现方式中的方法。
可以理解地，上述第八方面中，处理器可以通过硬件来实现也可以通过软件来实现。当通过硬件实现时，该处理器可以是逻辑电路、集成电路等。当通过软件来实现时，该处理器可以是一个通用处理器，通过读取存储器中存储的软件代码来实现。此外，以上处理器可以为一个或多个，存储器可以为一个或多个。存储器可以与处理器集成在一起，或者存储器与处理器分离设置。在具体实现过程中，存储器可 以与处理器集成在同一块芯片上，也可以分别设置在不同的芯片上，本申请实施例对存储器的类型以及存储器与处理器的设置方式不做限定。
第九方面，本申请提供一种计算机可读存储介质。计算机存储介质中存储有计算机可读指令，当计算机读取并执行计算机可读指令时，使得计算机执行上述第一方面至第四方面的任一种可能的设计中的方法。
第十方面，本申请提供一种计算机程序产品，当计算机读取并执行计算机程序产品时，使得计算机执行上述第一方面至第四方面的任一种可能的设计中的方法。
第十一方面，本申请提供一种芯片，包括处理器，处理器与存储器耦合，用于读取并执行存储器中存储的软件程序，以实现第二方面至第四方面以及第六方面和第七方面的任一种可能的设计中的方法。
附图说明
图1为本申请实施例适用的一种通信系统示意图；
图2为本申请实施例提供的一种直方图统计的示意图；
图3为本申请实施例提供的区域1内包括不同厂商的终端设备示意图；
图4为本申请实施例提供的不同特征分布示意图；
图5为本申请实施例一提供的通信方法所对应的流程示意图；
图6为本申请实施例二提供的通信方法所对应的流程示意图；
图7为本申请实施例三提供的通信方法所对应的流程示意图；
图8为本申请实施例四提供的通信方法所对应的流程示意图；
图9为本申请实施例中所涉及的装置的可能的示例性框图；
图10为本申请实施例提供的一种终端设备的结构示意图；
图11为本申请实施例提供的一种网络设备的结构示意图；
图12为本申请实施例提供的一种服务器的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行描述。本申请实施例中的技术方案可以应用于各种通信系统，例如通用移动通信系统(universal mobile telecommunications system，UMTS)、无线局域网(wireless local area network，WLAN)、无线保真(wireless fidelity，Wi-Fi)系统、全球互联微波接入(worldwide interoperability for microwave access，WiMAX)通信系统、车到任意物体(vehicle to everything，V2X)通信系统、设备间(device-to-device，D2D)通信系统、车联网通信系统、第4代(4th generation，4G)移动通信系统，如长期演进(long term evolution，LTE)系统)、第五代(5th generation，5G)移动通信系统，如新空口(new radio，NR)系统，以及未来演进的通信系统，如第六代(6th generation，6G)移动通信系统等。
本申请将围绕可包括多个设备、组件、模块等的系统来呈现各个方面、实施例或特征。应当理解和明白的是，各个系统可包括另外的设备、组件、模块等，并且/或者可以并不包括结合附图讨论的所有设备、组件、模块等。此外，还可以使用这些方案的组合。
另外，在本申请实施例中，“示例性地”、“比如”等词语用于表示例子、例证或说明。本申请中被描述为“示例”的任何实施例或设计方案不应被解释为比其它实施例或设计方案更优选或更具优势。确切而言，使用“示例”一词旨在以具体方式呈现概念。本申请实施例中，“的(of)”，“相应的(corresponding，relevant)”和“对应的(corresponding)”有时可以混用，应当指出的是，在不强调其区别时，其所要表达的含义是一致的。
本申请实施例描述的网络架构以及业务场景是为了更加清楚的说明本申请实施例的技术方案，并不构成对于本申请实施例提供的技术方案的限定，本领域普通技术人员可知，随着网络架构的演变和新业务场景的出现，本申请实施例提供的技术方案对于类似的技术问题，同样适用。
为便于理解本申请实施例，首先以图1中示出的通信系统为例详细说明适用于本申请实施例的通信系统。如图1所示，该通信系统包括多个终端设备(比如终端设备101、终端设备102和终端设备103)和一个或多个网络设备(比如网络设备110)，可选地，还可以包括一个或多个服务器(比如服务器121、服务器122)。
(1)终端设备
终端设备可以为接入上述通信系统，且具有无线收发功能的终端或可设置于该终端的芯片或芯片系统。该终端设备也可以称为用户设备(user equipment，UE)、用户装置、接入终端、用户单元、用户站、移动站、移动台(mobile station，MS)、远方站、远程终端、移动设备、用户终端、终端、终端单元、终端站、终端装置、无线通信设备、用户代理或用户装置。
例如，本申请的实施例中的终端设备可以是手机(mobile phone)、无线数据卡、个人数字助理(personal digital assistant，PDA)电脑、膝上型电脑(laptop computer)、平板电脑(Pad)、无人机、带无线收发功能的电脑、机器类型通信(machine type communication，MTC)终端、虚拟现实(virtual reality，VR)终端设备、增强现实(augmented reality，AR)终端设备、物联网(internet of things，IoT)终端设备、工业控制(industrial control)中的无线终端、无人驾驶(self driving)中的无线终端、远程医疗(remote medical)中的无线终端、智能电网(smart grid)中的无线终端、运输安全(transportation safety)中的无线终端、智慧城市(smart city)中的无线终端、智慧家庭(smart home)中的无线终端(例如游戏机、智能电视、智能音箱、智能冰箱和健身器材等)、车载终端、具有终端功能的RSU。接入终端可以是蜂窝电话(cellular phone)、无绳电话、会话启动协议(session initiation protocol，SIP)电话、无线本地环路(wireless local loop，WLL)站、个人数字助理(personal digital assistant，PDA)、具有无线通信功能的手持设备(handset)、计算设备或连接到无线调制解调器的其它处理设备、可穿戴设备等。
又例如，本申请实施例中的终端设备可以是智慧物流中的快递终端(例如可监控货物车辆位置的设备、可监控货物温湿度的设备等)、智慧农业中的无线终端(例如可收集禽畜的相关数据的可穿戴设备等)、智慧建筑中的无线终端(例如智慧电梯、消防监测设备、以及智能电表等)、智能医疗中的无线终端(例如可监测人或动物的生理状态的可穿戴设备)、智能交通中的无线终端(例如智能公交车、智能车辆、共享单车、充电桩监测设备、智能红绿灯、以及智能监控以及智能停车设备等)、智能零售中的无线终端(例如自动售货机、自助结账机、以及无人便利店等)。又例如，本申请的终端设备可以是作为一个或多个部件或者单元而内置于车辆的车载模块、车载模组、车载部件、车载芯片或者车载单元，车辆通过内置的所述车载模块、车载模组、车载部件、车载芯片或者车载单元可以实施本申请提供的方法。
(2)网络设备
网络设备可以为接入网设备和核心网网元中的一个设备，或者，网络设备可以为核心网网元中的一个或多个设备与接入网设备的集成的设备。
上述接入网设备为位于上述通信系统的网络侧，且具有无线收发功能的设备或可设置于该设备的芯片或芯片系统。该接入网设备包括但不限于：无线保真(wireless fidelity，Wi-Fi)系统中的接入点(access point，AP)，如家庭网关、路由器、服务器、交换机、网桥等，演进型节点B(evolved Node B，eNB)、无线网络控制器(radio network controller，RNC)、节点B(Node B，NB)、基站控制器(base station controller，BSC)、基站收发台(base transceiver station，BTS)、家庭基站(例如，home evolved NodeB，或home Node B，HNB)、基带单元(baseband unit，BBU)，无线中继节点、无线回传节点、传输点(transmission and reception point，TRP或者transmission point，TP)等，还可以为5G，如，新空口(new radio，NR)系统中的gNB，或，传输点(TRP或TP)，5G系统中的基站的一个或一组(包括多个天线面板)天线面板，或者，还可以为构成gNB或传输点的网络节点，如基带单元(BBU)，或，分布式单元(distributed unit，DU)、具有基站功能的路边单元(road side unit，RSU)等，或者还可以为卫星、或未来各种形式的基站。
上述核心网网元可以包括但不限于如下一项或多项：用户面网元、移动性管理网元、会话管理网元、策略控制网元、存储功能网元。
其中，用户面网元：作为和数据网络的接口，完成用户面数据转发、基于会话/流级的计费统计，带宽限制等功能。即分组路由和转发以及用户面数据的服务质量(quality of service，QoS)处理等。在5G通信系统中，该用户面网元可以是用户面功能(user plane function，UPF)网元。
移动性管理网元主要用于移动性管理和接入管理等。在5G通信系统中，该接入管理网元可以是接入和移动性管理功能(access and mobility management function，AMF)网元，主要进行移动性管理、接入鉴权/授权等功能。此外，移动性管理网元还负责在终端与策略控制功能(policy  control function，PCF)网元间传递用户策略。
会话管理网元：主要用于会话管理(例如创建、删除等)、维护会话上下文及用户面转发管道信息、用户设备的网络互连协议(internet protocol，IP)地址分配和管理、选择可管理用户平面功能、策略控制和收费功能接口的终结点以及下行数据通知等。在5G通信系统中，该会话管理网元可以是会话管理功能(session management function，SMF)网元，完成终端IP地址分配，UPF选择，及计费与QoS策略控制等。
策略控制网元：包括用户签约数据管理功能、策略控制功能、计费策略控制功能、服务质量(quality of service，QoS)控制等，用于指导网络行为的统一策略框架，为控制面功能网元(例如AMF，SMF网元等)提供策略规则信息等。在5G通信系统中，该策略控制网元可以是PCF。
存储功能网元：为其他核心网元提供网络功能实体信息的存储功能和选择功能。在5G通信系统中，该网元可以是网络功能存储库功能(network function repository function，NRF)网元。
(3)服务器
服务器也可以称为在网络上层(over the top，OTT)服务器。服务器可以与多个终端设备相关联；当终端设备与服务器相关联时，终端设备可以与服务器通信。如图1所示，终端设备101和终端设备102与服务器121相关联，终端设备103与服务器122相关联。以终端设备101为例，终端设备101可以与服务器121通信；比如，终端设备101可以向服务器121发送终端设备101采集的数据，终端设备101采集的数据可用于服务器121生成AI模型，以及服务器121可以向终端设备101发送AI模型的配置信息，进而终端设备101可以根据AI模型的配置信息使用AI模型。
其中，服务器与终端设备相关联的场景可以有多种。在一种可能的场景中，图1所示意的通信系统中可以包括多个终端设备，这多个终端设备可能属于不同的终端设备厂商，或者这多个终端设备所使用的芯片可能属于不同的芯片厂商。其中，一个或多个终端设备厂商可构成一个群体(或者说利益体)，或者一个或多个芯片厂商可构成一个群体，又或者一个或多个终端设备厂商与一个或多个芯片厂商可构成一个群体。一个群体可维护一个服务器或者一个服务器集群(一个服务器集群中可以包括多个服务器)，本申请实施例中将以“一个群体维护一个服务器”为例进行描述。一个群体维护的服务器与该群体的终端设备相关联。
举个例子，终端设备101和终端设备102均属于终端设备厂商1，服务器121为终端设备厂商1维护的服务器，因此，终端设备101和终端设备102与服务器121相关联；终端设备103属于终端设备厂商2，服务器122为终端设备厂商2维护的服务器，因此，终端设备103与服务器122相关联。
可以理解的是，本申请实施例提供的通信方法，可以适用于图1所示的服务器、终端设备与网络设备之间，或者也可以适用于终端设备与网络设备之间，具体实现可以参考下述方法实施例。本申请实施例中的技术方案还可以应用于其他通信系统中，相应的名称也可以用其他通信系统中的对应功能的名称进行替代。图1仅为便于理解而示例的简化示意图，该通信系统中还可以包括其它网络设备、其它终端设备、其它服务器中的至少一项，图1中未予示出。
下面先对本申请实施例所涉及的相关内容进行解释说明。需要说明的是，这些解释是为了让本申请实施例更容易被理解，而不应该视为对本申请所要求的保护范围的限定。
(1)一组数据的特征信息
本申请实施例中，一组数据也可以称为一个数据，此处以“一组数据为第一数据”为例进行描述。示例性地，第一数据可以是根据终端设备的下行信道状态信息确定的；或者，第一数据可以是与终端设备的定位有关的数据。本申请实施例不对第一数据进行限定，对应不同的应用场景，第一数据可以不同。
第一数据的特征信息用于指示第一数据的特征。可选地，第一数据的特征可以是采用第一方式对第一数据进行处理得到的。
其中，第一方式可以包括第一处理过程，比如第一处理过程可以包括直方图(histogram)统计。可选地，第一方式还可以包括第二处理过程和/或第三处理过程，第二处理过程为第一处理过程之前的处理过程，第三处理过程为第一处理过程之后的处理过程。
也就是说，第一数据的特征可以是根据第一处理过程获得的，或者，第一数据的特征可以是根据第二处理过程和第一处理过程获得的，或者，第一数据的特征可以是根据第一处理过程和第 三处理过程获得的，或者，第一数据的特征可以是根据第一处理过程、第二处理过程和第三处理过程获得的。
(1.1)第一处理过程
如上所述，第一处理过程可以包括直方图统计。图2为本申请实施例提供的一种直方图统计的示意图。结合图2，假设第一数据为三维数据，例如维度为W×H×D，W和H可以为与第一数据相关的信息，D可以为2，D代表实部和虚部。以单元格P×Q×T为单位遍历第一数据，1≤P≤W，1≤Q≤H，1≤T≤D，遍历过程可以有交叠也可以无交叠，并针对每个单元格P×Q×T进行直方图统计，可以获得直方图数据，得到的所有直方图数据构成第一数据的特征(矩阵或向量)。其中，该直方图数据可以是基于单元格P×Q×T中各元素的数值，也可以是基于该单元格P×Q×T中各元素的方向和强度。可以理解的是，本申请不对第一数据的维度进行限定，图2仅为本申请提供的示例。
在一个示例中，第一数据是根据下行信道状态信息确定的，比如第一数据包括下行信道状态信息。下行信道状态信息可以通过第一维度、第二维度和第三维度表示。可选地，第一维度与载波数量或子带数量对应，第二维度与网络设备的天线端口数量或射频链数量对应，第三维度表示实部和虚部。假设第一维度为W，第二维度为H，第三维度为D，则下行信道状态信息可以为W×H×D，结合图2，以单元格P×Q×T为单位遍历下行信道状态信息，并针对每个单元格P×Q×T进行直方图统计，获得直方图数据，得到的所有直方图数据构成下行信道状态信息的特征(矩阵或向量)。
可以理解的是，在其它可能的实现中，第一处理过程可以包括无线信道特征提取，其中，无线信道特征可能包括但不限于如下特征中的一种或几种：功率时延谱(power delay profile，PDP)、时变多普勒谱、角度功率谱(angular power spectrum，APS)、PDP稀疏度、多普勒谱稀疏度、APS稀疏度。
(1.2)第二处理过程
可选地，第二处理过程可以包括但不限于如下一项或多项：傅里叶变换、快速傅里叶变换(fast fourier transform，FFT)、离散傅里叶变换(discrete fourier transform，DFT)、数据截断、坐标转换、取幅度、降噪、归一化、灰度化、数据矫正(例如伽马矫正)、量化、奇异值分解(singular value decomposition，SVD)、平移、数据扩充、求梯度、特征模板滤波、部分选取、和对多种处理结果进行组合。
其中，求梯度可以包括局部二进制模式算法中的二值化梯度。特征模板滤波可以包括对数据(例如特征)进行索贝尔算子(sobel operators)运算、计算获取哈尔特征(Haar-like features)、和/或通过canny边缘检测算法获取数据中的边缘等。部分选取可以包括：使用尺度不变特征转换等方法提取感兴趣区域(region of interest，ROI)。对多种处理结果进行组合可以包括：对经过多个不同索贝尔算子计算得到的数据进行组合。
需要说明的是，当第二处理过程包括多项时，本申请不对傅里叶变换、FFT、DFT、数据截断、坐标转换、取幅度、降噪、归一化、灰度化、数据矫正(例如伽马矫正)、量化、SVD、平移、数据扩充、求梯度、特征模板滤波、部分选取、和对多种处理结果进行组合的执行顺序进行限定。
(1.3)第三处理过程
可选地，第三处理过程可以包括但不限于如下一项或多项：归一化、特征组合、降低维度、量化、和编码。
其中，归一化可以获得基于单个单元格或多个单元格的直方图数据。特征组合可以包括：通过多种尺度(例如预处理插值或降维生成不同尺度的多个图像，然后分别提取特征)、或不同配置(例如不同尺寸的单元格)得到的直方图特征组合。降低维度可以包括但不限于如下一项或多项：对直方图数据(进行直方图统计后获得的数据)进行组合求和、和采用降低特征维度的方法压缩直方图数据的维度。
可选地，对直方图数据进行组合求和可以得到更少的特征维度。例如，对直方图数据进行组合求和可以包括：可变形部件模型算法中对直方图数据矩阵的行和列分别相加。可选地，降低特征维度的方法可以包括但不限于：主成分分析(principal component analysis，PCA)和线性判别分析(linear discriminant analysis，LDA)。
(1.4)第一方式的具体示例
基于上述第一处理过程、第二处理过程和第三处理过程的描述，在一个示例中，第一方式可以为以下任一项：方向梯度直方图(histogram of oriented gradient，HOG)、局部二进制模式(local binary patterns，LBP)、边缘方向直方图(edge orientation histogram，EOH)、边缘直方图描述符(edge histogram descriptor，EHD)、边缘方向直方图(edge direction histogram，EDH)、可变形部件模型(deformable part model，DPM)、尺度不变特征转换(scale-invariant feature transform，SIFT)或加速鲁棒特征(speed up robust features，SURF)。
其中，HOG方法除了包括第一处理过程，还包括灰度化、求梯度等第二处理过程，和归一化等第三处理过程。类似地，局部二进制模式LBP方法包括第一处理过程，还包括第二处理过程和第三处理过程。本申请不一一阐述。
如此，在对第一数据进行直方图统计之前，采用第二处理过程对第一数据进行处理，和/或，在对第一数据进行直方图统计后，采用第三处理过程对直方图统计的结果进行处理，可以进一步增加第一数据的特征的准确性。
可以理解的是，除采用第一方式外，也可以采用其它可能的方式获得第一数据的特征。比如，可以将第一数据输入AI模型；作为一种实现方式，该AI模型可以是自编码器(Autoencoder)的编码部分，也可以是AI分类器的骨干网络部分。进而，AI模型可以输出第一数据的特征。可选地，在将第一数据输入AI模型之前，可以对第一数据进行预处理，具体的预处理方式不做限定。
(2)多组数据的特征分布信息
参照上述“一组数据”，多组数据可以是根据多组下行信道状态信息确定的，比如多组数据包括多组下行信道状态信息；或者，多组数据是与终端设备的定位有关的数据。在一个示例中，多组数据可以包括上述第一数据，或者说上述第一数据为多组数据中的一组数据。
多组数据的特征分布信息用于指示多组数据的特征分布。其中，多组数据的特征分布可以表示属于多个类别中每个类别的数据的数量或比例。可选地，多组数据的特征分布可以是采用第二方式对多组数据进行处理得到的。下面结合示例1和示例2对第二方式进行描述。
(2.1)示例1
在示例1中，第二方式可以包括第一方式和第三方式。此种情形下，采用第二方式对多组数据进行处理得到多组数据的特征分布，可以是指：采用第一方式对多组数据中的每组数据进行处理，得到多组数据中的每组数据的特征；进一步地，采用第三方式对多组数据的特征进行处理，得到多组数据的特征分布。
示例性地，第三方式可以包括但不限于以下至少一项：针对多组数据的特征中的每个元素在组的维度上取均值、针对多组数据的特征中的每个元素在组的维度上统计得到分位数、针对多组数据的特征中的每个元素在组的维度上进行高斯拟合得到均值和方差、或其它分布的拟合得到其参数值。
以多组数据包括A组数据，比如A组下行信道状态信息为例，当下行信道状态信息通过第一维度、第二维度和第三维度表示时，A组下行信道状态信息可以包括A组W×H×D的数据，其中A为大于1的整数。例如，对A组中的每组数据进行直方图统计，获得A组数据的特征，采用高斯拟合对A组数据特征的每个元素在A的维度上进行处理，获得A组数据的特征分布。
(2.2)示例2
在示例2中，第二方式可以包括第一方式，或者说第二方式与第一方式相同。此种情形下，采用第二方式对多组数据进行处理得到多组数据的特征分布，可以是指：采用第一方式对多组数据进行处理，得到多组数据的特征分布。
以多组数据包括A组数据，比如A组下行信道状态信息为例，上述下行信道状态信息通过第一维度、第二维度和第三维度表示，可以包括：下行信道状态信息通过第一维度、第二维度、第三维度和第四维度表示。假设第一维度用W表示，第二维度用H表示，第三维度用D表示，第四维度用A表示，则A组下行信道状态信息的示例可参照下述方式a或方式b。
方式a，A组下行信道状态信息可以为W×H×(D×A)、或W×(H×A)×D、或(W×A)×H×D的数据，可以将第一维度、第二维度和第三维度中的任一个维度扩展至A倍。
方式b，A组下行信道状态信息可以为W×H×D×A的数据，即A组下行信道状态信息可以通过四维数据表示。
针对于方式a：
当A组下行信道状态信息为W×H×(D×A)、或W×(H×A)×D、或(W×A)×H×D的数据时，根据第一方式获得A组下行信道状态信息的特征分布的实现方式，可参照下行信道状态信息通过第一维度、第二维度和第三维度表示时对应的实现方式。
例如，以下行信道状态信息为W×H×(D×A)的数据为例，以单元格P×Q×T为单位遍历下行信道状态信息，1≤P≤W，1≤Q≤H，1≤T≤D×A，遍历过程可以有交叠也可以无交叠，并针对每个单元格P×Q×T进行直方图统计，获得直方图数据，得到的所有直方图数据构成A组数据的特征分布。
针对于方式b：
当下行信道状态信息为W×H×D×A的数据时，根据第一方式获得A组下行信道状态信息的特征分布的实现方式，可参照下行信道状态信息通过第一维度、第二维度和第三维度表示时对应的实现方式。
例如，以单元格P×Q×T×R为单位遍历下行信道状态信息，1≤P≤W，1≤Q≤H，1≤T≤D，1≤R≤A，遍历过程可以有交叠也可以无交叠，并针对每个单元格P×Q×T×R进行直方图统计，获得直方图数据，得到的所有直方图数据构成A组数据的特征分布。
可选地，上述示例1和示例2中，A组下行信道状态信息可以包括终端设备在一个时间段内的不同时间点采集到的A组下行信道状态信息；比如，终端设备在时间点1采集一组或多组下行信道状态信息，在时间点2采集一组或多组下行信道状态信息，依次类推，在一个时间段内终端设备共采集到A组下行信道状态信息。
在无线通信系统中，应用AI模型可以显著提升通信系统的性能。由于AI模型的训练过程对算力的要求较高，因此，终端设备使用的AI模型往往不是终端设备自身生成的。
针对终端设备使用的AI模型，一种可能的生成方式为：终端设备可以将终端设备位于特定区域(比如区域1)内采集到的数据(可称为训练数据)发送给服务器。其中，由于终端设备采集的数据较为敏感，因此，终端设备是将采集的数据上报给与终端设备相关联的服务器。也就是说，服务器接收的训练数据均为服务器关联的终端设备上报的。进一步地，服务器根据训练数据生成AI模型，并将AI模型的配置信息发送给终端设备，进而终端设备可以根据AI模型的配置信息，在区域1内使用AI模型。
如图3所示，以两家终端设备厂商为例，两家终端设备厂商分别为厂商1和厂商2，厂商1管理服务器1，厂商2管理服务器2。在区域1内，厂商1的终端设备包括m1个终端设备，厂商2的终端设备包括m2个终端设备，m1与m2均为正整数，m1与m2可能相同或不同。m1个终端设备分别向服务器1发送m1个终端设备在区域1内采集的训练数据，进而服务器1可以根据接收的训练数据生成AI模型1；类似地，m2个终端设备分别向服务器2发送m2个终端设备在区域1内采集的训练数据，进而服务器2可以根据接收的训练数据生成AI模型2。
考虑到两家厂商的终端设备在区域1中的分布可能不一样(比如两家厂商的终端设备的数量不同等)，上传到各自服务器的训练数据的特征分布也会不同。比如AI模型1的训练数据的特征分布如图4中的(a)所示，AI模型2的训练数据的特征分布如图4中的(b)所示。进一步地，以区域1内包括m1个终端设备和m2个终端设备为例，参见图4中的(c)所示，为区域1内总的训练数据(包括m1终端设备在区域1内采集的训练数据和m2个终端设备在区域1内采集的训练数据)的特征分布示意图。
根据图4可以看出，AI模型1的训练数据的特征分布不同于总的训练数据的特征分布，从而会导致厂商1的终端设备在区域1内使用AI模型1时，AI模型1的性能较差。比如，厂商1的终端设备在区域1内使用AI模型1对下行信道状态信息进行编码，并发送给网络设备后，网络设备解码出的下行信道状态信息发生失真；又比如，厂商1的终端设备在区域1内使用AI模型获得的定位结果不准确。同样地，AI模型2的训练数据的特征分布也不同于总的训练数据的特征分布不同，从而会导致厂商2的终端设备在区域1内使用AI模型2时，AI模型2的性能较差。
基于此，本申请实施例提供一种通信方法，用于解决AI模型的训练数据的特征分布与特定区域内总的训练数据的特征分布不同，而影响AI模型的性能的问题。
实施例一
图5为本申请实施例一提供的通信方法所对应的流程示意图。图5中以终端设备、网络设备和服务器作为该交互示意的执行主体为例来示意该方法，但本申请并不限制该交互示意的执行主体。例如，图5中的终端设备也可以是支持该终端设备实现该方法的芯片、芯片系统、或处理器；图5中的网络设备也可以是支持该接入网设备实现该方法的芯片、芯片系统、或处理器，还可以是能实现全部或部分接入网设备功能的逻辑模块或软件；图5中的服务器也可以是支持该服务器实现该方法的芯片、芯片系统、或处理器。
如图5所示，该方法包括如下步骤：
S501，第一终端设备向网络设备发送第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数。
本申请实施例中，划分第一区域的方式可以有多种。在一个示例中，第一区域可以包括多个小区。这多个小区可以属于同一个基站，或者也可以属于不同基站。下文中将以“这多个小区属于同一个基站，且网络设备为这多个小区所属的基站”为例进行描述。
示例性地，考虑到终端设备具有移动性，因此，第一终端设备在第一时间段内可能始终位于第一区域，或者第一终端设备也可能会在第一时间段结束之前移出第一区域。若第一终端设备在第一时间段内始终位于第一区域，则第一终端设备可以在第一时间段结束后，根据在第一时间段内采集的P1组数据，获得P1组数据的特征信息和/或特征分布信息，并向网络设备发送第一信息。或者，若第一终端设备在第一时间段结束之前移出第一区域，比如第一时间段的起始时间为时间点t1，结束时间为时间点t2，第一终端设备在时间点t3移出第一区域，时间点t3位于时间点t2之前，则在移出第一区域之前，第一终端设备可以根据在第一时间段内采集的P1组数据(即是指在时间点t1至时间点t3这一段时间内采集的P1组数据)，获得P1组数据的特征信息和/或特征分布信息，并向网络设备发送第一信息。其中，第一终端设备获得P1组数据的特征信息和/或特征分布信息的实现可以参照前文，不再赘述。
可选地，在S501之前，上述方法还可以包括：网络设备可以向位于第一区域的终端设备发送第一请求信息，第一请求信息用于请求位于第一区域的终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。位于第一区域的终端设备包括第一终端设备，进而第一终端设备可以接收到第一请求信息。
其中，网络设备发送第一请求信息的方式可以有多种；比如，网络设备可以通过广播的方式向位于第一区域的终端设备发送第一请求信息。
示例性地，第一请求信息可以包括以下至少一项：
◇第一指示信息，第一指示信息指示请求终端设备上报的信息为终端设备采集的数据的特征信息和/或特征分布信息。
◇第二指示信息，第二指示信息指示第一方式和/或第二方式，第一方式用于确定终端设备采集的数据的特征信息，第二方式用于确定终端设备采集的数据的特征分布信息。其中，第一方式和第二方式的描述可以参照前文。
◇终端配置信息，终端配置信息指示第一终端设备采集P1组数据时应满足的配置条件。示例性地，终端配置信息包括：第一频段的标识，和/或，第一天线数量；此种情形下，第一终端设备采集P1组数据时应满足的配置条件是指：第一终端设备所使用的频段为第一频段，和/或，第一终端设备的天线数量等于第一天线数量。
◇触发信息，触发信息包括触发方式和/或触发配置，触发方式包括以下至少一项：事件触发、周期触发，触发配置包括以下至少一项：触发事件、触发周期。示例性地，触发事件比如为通信系统性能满足某种条件，通信系统性能可能包括以下至少一项：吞吐率、比特误码率(bit error rate，BER)、数据块差错率(block error rate，BLER)、确认回答(acknowledgement，ACK)/否定回答(negative acknowledgement，NACK)比例、重传率等；又比如，触发事件可以为终端设备测量量满足某种条件，测量量可能包括以下至少一项：信号与干扰加噪声比(Signal to Interference plus Noise Ratio，SINR)、参考信号接收功率(Reference Signal Receiving Power，RSRP)、接收信号强度指示(Received Signal Strength Indication，RSSI)、参考信号接收质量(Reference Signal Receiving Quality，RSRQ)、接收信号码功率(Received Signal Code Power，RSCP)。如此，终端设备接收到第一请求信息，可以根据触发信息在第一区域且在第一时间段内采集数据。
◇数量信息，数量信息用于指示P1的数值。
◇区域信息，区域信息用于指示第一区域。比如，第一区域包括多个小区，则区域信息可以包括这多个小区的标识，以多个小区中的小区1为例，小区1的标识可以为小区1的物理小区标识(physical cell identifier，PCI)。
◇时间段信息，时间段信息用于指示第一时间段。可选地，时间段信息可以包括第一时间段的起始时间和结束时间，或者包括第一时间段的起始时间(或结束时间)和第一时间段的时长。比如，第一时间段的起始时间为X1年X2月X3日X4时X5分，结束时间为Y1年Y2月Y3日Y4时Y5分，则时间段信息可以包括X1至X5以及Y1至Y5的取值。
◇第三指示信息，第三指示信息指示终端设备采集的数据的至少一个特征需要满足的条件。其中，以至少一个特征中的一个特征为例，该特征需要满足的条件，可以是指：该特征与某个向量的距离大于或等于阈值时，其中某个向量、阈值、距离的定义应包含在第三指示信息中或提前约定。作为一种实现方法，某个向量可以是当前AI模型的训练集的特征，距离可以是欧式距离、曼哈顿距离、切比雪夫距离、闵可夫斯基距离、余弦相似度、马氏距离、汉明距离等距离定义中的一种。如此，当终端设备采集到一组数据后，若该组数据的至少一个特征不满足上述条件，则终端设备可以确定该组数据为无效数据，上述P1组数据均为有效数据，而不包括无效数据。
可以理解的是，当请求信息不包括第一指示信息时，终端设备上报特征信息还是特征分布信息可以是协议预先定义的，或者预先配置的。当请求信息不包括第二指示信息时，确定特征信息和/或特征分布信息的方式可以是协议预先定义的，或者预先配置的。当请求信息不包括区域信息时，第一区域的范围可以是协议预先定义的，或者预先配置的。当请求信息不包括第一时间段时，比如第一时间段的起始时间可以根据请求信息的接收时间确定，第一时间段的时长可以为协议预先定义的，或者预先配置的。当请求信息不包括第三指示信息时，终端设备可以认为终端设备在第一区域且在第一时间段内采集的多组数据均为有效数据。其它项信息可以参照理解，不再一一赘述。
可选地，在上述S501中，终端设备还可以向网络设备发送指示信息，指示信息指示P1组数据对应的终端配置信息、触发信息、数量信息、区域和时间段中的至少一项，比如指示信息包括上述区域信息和时间段信息。
S502，网络设备向第一终端设备发送第一特征分布信息，相应地，第一终端设备接收第一特征分布信息。
如上所述，网络设备可以接收第一终端设备上报的第一信息，此外，网络设备还可以接收其它终端设备上报的信息，比如第二终端设备上报的第二信息、第三终端设备上报的第三信息。第二信息包括以下至少一项：P2组数据的特征信息、P2组数据的特征分布信息，P2组数据为第二终端设备位于第一区域且在第一时间段内采集的数据，P2为正整数。第三信息包括以下至少一项：P3组数据的特征信息、P3组数据的特征分布信息，P3组数据为第三终端设备位于第一区域且在第一时间段内采集的数据，P3为正整数。
如此，网络设备接收到多个终端设备上报的信息后，可以汇总得到第一特征分布信息，也即，第一特征分布信息根据第一信息(可选地，以及第二信息和第三信息)确定。第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布。其中，位于第一区域的终端设备可以包括不同服务器关联的终端设备，比如位于第一区域的终端设备包括第一服务器关联的终端设备(如第一终端设备、第二终端设备)，第二服务器关联的终端设备(如第三终端设备)。也就是说，第一特征分布信息用于指示第一区域内的总的训练数据的特征分布。
本申请实施例对“网络设备汇总得到第一特征分布信息”的具体实现不做限定。作为一种可能的实现，以网络设备接收到第一信息和第二信息为例，第一信息包括P1组数据的特征信息，第二信息包括P2组数据的特征信息，进而网络设备可以对P1组数据的特征和P2组数据的特征进行处理(比如采用前文所述的第三方式对P1组数据的特征和P2组数据的特征进行处理)，得到第一特征分布信息。
网络设备向第一终端设备发送第一特征分布信息的实现可以有多种，下面结合实现方式1至实现方式3描述三种可能的实现。
(1)实现方式1
在实现方式1中，网络设备可以主动向位于第一区域的终端设备发送第一特征分布信息；比如，网络设备通过广播的方式，向位于第一区域的终端设备发送第一特征分布信息。
(2)实现方式2
在实现方式2中，网络设备可以基于第一终端设备的请求，向第一终端设备发送第一特征分布信息。比如，在S502之前，第一终端设备可以向网络设备发送第二请求信息，第二请求信息用于请求向第一终端设备发送第一特征分布信息；相应地，网络设备接收到第二请求信息后，可以向第一终端设备发送第一特征分布信息。
进一步可选地，在第一终端设备向网络设备发送第二请求信息之前，第一终端设备可以接收来自第一服务器的第三请求信息，第三请求信息用于请求向第一服务器发送第一特征分布信息。其中，第一服务器与第一终端设备相关联。也就是说，第一终端设备可以基于第一服务器的请求，向网络设备请求第一特征分布信息。
示例性地，第三请求信息可以包括区域信息和/或时间段信息，第二请求信息所包括的内容可以根据第三请求信息所包括的内容确定，比如第二请求信息所包括的内容和第三请求信息所包括的内容相同。
(3)实现方式3
在实现方式3中，第一终端设备可以向网络设备发送第二请求信息，第二请求信息用于请求向第一终端设备发送第一特征分布信息；相应地，网络设备接收第二请求信息后，可以获知需要向第一终端设备发送第一特征分布信息。比如，第一终端设备向网络设备发送第二请求信息，可以是在网络设备发送第一请求信息(参见S501中的描述)之前；也就是说，网络设备接收到第二请求信息后，可以根据第二请求信息发送第一请求信息。
可选地，在第一终端设备向网络设备发送第二请求信息之前，第一终端设备可以接收来自第一服务器的第三请求信息，第三请求信息用于请求向第一服务器发送第一特征分布信息。
示例性地，第三请求信息可以包括以下至少一项：第一指示信息、第二指示信息、终端配置信息、触发信息、数量信息、区域信息、时间段信息、第三指示信息。第二请求信息所包括的内容可以根据第三请求信息所包括的内容确定，比如第二请求信息所包括的内容和第三请求信息所包括的内容相同。可选地，第一请求信息所包括的内容可以根据第二请求信息所包括的内容确定，比如第一请求信息所包括的内容和第二请求信息所包括的内容相同。
可以理解的是，上述实现方式2和实现方式3的主要区别在于：实现方式2中，网络设备自主确定第一特征分布信息，并基于第一终端设备的请求，向第一终端设备发送第一特征分布信息；实现方式3中，网络设备基于第一终端设备的请求，确定第一特征分布信息，并向第一终端设备发送第一特征分布信息。
可选地，在上述S502中，网络设备还可以向第一终端设备发送指示信息，指示信息指示第一特征分布信息对应的终端配置信息、触发信息、数量信息、区域信息和时间段信息中的至少一项，比如指示信息包括上述区域信息和时间段信息。
可选地，在其它可能的实现中，上述S501中的第一信息也可以包括P1组数据，而不包括P1组数据的特征信息、P1组数据的特征分布信息。此种情形下，网络设备可以根据P1组数据(以及可选地，P2组数据和P3组数据)，确定第一特征分布信息。
可选地，上述S501为可选步骤，比如在第一时间段内，第一终端设备可能未处于第一区域，进而第一终端设备可以不向网络设备发送第一信息。
本申请实施例中，向网络设备上报特征信息和/或特征分布信息的终端设备，与接收第一特征分布信息的终端设备可以存在交集。也就是说，某一终端设备可能向网络设备上报了特征信息和/或特征分布信息，但未接收到来自网络设备的第一特征分布信息；或者，某一终端设备可能未向网络设备上报特征信息和/或特征分布信息，但接收到了来自网络设备的第一特征分布信息；又或者，某一终端设备向网络设备上报了特征信息和/或特征分布信息，也接收到了来自网络设备的第一特征分布信息。
S503，第一终端设备向第一服务器发送第一特征分布信息，相应地，第一服务器接收来自第一终端设备的第一特征分布信息。
可选地，第一终端设备还可以向第一服务器发送指示信息，指示信息指示第一特征分布信息对应的终端配置信息、触发信息、数量信息、区域信息和时间段信息中的至少一项，比如指示信息包括上述区域信息和时间段信息。
S504，第一服务器根据第一特征分布信息和第一训练数据，生成第一AI模型。
示例性地，第一服务器将第一特征分布信息作为生成第一AI模型的辅助信息，比如第一服务器可 以基于第一特征分布信息，对训练过程进行调整，从而使得根据第一训练数据生成的第一AI模型接近于根据第一区域内总的训练数据生成的AI模型。其中，第一训练数据参见下文的描述。本申请实施例对“第一服务器根据第一特征分布信息和第一训练数据，生成第一AI模型”的具体实现不做限定。
可选地，第一服务器接收到第一特征分布信息后，便可以根据第一特征分布信息和第一训练数据，生成第一AI模型。
或者，第一服务器接收到第一特征分布信息后，可以先判断第一特征分布信息和第二特征分布信息的差异是否满足第一条件。若是，则可以根据第一训练数据和第一特征分布信息，生成第一AI模型；若否，则无需生成第一AI模型。
其中，第二特征分布信息为第一服务器预先获取的。比如，在S504之前，第一服务器根据第二训练数据和第二特征分布信息，生成了第二AI模型。进而，在第一服务器接收到第一特征分布信息后，若第一服务器确定第一特征分布信息和第二特征分布信息的差异满足第一条件，则说明第二AI模型不适用于当前的数据，需要对第二AI模型进行更新；进而，第一服务器可以根据第一训练数据和第一特征分布信息，生成第一AI模型。若第一服务器确定第一特征分布信息和第二特征分布信息的差异不满足第一条件，则说明第二模型仍然可以适用于当前的数据，可以无需对第二AI模型进行更新。
如此，本申请实施例提供了一种更新AI模型的判断依据，即第一服务器可以依据第一特征分布信息和第二特征分布信息来判断AI模型是否需要更新，从而能够较为简便快捷地识别出AI模型是否需要更新，以便于更合理地对AI模型进行更新，避免频繁更新AI模型导致服务器的处理负担较重，或者长时间未更新AI模型导致AI模型的性能较差。
(1)对第一特征分布信息和第二特征分布信息的差异满足第一条件进行描述。
示例性地，衡量第一特征分布信息和第二特征分布信息的差异的量化指标可以有多种。
在一个示例中，量化指标为距离，则第一特征分布信息和第二特征分布信息的差异满足第一条件，可以是指：第一特征分布信息和第二特征分布信息之间的距离大于或等于距离阈值。可选地，距离阈值可以为预先配置的。
其中，第一特征分布信息和第二特征分布信息之间的距离可以包括但不限于如下一项或多项：余弦相似度、欧氏距离、曼哈顿距离(manhattan distance)、标准欧氏距离(standardized Euclidean distance)、平方欧式距离(squared fuclidean distance)、坎贝拉距离(canberra distance)、切比雪夫距离(chebyshev distance)、相关系数距离(correlation distance)、马氏距离(mahalanobis distance)和闵可夫斯基距离(minkowski distance)。
在另一个示例中，量化指标为两种分布间的偏差度量，则第一特征分布信息和第二特征分布信息的差异满足第一条件，可以是指：第一特征分布信息和第二特征分布信息之间的距离大于或等于两种分布间的偏差度量阈值。可选地，偏差度量阈值可以为预先配置的。
其中，第一特征分布信息和第二特征分布信息之间的偏差度量可以包括但不限于如下一项或多项：KL散度、柯尔莫可洛夫-斯米洛夫检验(Kolmogorov-Smirnov test，K-S test)、模型稳定度指标(population stability index，PSI)、海林格距离(Hellinger distance)、卡方检验(chi-squared test)、交叉熵、特征基数或频率等。
(2)对第一训练数据进行描述。
(2.1)作为一种可能的实现，第一服务器可以向第一服务器关联的终端设备发送第四请求信息，第四请求信息用于请求终端设备上报终端设备位于第一区域且在第一时间段内采集的数据，也即，第一服务器可以请求终端设备上报终端设备之前采集的数据。
比如，第一服务器关联的终端设备包括第一终端设备，则第一终端设备可以向第一服务器发送P1组数据，此种情形下，第一训练数据包括P1组数据；又比如，第一服务器关联的终端设备包括第二终端设备，则第二终端设备可以向第一服务器发送P2组数据，此种情形下，第一训练数据包括P2组数据。
(2.2)作为又一种可能的实现，第一服务器可向第一服务器关联的终端设备发送第四请求信息，第四请求信息用于请求终端设备上报终端设备位于第一区域且在第二时间段内采集的数据。其中第二时间段与第一时间段的时长相同，且第二时间段的起始时间位于第一时间段的结束时间之后。也即，第一服务器可以请求终端设备在第二时间段内再次采集数据，并上报采集的数据。通常情况下，第二时间段的起始时间与第一时间段的结束时间之间的时间间隔较小，因此，可以认为：终端设备位于第一区域且 在第一时间段内采集的数据，与终端设备位于第一区域且在第二时间段内采集的数据的特征分布相同。也就是说，可以认为：第一特征分布信息也用于指示位于第一区域的终端设备在第二时间段内采集的数据的特征分布。
比如，第一服务器关联的终端设备包括第一终端设备，第一终端设备在第一区域且在第二时间段内采集Q1组数据，并上报给第一服务器，此种情形下，第一训练数据包括Q1组数据；又比如，第一服务器关联的终端设备包括第二终端设备，第二终端设备在第一区域且在第二时间段内采集Q2组数据，并上报给第一服务器，此种情形下，第一训练数据包括Q2组数据。Q1和Q2均为正整数。
如此，当第一服务器确定需要更新AI模型时，可以向第一终端设备发送第四请求信息，进而第一终端设备基于第四请求向第一服务器上报采集的数据；也就是说，当第一服务器不需要更新AI模型时，第一终端设备可无需向第一服务器上报采集的数据，从而可以有效节省传输资源。
可以理解的是，向第一服务器发送训练数据的终端设备，与向第一服务器发送第一特征分布信息的终端设备可以存在交集。
此外，在其它可能的实现中，第一训练数据也可以为第一服务器预先获取的数据，比如预先存储在第一服务器中的数据。
S505，第一服务器向第一终端设备发送第一AI模型的配置信息。
示例性地，第一服务器可以主动向第一服务器关联的某一或某些终端设备发送第一AI模型的配置信息；或者，第一服务器也可以基于第一服务器关联的某一或某些终端设备的请求，向终端设备发送第一AI模型的配置信息。也就是说，向第一服务器发送第一特征分布信息的终端设备与接收第一AI模型的配置信息的终端设备可能为相同的终端设备，或者也可能不同。此处是以第一服务器向第一终端设备发送第一AI模型的配置信息为例进行描述的。
示例性地，第一AI模型的配置信息用于配置第一AI模型。比如，第一AI模型的配置信息包括第一AI模型的参数，第一AI模型的参数比如包括第一AI模型的结构参数和/或权重；又比如，第一AI模型的配置信息包括第一AI模型的配置文件(profile)；又比如，第一AI模型的配置信息包括第一AI模型的获取信息，例如获取地址、获取协议或模型文件格式等。其中，获取地址比如可以为网络协议(internet protocol，IP)地址，或者也可以为设备的标识和/或地址。如此，第一终端设备接收到第一AI模型的配置信息后，可以根据第一AI模型的配置信息，获取并在第一区域内使用第一AI模型。
可选地，本申请实施例不对第一AI模型的应用场景进行限定，对应不同的应用场景，AI模型可以为具有不同功能的算法模型。比如，第一AI模型的应用场景可以包括但不限于以下一项或多项：终端设备侧的下行信道状态信息编码、终端设备侧的波束管理、终端设备侧的定位。
可选地，在上述S505之后，若第一服务器接收到第三特征分布信息，则可以判断第三特征分布信息与第一特征分布信息之间的差异是否满足第一条件，若满足第一条件，则可以根据第三特征分布信息，生成一个新的AI模型，以实现AI模型的更新。
采用上述方法，由于网络设备将第一特征分布信息发送给第一终端设备，以及第一终端设备将第一特征分布信息发送给第一终端设备关联的第一服务器。如此，虽然第一服务器生成AI模型所使用的第一训练数据是第一服务器关联的终端设备在第一区域内采集的数据，第一训练数据的特征分布可能不同于第一区域内总的训练数据的特征分布，但由于第一服务器可以获取到第一特征分布信息(用于指示第一区域内总的训练数据的特征分布)，因此，第一服务器可以将第一特征分布信息作为生成AI模型的辅助信息，从而使得根据第一训练数据生成的AI模型接近于根据第一区域内总的训练数据生成的AI模型，能够有效提高AI模型的性能。基于实施例一的方案，下面结合实施例二和实施例三描述两种可能的实现流程。
实施例二
图6为本申请实施例二提供的通信方法所对应的流程示意图，图6所示意的流程可以对应于实施例一中的实现方式1。
如图6所示，该方法包括如下步骤：
S601，网络设备向位于第一区域的终端设备发送第一请求信息；相应地，位于第一区域的M个终端设备接收到第一请求信息，M为正整数。
此处，第一请求信息用于请求位于第一区域的终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。比如，网络设备可以通过广播的方式发送第一请求信息。
示例性地，触发网络设备发送第一请求信息的因素可以有多种。比如，网络设备周期性确定第一区域的设定时间段内总的训练数据的特征分布，进而网络设备可以周期性发送请求信息，以请求位于第一区域的终端设备在设定时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
S602，M个终端设备在第一区域且在第一时间段内采集数据，并向网络设备上报采集到的数据的特征信息和/或特征分布信息。
其中，M个终端设备关联不同的服务器，比如M个终端设备中的部分终端设备关联厂商1维护的服务器，M个终端设备中的另一部分终端设备关联厂商2维护的服务器。
S603，网络设备根据M个终端设备上报的特征信息和/或特征分布信息，确定第一特征分布信息。
S604，网络设备向位于第一区域的终端设备发送第一特征分布信息；相应地，位于第一区域的N个终端设备接收到第一特征分布信息，N为正整数。
比如，网络设备可以通过广播的方式发送第一特征分布信息。
可选地，M个终端设备和N个终端设备可以存在交集。比如，若M个终端设备仍然都位于第一区域，则N个终端设备可以包括M个终端设备。
S605，N个终端设备分别向各自关联的服务器发送第一特征分布信息。比如，N个终端设备包括第一终端设备，第一终端设备关联第一服务器。
S606，第一服务器接收第一特征分布信息，确定第一特征分布信息与第二特征分布信息的差异满足第一条件。
S607，第一服务器向K个终端设备发送第四请求信息，第四请求信息用于请求K个终端设备在第一区域且在第二时间段内采集数据，并上报采集的数据，K为正整数。
可选地，K个终端设备与N个终端设备可以存在交集。
S608，K个终端设备在第一区域且在第二时间段内采集数据，并向第一服务器上报采集的数据。
S609，第一服务器根据第一特征分布信息和第一训练数据，生成第一AI模型。其中，第一训练数据包括K个终端设备在第一区域且在第二时间段内采集的数据。
S610，第一服务器向K个终端设备发送第一AI模型的配置信息。
实施例三
图7为本申请实施例三提供的通信方法所对应的流程示意图，图7所示意的流程可以对应于实施例一中的实现方式3。
如图7所示，该方法包括如下步骤：
S701，第一服务器向第一服务器关联的K个终端设备发送第三请求信息，第三请求信息用于请求K个终端设备向第一服务器发送第一特征分布信息。
示例性地，第三请求信息可以包括以下至少一项：第一指示信息、第二指示信息、终端配置信息、触发信息、数量信息、区域信息、时间段信息、第三指示信息。
可选地，K个终端设备为位于第一区域的终端设备。
S702，K个终端设备向网络设备发送第二请求信息。
比如，K个终端设备包括第一终端设备，第一终端设备发送的第二请求信息用于请求向第一终端设备发送第一特征分布信息。
S703，网络设备接收到K个终端设备的第二请求信息后，向位于第一区域的终端设备发送第一请求信息；相应地，位于第一区域的M个终端设备接收到请求信息。
比如，网络设备通过广播的方式发送第一请求信息。
可选地，M个终端设备包括K个终端设备，还可以包括第二服务器或者其它服务器关联的终端设备。
S704，M个终端设备在第一区域且在第一时间段内采集数据，并向网络设备上报采集到的数据的特征信息和/或特征分布信息。
S705，网络设备根据M个终端设备上报的特征信息和/或特征分布信息，确定第一特征分布信息。
S706，网络设备向K个终端设备发送第一特征分布信息；相应地，K个终端设备接收第一特征分 布信息。
比如，K个终端设备仍然位于第一区域，而未移出第一区域，则网络设备可以向K个终端设备发送第一特征分布信息。
S707，K个终端设备向第一服务器发送第一特征分布信息。
S708，第一服务器接收到第一特征分布信息后，确定第一特征分布信息与第二特征分布信息的差异满足第一条件。
S709，第一服务器向K个终端设备发送第四请求信息，第四请求信息用于请求K个终端设备在第一区域且在第二时间段内采集数据，并上报采集的数据。
S710，K个终端设备在第一区域且在第二时间段内采集数据，并向第一服务器上报采集的数据。
S711，第一服务器根据第一特征分布信息和第一训练数据，生成第一AI模型。其中，第一训练数据包括K个终端设备在第一区域且在第二时间段内采集的数据。
S712，第一服务器向K个终端设备发送第一AI模型的配置信息。
可以理解的是，上述是以向第一服务器发送第一特征分布信息的终端设备与向第一服务器发送数据的终端设备均为K个终端设备为例进行描述，具体实施中不限于此。
实施例四
图8为本申请实施例四提供的通信方法所对应的流程示意图。图8中以终端设备和网络设备作为该交互示意的执行主体为例来示意该方法，但本申请并不限制该交互示意的执行主体。例如，图8中的终端设备也可以是支持该终端设备实现该方法的芯片、芯片系统、或处理器；图8中的网络设备也可以是支持该接入网设备实现该方法的芯片、芯片系统、或处理器，还可以是能实现全部或部分接入网设备功能的逻辑模块或软件。
如图8所示，该方法包括如下步骤：
S801，第一终端设备向网络设备发送第一信息，第一信息包括以下至少一项：P1组数据的特征信息、P1组数据的特征分布信息，P1组数据为第一终端设备位于第一区域且在第一时间段内采集的数据。
可选地，在S801之前，上述方法还可以包括：网络设备可以向位于第一区域的终端设备发送第一请求信息，第一请求信息用于请求位于第一区域的终端设备在第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。位于第一区域的终端设备包括第一终端设备，进而第一终端设备可以接收到第一请求信息。
可选地，在上述S801中，终端设备还可以向网络设备发送指示信息，指示信息指示P1组数据对应的终端配置信息、触发信息、数量信息、区域信息和时间段信息中的至少一项，比如指示信息包括上述区域信息和时间段信息。
S802，网络设备根据第一信息，确定第一特征分布信息。
如上所述，网络设备可以接收第一终端设备上报的第一信息，此外，网络设备还可以接收其它终端设备上报的信息，比如第二终端设备上报的第二信息、第三终端设备上报的第三信息。第二信息包括以下至少一项：P2组数据的特征信息、P2组数据的特征分布信息，P2组数据为第二终端设备位于第一区域且在第一时间段内采集的数据。第三信息包括以下至少一项：P3组数据的特征信息、P3组数据的特征分布信息，P3组数据为第三终端设备位于第一区域且在第一时间段内采集的数据，P3为正整数。
如此，网络设备接收到多个终端设备上报的信息后，可以汇总得到第一特征分布信息。第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布。比如，“位于第一区域的终端设备在第一时间段内采集的数据”包括P1组数据、P2组数据和P3组数据，则第一特征分布信息用于指示P1组数据、P2组数据和P3组数据的特征分布，或者说，第一特征分布信息用于指示第三训练数据的特征分布，第三训练数据参见下文。
S803，网络设备确定第一特征分布信息与第二特征分布信息的差异满足第一条件。
可选地，第二特征分布信息为网络设备预先获取的。比如，第二特征分布信息用于指示第四AI模型的训练数据的特征分布，第四AI模型为位于第一区域的终端设备当前使用的AI模型。
若网络设备确定第一特征分布信息和第二特征分布信息的差异满足第一条件，说明第四AI模型不适用于当前的数据，即当前第四AI模型的性能较差。若网络设备确定第一特征分布信息和第二特征分布信息的差异不满足第一条件，则说明第四AI模型仍然可以适用于当前的数据。
示例性地，网络设备确定第一特征分布信息与第二特征分布信息的差异满足第一条件后，可以执行S804-a(即情形a)，或者执行S804-b(即情形b)，或者执行S804-c和S805-c(即情形c)。
S804-a，网络设备向位于第一区域的终端设备(比如第一终端设备)发送去激活信息，去激活信息用于指示去激活AI模型或AI模式；相应地，位于第一区域的终端设备接收去激活信息。
此处，去激活信息用于指示去激活AI模型或AI模式，可以理解为：不再使用AI模型或者退出AI模式。比如，第一终端设备在接收到去激活信息之前，是使用第二AI模型对下行信道状态信息进行编码；则第一终端设备在接收到去激活信息之后，可以采用传统方法对下行信道状态信息进行编码，而不再使用AI模型对下行信道状态信息进行编码。
S804-b，网络设备向位于第一区域的终端设备(比如第一终端设备)发送切换信息，切换信息用于指示将使用的AI模型切换为目标AI模型；相应地，第一区域的终端设备接收切换指示信息。
可选地，切换信息可以包括以下至少一项：目标AI模型的标识信息、目标AI模型的配置信息。其中，目标AI模型的配置信息可以参照上文第一AI模型的配置信息的描述。
举个例子，网络设备预先向第一终端设备发送了多个AI模型(比如AI模型1、AI模型2和AI模型3)的配置信息。其中，特征分布信息1用于指示AI模型1的训练数据的特征分布，特征分布信息2用于指示AI模型2的训练数据的特征分布，特征分布信息3用于指示AI模型3的训练数据的特征分布。第一终端设备当前使用的AI模型为AI模型1，则网络设备确定第一特征分布信息与特征分布信息1之间的差异满足第一条件后，可以根据特征分布信息2、特征分布信息3分别与第一特征分布信息的差异，从AI模型2和AI模型3中选择目标AI模型。比如，特征分布信息2和第一特征分布信息之间的差异较小，则可以选择AI模型2作为目标AI模型。此种情形下，切换信息可以包括AI模型2的标识信息。
S804-c，网络设备根据第三训练数据生成第三AI模型。
此处对第三训练数据进行描述。
(1)作为一种可能的实现，网络设备确定第一特征分布信息与第二特征分布信息的差异满足第一条件后，可以向位于第一区域的终端设备发送第五请求信息，第五请求信息用于请求终端设备上报终端设备位于第一区域且在第一时间段内采集的数据，也即，网络设备可以请求终端设备上报终端设备之前采集的数据。比如，位于第一区域的终端设备包括第一终端设备、第二终端设备和第三终端设备，则第一终端设备可以向网络设备发送P1组数据，第二终端设备可以向网络设备发送P2组数据，第三终端设备可以向网络设备发送P3组数据；此种情形下，第三训练数据包括P1组数据、P2组数据和P3组数据。
或者，以第一终端设备为例，P1组数据也可以包含在第一信息中，此种情形下，网络设备无需在发送第五请求信息。
(2)作为又一种可能的实现，网络设备确定第一特征分布信息与第二特征分布信息的差异满足第一条件后，可以位于第一区域的终端设备发送第五请求信息，第五请求信息用于请求终端设备上报终端设备位于第一区域且在第二时间段内采集的数据。其中，第二时间段与第一时间段的时长相同，且第二时间段的起始时间位于第一时间段的结束时间之后。也即，网络设备可以请求终端设备在第二时间段内再次采集数据，并上报采集的数据。通常情况下，第二时间段的起始时间与第一时间段的结束时间之间的时间间隔较小，因此，可以认为：终端设备位于第一区域且在第一时间段内采集的数据，与终端设备位于第一区域且在第二时间段内采集的数据的特征分布相同。也就是说，可以认为：第一特征分布信息也用于指示位于第一区域的终端设备在第二时间段内采集的数据的特征分布。
比如，位于第一区域内的终端设备包括第一终端设备、第二终端设备和第三终端设备；第一终端设备在第一区域且在第二时间段内采集Q1组数据，并上报给网络设备；第二终端设备在第一区域且在第二时间段内采集Q2组数据，并上报给网络设备；第三终端设备在第一区域且在第二时间段内采集Q3组数据，并上报给网络设备。此种情形下，第三训练数据包括Q1组数据、Q2组数据和Q3组数据。Q1、Q2和3均为正整数。
S805-c，网络设备向位于第一区域的终端设备(比如第一终端设备)发送第三AI模型的配置信息；相应地，位于第一区域的终端设备可以接收第三AI模型的配置信息。
示例性地，第三AI模型的配置信息可以参照实施例一中第一AI模型的配置信息的描述。以第一终端设备为例，第一终端设备接收到第三AI模型的配置信息后，可以在第一区域内使用第三 AI模型，而不再使用第四AI模型。
采用上述方法，网络设备可以依据第一特征分布信息和第二特征分布信息，来判断当前AI模型的性能，从而能够较为简便快捷地识别出AI模型的性能是否较差。以及，在AI模型的性能较差的情况下，网络设备可以采取一系列的解决措施；比如，在AI模型的性能较差的情况下，网络设备可以对AI模型进行更新。进一步地，本申请实施例中，由于网络设备是在确定需要更新AI模型时，向第一终端设备发送第五请求信息，进而第一终端设备基于第五请求向网络设备上报P1组数据；也就是说，当网络设备不需要更新AI模型时，第一终端设备可无需向网络设备发送P1组数据，从而可以有效节省传输资源。
针对于上述实施例，可以理解的是：
(1)上述侧重描述了实施例一至实施例四中不同实施例之间的差异之处，除差异之处的其它内容，实施例一至实施例四之间可以相互参照；此外，同一实施例中，不同实现方式或不同示例或不同步骤之间也可以相互参照。
(2)上述实施例一至实施例四中，是以“第一区域包括的多个小区属于同一个基站，且网络设备为这多个小区所属的基站”为例进行描述的。
当第一区域包括的多个小区属于同一个基站时，该基站可以根据接收到的特征信息和/或特征分布信息，汇总得到第一特征分布信息。而当第一小区包括的多个小区属于多个基站时，此处提供两种确定第一特征分布信息的方式。
方式1：多个基站可以分别将接收到的特征信息和/或特征分布信息发送给核心网设备，进而该核心网设备可以汇总得到第一特征分布信息，并将第一特征分布信息发送给这多个基站。
方式2：多个基站中包括第一基站，多个基站中除第一基站以外的其它基站可以分别将接收到的特征信息和/或特征分布信息发送给第一基站，进而第一基站可以汇总得到第一特征分布信息，并将第一特征分布信息发送给其它基站。
此外，除上述区别外，“第一小区包括的多个小区属于多个基站”的实现可以参照“第一小区包括的多个小区属于同一基站”的实现，不再赘述。
(3)上述实施例四中，是以“网络设备生成AI模型”为例进行描述的，在其它可能的实现中，网络设备可以将训练数据发送给网络设备对应的服务器，进而由网络设备对应的服务器根据训练数据生成AI模型，并发送给网络设备。
(4)本申请实施例中，“多组数据的特征分布信息”也可以替换为“多组数据的分布信息”。其中，多组数据的特征分布信息用于指示多组数据的特征分布，多组数据的分布信息用于指示多组数据的分布。类似于多组数据的特征分布，多组数据的分布可以表示属于多个类别中每个类别的数据的数量或比例。多组数据的特征分布与多组数据的分布的区别之处在于：多组数据的特征分布是基于提取的特征来确定数据的分布，而多组数据的分布是基于数据本身来确定数据的分布。
(5)本申请实施例中，终端设备向网络设备或第一服务器上报的数据，可以为终端设备采集的未经过预处理的数据，或者也可以是终端设备采集的经过预处理的数据，具体预处理的方式不做限定。
(6)本申请实施例中，“根据训练数据和第一特征分布信息，生成AI模型”，可以包括：根据训练数据和第一特征分布信息，对已有的AI模型进行微调(finetuning)得到新的AI模型。类似之处可参照处理。
上述主要从设备之间交互的角度对本申请实施例提供的方案进行了介绍。可以理解的是，为了实现上述功能，各个设备可以包括执行各个功能相应的硬件结构和/或软件模块。本领域技术人员应该很容易意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，本申请的实施例能够以硬件或硬件和计算机软件的结合形式来实现。某个功能究竟以硬件还是计算机软件驱动硬件的方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
本申请实施例可以根据上述方法示例对网络设备、终端设备和服务器进行功能单元的划分，例如，可以对应各个功能划分各个功能单元，也可以将两个或两个以上的功能集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
在采用集成的单元的情况下，图9示出了本申请实施例中所涉及的装置的可能的示例性框图。如图9所示，装置900可以包括：处理单元902和通信单元903。处理单元902用于对装置900的动作进行 控制管理。通信单元903用于支持装置900与其他设备的通信。可选地，通信单元903也称为收发单元，可以包括接收单元和/或发送单元，分别用于执行接收和发送操作。装置900还可以包括存储单元901，用于存储装置900的程序代码和/或数据。
该装置900可以为上述实施例中的第一终端设备。处理单元902可以支持装置900执行上文中各方法示例中第一终端设备的动作。或者，处理单元902主要执行方法示例中第一终端设备的内部动作，通信单元903可以支持装置900与其它设备之间的通信。
比如，在一个实施例中，通信单元903用于：向网络设备发送第一信息，所述第一信息包括以下至少一项：P1组数据的特征信息、所述P1组数据的特征分布信息，所述P1组数据为所述第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；以及，接收来自所述网络设备的第一特征分布信息，所述第一特征分布信息用于指示位于所述第一区域的终端设备在所述第一时间段内采集的数据的特征分布，所述终端设备包括所述第一终端设备；以及，向第一服务器发送所述第一特征分布信息，所述第一服务器与所述第一终端设备相关联。
该装置900可以为上述实施例中的网络设备。处理单元902可以支持装置900执行上文中各方法示例中网络设备的动作。或者，处理单元902主要执行方法示例中网络设备的内部动作，通信单元903可以支持装置900与其它设备之间的通信。
比如，在一个实施例中，通信单元903用于：接收来自第一终端设备的第一信息，所述第一信息包括以下至少一项：P1组数据的特征信息、所述P1组数据的特征分布信息，所述P1组数据为所述第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；以及，向所述第一终端设备发送第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布，所述终端设备包括所述第一终端设备。
该装置900可以为上述实施例中的第一服务器。处理单元902可以支持装置900执行上文中各方法示例中第一服务器的动作。或者，处理单元902主要执行方法示例中第一服务器的内部动作，通信单元903可以支持装置900与其它设备之间的通信。
比如，在一个实施例中，通信单元903用于：接收来自第一终端设备的第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布，所述第一服务器与所述第一终端设备相关联；处理单元902用于：根据所述第一特征分布信息，生成AI模型；以及，通信单元903还用于：向所述第一终端设备发送所述AI模型的配置信息。
应理解以上装置中单元的划分仅仅是一种逻辑功能的划分，实际实现时可以全部或部分集成到一个物理实体上，也可以物理上分开。且装置中的单元可以全部以软件通过处理元件调用的形式实现；也可以全部以硬件的形式实现；还可以部分单元以软件通过处理元件调用的形式实现，部分单元以硬件的形式实现。例如，各个单元可以为单独设立的处理元件，也可以集成在装置的某一个芯片中实现，此外，也可以以程序的形式存储于存储器中，由装置的某一个处理元件调用并执行该单元的功能。此外这些单元全部或部分可以集成在一起，也可以独立实现。这里所述的处理元件又可以成为处理器，可以是一种具有信号的处理能力的集成电路。在实现过程中，上述方法的各操作或以上各个单元可以通过处理器元件中的硬件的集成逻辑电路实现或者以软件通过处理元件调用的形式实现。
在一个例子中，以上任一装置中的单元可以是被配置成实施以上方法的一个或多个集成电路，例如：一个或多个特定集成电路(application specific integrated circuit，ASIC)，或，一个或多个微处理器(digital singnal processor，DSP)，或，一个或者多个现场可编程门阵列(field programmable gate array，FPGA)，或这些集成电路形式中至少两种的组合。再如，当装置中的单元可以通过处理元件调度程序的形式实现时，该处理元件可以是处理器，比如通用中央处理器(central processing unit，CPU)，或其它可以调用程序的处理器。再如，这些单元可以集成在一起，以片上系统(system-on-a-chip，SOC)的形式实现。
以上用于接收的单元是一种该装置的接口电路，用于从其它装置接收信号。例如，当该装置以芯片的方式实现时，该接收单元是该芯片用于从其它芯片或装置接收信号的接口电路。以上用于发送的单元是一种该装置的接口电路，用于向其它装置发送信号。例如，当该装置以芯片的方式实现时，该发送单元是该芯片用于向其它芯片或装置发送信号的接口电路。
参见图10，为本申请实施例提供的一种终端设备的结构示意图，该终端设备可应用于如图1所示的通信系统中，用于实现以上实施例中终端设备的操作。如图10所示，该终端设备包括：天线1010、射频部分1020、信号处理部分1030。天线1010与射频部分1020连接。在下行方向上，射频部分1020 通过天线1010接收接入网设备发送的信息，将接入网设备发送的信息发送给信号处理部分1030进行处理。在上行方向上，信号处理部分1030对终端设备的信息进行处理，并发送给射频部分1020，射频部分1020对终端设备的信息进行处理后经过天线1010发送给接入网设备。
信号处理部分1030可以包括调制解调子系统，用于实现对数据各通信协议层的处理；还可以包括中央处理子系统，用于实现对终端设备操作系统以及应用层的处理；此外，还可以包括其它子系统，例如多媒体子系统，周边子系统等，其中多媒体子系统用于实现对终端设备相机，屏幕显示等的控制，周边子系统用于实现与其它设备的连接。调制解调子系统可以为单独设置的芯片。
调制解调子系统可以包括一个或多个处理元件1031，例如，包括一个主控CPU和其它集成电路。此外，该调制解调子系统还可以包括存储元件1032和接口电路1033。存储元件1032用于存储数据和程序，但用于执行以上方法中终端设备所执行的方法的程序可能不存储于该存储元件1032中，而是存储于调制解调子系统之外的存储器中，使用时调制解调子系统加载使用。接口电路1033用于与其它子系统通信。
该调制解调子系统可以通过芯片实现，该芯片包括至少一个处理元件和接口电路，其中处理元件用于执行以上终端设备执行的任一种方法的各个步骤，接口电路用于与其它装置通信。在一种实现中，终端设备实现以上方法中各个步骤的单元可以通过处理元件调度程序的形式实现，例如用于终端设备的装置包括处理元件和存储元件，处理元件调用存储元件存储的程序，以执行以上方法实施例中终端设备执行的方法。存储元件可以为与处理元件处于同一芯片上的存储元件，即片内存储元件。
在另一种实现中，用于执行以上方法中终端设备所执行的方法的程序可以在与处理元件处于不同芯片上的存储元件，即片外存储元件。此时，处理元件从片外存储元件调用或加载程序于片内存储元件上，以调用并执行以上方法实施例中终端设备执行的方法。
在又一种实现中，终端设备实现以上方法中各个步骤的单元可以是被配置成一个或多个处理元件，这些处理元件设置于调制解调子系统上，这里的处理元件可以为集成电路，例如：一个或多个ASIC，或，一个或多个DSP，或，一个或者多个FPGA，或者这些类集成电路的组合。这些集成电路可以集成在一起，构成芯片。
终端设备实现以上方法中各个步骤的单元可以集成在一起，以SOC的形式实现，该SOC芯片，用于实现以上方法。该芯片内可以集成至少一个处理元件和存储元件，由处理元件调用存储元件的存储的程序的形式实现以上终端设备执行的方法；或者，该芯片内可以集成至少一个集成电路，用于实现以上终端设备执行的方法；或者，可以结合以上实现方式，部分单元的功能通过处理元件调用程序的形式实现，部分单元的功能通过集成电路的形式实现。
可见，以上用于终端设备的装置可以包括至少一个处理元件和接口电路，其中至少一个处理元件用于执行以上方法实施例所提供的任一种终端设备执行的方法。处理元件可以以第一种方式：即调用存储元件存储的程序的方式执行终端设备执行的部分或全部步骤；也可以以第二种方式：即通过处理器元件中的硬件的集成逻辑电路结合指令的方式执行终端设备执行的部分或全部步骤；当然，也可以结合第一种方式和第二种方式执行终端设备执行的部分或全部步骤。
这里的处理元件同以上描述，可以通过处理器实现，处理元件的功能可以和图9中所描述的处理单元的功能相同。示例性地，处理元件可以是通用处理器，例如CPU，还可以是被配置成实施以上方法的一个或多个集成电路，例如：一个或多个ASIC，或，一个或多个微处理器DSP，或，一个或者多个FPGA等，或这些集成电路形式中至少两种的组合。存储元件可以通过存储器实现，存储元件的功能可以和图9中所描述的存储单元的功能相同。存储元件可以是一个存储器，也可以是多个存储器的统称。
图10所示的终端设备能够实现上述方法实施例中涉及终端设备的各个过程。图10所示的终端设备中的各个模块的操作和/或功能，分别为了实现上述方法实施例中的相应流程。具体可参见上述方法实施例中的描述，为避免重复，此处适当省略详述描述。
参见图11，为本申请实施例提供的一种网络设备的结构示意图，该网络设备(或基站)可应用于如图1所示的通信系统中，执行上述方法实施例中网络设备的功能。如图11所示，网络设备110可包括一个或多个DU 1101和一个或多个CU 1102。所述DU 1101可以包括至少一个天线11011，至少一个射频单元11012，至少一个处理器11013和至少一个存储器11014。所述DU 1101部分主要用于射频信号的收发以及射频信号与基带信号的转换，以及部分基带处理。CU1102可以包括至少一个处理器11022和至少一个存储器11021。
所述CU 1102部分主要用于进行基带处理，对网络设备进行控制等。所述DU 1101与CU 1102可以是物理上设置在一起，也可以物理上分离设置的，即分布式基站。所述CU 1102为网络设备的控制中心，也可以称为处理单元，主要用于完成基带处理功能。例如所述CU 1102可以用于控制网络设备执行上述方法实施例中关于网络设备的操作流程。
此外，可选的，网络设备110可以包括一个或多个射频单元，一个或多个DU和一个或多个CU。其中，DU可以包括至少一个处理器11013和至少一个存储器11014，射频单元可以包括至少一个天线11011和至少一个射频单元11012，CU可以包括至少一个处理器11022和至少一个存储器11021。
在一个实例中，所述CU1102可以由一个或多个单板构成，多个单板可以共同支持单一接入指示的无线接入网(如5G网)，也可以分别支持不同接入制式的无线接入网(如LTE网，5G网或其他网)。所述存储器11021和处理器11022可以服务于一个或多个单板。也就是说，可以每个单板上单独设置存储器和处理器。也可以是多个单板共用相同的存储器和处理器。此外每个单板上还可以设置有必要的电路。所述DU1101可以由一个或多个单板构成，多个单板可以共同支持单一接入指示的无线接入网(如5G网)，也可以分别支持不同接入制式的无线接入网(如LTE网，5G网或其他网)。所述存储器11014和处理器11013可以服务于一个或多个单板。也就是说，可以每个单板上单独设置存储器和处理器。也可以是多个单板共用相同的存储器和处理器。此外每个单板上还可以设置有必要的电路。
图11所示的网络设备能够实现上述方法实施例中涉及网络设备的各个过程。图11所示的网络设备中的各个模块的操作和/或功能，分别为了实现上述方法实施例中的相应流程。具体可参见上述方法实施例中的描述，为避免重复，此处适当省略详述描述。
参考图12，为本申请实施例提供的一种服务器的结构示意图，用于实现以上实施例中第一服务器的操作。如图12所示，服务器1200可包括处理器1201、存储器1202以及接口电路1203。处理器1201可用于对通信协议以及通信数据进行处理，以及对通信装置进行控制。存储器1202可用于存储程序和数据，处理器1201可基于该程序执行本申请实施例中由第一服务器执行的方法。接口电路1203可用于服务器1200与其他设备进行通信，该通信可以为有线通信或无线通信，该接口电路例如可以是服务化通信接口。
以上存储器1202也可以是外接于服务器1200的，此时服务器1200可包括接口电路1203以及处理器1201。以上接口电路1203也可以是外接于服务器1200的，此时服务器1200可包括存储器1202以及处理器1201。当接口电路1203以及存储器1202均外接于服务器1200时，服务器1200可包括处理器1201。
图12所示的服务器能够实现上述方法实施例中涉及第一服务器的各个过程。图12所示的服务器中的各个模块的操作和/或功能，分别为了实现上述方法实施例中的相应流程。具体可参见上述方法实施例中的描述，为避免重复，此处适当省略详述描述。
本申请实施例中的术语“系统”和“网络”可被互换使用。“至少一种”是指一种或者多种，“多个”是指两个或两个以上。“和/或”，描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A、同时存在A和B、单独存在B的情况，其中A，B可以是单数或者复数。字符“/”一般表示前后关联对象是一种“或”的关系。“以下至少一项(个)”或其类似表达，是指的这些项中的任意组合，包括单项(个)或复数项(个)的任意组合。例如“A，B 和C中的至少一个”包括A，B，C，AB，AC，BC或ABC。以及，除非有特别说明，本申请实施例提及“第一”、“第二”等序数词是用于对多个对象进行区分，不用于限定多个对象的顺序、时序、优先级或者重要程度。
本领域内的技术人员应明白，本申请的实施例可提供为方法、系统、或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、光学存储器等)上实施的计算机程序产品的形式。
本申请是参照根据本申请的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
显然，本领域的技术人员可以对本申请进行各种改动和变型而不脱离本申请的精神和范围。这样，倘若本申请的这些修改和变型属于本申请权利要求及其等同技术的范围之内，则本申请也意图包含这些改动和变型在内。

Claims (30)

  1. 一种通信系统，其特征在于，所述通信系统包括网络设备、第一终端设备和第一服务器，所述第一终端设备与所述第一服务器相关联；其中，
    所述网络设备用于，向第一终端设备发送第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布；
    所述第一终端设备用于，接收所述第一特征分布信息，向所述第一服务器发送所述第一特征分布信息；
    所述第一服务器用于，接收所述第一特征分布信息，根据所述第一特征分布信息和训练数据，生成人工智能AI模型；以及，向所述第一终端设备或与所述第一服务器关联的第二终端设备发送所述AI模型的配置信息。
  2. 根据权利要求1所述的系统，其特征在于，
    所述第一终端设备还用于，向所述网络设备发送第一信息，所述第一信息包括以下至少一项：P1组数据的特征信息、所述P1组数据的特征分布信息，所述P1组数据为所述第一终端设备位于所述第一区域且在所述第一时间段内采集的数据，P1为正整数；所述终端设备包括所述第一终端设备；
    所述网络设备还用于，根据所述第一信息，确定所述第一特征分布信息。
  3. 根据权利要求2所述的系统，其特征在于，
    所述网络设备还用于，发送第一请求信息，所述第一请求信息用于请求所述终端设备在所述第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息；
    所述第一终端设备在向所述网络设备发送所述第一信息之前，还用于接收所述第一请求信息。
  4. 根据权利要求3所述的系统，其特征在于，所述第一请求信息包括以下至少一项：
    第一指示信息，所述第一指示信息指示请求所述终端设备上报的信息为所述终端设备采集的数据的特征信息和/或特征分布信息；
    第二指示信息，所述第二指示信息指示第一方式和/或第二方式，所述第一方式用于确定所述终端设备采集的数据的特征信息，所述第二方式用于确定所述终端设备采集的数据的特征分布信息；
    终端配置信息，所述终端配置信息指示所述第一终端设备采集所述P1组数据时应满足的配置条件；
    触发信息，所述触发信息包括触发方式和/或触发配置，所述触发方式包括以下至少一项：事件触发、周期触发，所述触发配置包括以下至少一项：触发事件、触发周期；
    数量信息，所述数量信息用于指示所述P1的数值；
    区域信息，所述区域信息用于指示所述第一区域；
    时间段信息，所述时间段信息用于指示所述第一时间段。
  5. 根据权利要求1至4中任一项所述的系统，其特征在于，
    所述第一终端设备还用于，向所述网络设备发送第二请求信息，所述第二请求信息用于请求向所述第一终端设备发送所述第一特征分布信息；
    所述网络设备在向所述第一终端设备发送第一特征分布信息之前，还用于接收所述第二请求信息。
  6. 根据权利要求1至5中任一项所述的系统，其特征在于，
    所述第一服务器还用于，向所述第一终端设备发送第三请求信息，所述第三请求信息用于请求向所述第一服务器发送所述第一特征分布信息；
    所述第一终端设备在向所述第一服务器发送所述第一特征分布信息之前，还用于接收所述第三请求信息。
  7. 根据权利要求1至6中任一项所述的系统，其特征在于，所述第一服务器在根据所述第一特征分布信息和所述训练数据，生成所述AI模型之前，还用于确定所述第一特征分布信息和第二特征分布信息的差异满足第一条件，所述第二特征分布信息为所述第一服务器预先获取的。
  8. 根据权利要求1至7中任一项所述的系统，其特征在于，所述第一服务器还用于，接收来自所述第一终端设备的P1组数据，所述训练数据包括所述P1组数据，所述P1组数据为所述第一终端设备位于所述第一区域且在所述第一时间段内采集的数据，P1为正整数。
  9. 根据权利要求1至8中任一项所述的系统，其特征在于，所述第一服务器还用于，接收来自所述第二终端设备的P2组数据，所述训练数据包括所述P2组数据，所述P2组数据为所述第二终端设备位 于所述第一区域且在所述第一时间段内采集的数据，P2为正整数。
  10. 根据权利要求1至7中任一项所述的系统，其特征在于，所述训练数据包括所述第一服务器预先获取的数据。
  11. 一种通信方法，其特征在于，所述方法应用于第一终端设备，所述方法包括：
    接收来自网络设备的第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布；
    向第一服务器发送所述第一特征分布信息，所述第一服务器与所述第一终端设备相关联。
  12. 根据权利要求11所述的方法，其特征在于，所述方法还包括：
    向所述网络设备发送第一信息，所述第一信息包括以下至少一项：P1组数据的特征信息、所述P1组数据的特征分布信息，所述P1组数据为所述第一终端设备位于所述第一区域且在所述第一时间段内采集的数据，P1为正整数；
    其中，所述终端设备包括所述第一终端设备，所述第一特征分布信息根据所述第一信息确定。
  13. 根据权利要求12所述的方法，其特征在于，在向所述网络设备发送第一信息之前，所述方法还包括：
    接收来自所述网络设备的第一请求信息，所述第一请求信息用于请求所述终端设备在所述第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
  14. 根据权利要求13所述的方法，其特征在于，所述第一请求信息包括以下至少一项：
    第一指示信息，所述第一指示信息指示请求所述终端设备上报的信息为所述终端设备采集的数据的特征信息和/或特征分布信息；
    第二指示信息，所述第二指示信息指示第一方式和/或第二方式，所述第一方式用于确定所述终端设备采集的数据的特征信息，所述第二方式用于确定所述终端设备采集的数据的特征分布信息；
    终端配置信息，所述终端配置信息指示所述第一终端设备采集所述P1组数据时应满足的配置条件；
    触发信息，所述触发信息包括触发方式和/或触发配置，所述触发方式包括以下至少一项：事件触发、周期触发，所述触发配置包括以下至少一项：触发事件、触发周期；
    数量信息，所述数量信息用于指示所述P1的数值；
    区域信息，所述区域信息用于指示所述第一区域；
    时间段信息，所述时间段信息用于指示所述第一时间段。
  15. 根据权利要求11至14中任一项所述的方法，其特征在于，在接收来自所述网络设备的第一特征分布信息之前，所述方法还包括：
    向所述网络设备发送第二请求信息，所述第二请求信息用于请求向所述第一终端设备发送所述第一特征分布信息。
  16. 根据权利要求15所述的方法，其特征在于，在向所述网络设备发送第二请求信息之前，所述方法还包括：
    接收来自所述第一服务器的第三请求信息，所述第三请求信息用于请求向所述第一服务器发送所述第一特征分布信息。
  17. 一种通信方法，其特征在于，所述方法应用于网络设备，所述方法包括：
    确定第一特征分布信息；
    向第一终端设备发送所述第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布。
  18. 根据权利要求17所述的方法，其特征在于，
    所述方法还包括：接收来自所述第一终端设备的第一信息，所述第一信息包括以下至少一项：P1组数据的特征信息、所述P1组数据的特征分布信息，所述P1组数据为所述第一终端设备位于所述第一区域且在所述第一时间段内采集的数据，P1为正整数；所述终端设备包括所述第一终端设备；
    确定第一特征分布信息，包括：根据所述第一信息，确定所述第一特征分布信息。
  19. 根据权利要求18所述的方法，其特征在于，在接收来自所述第一终端设备的所述第一信息之前，所述方法还包括：
    向所述终端设备发送第一请求信息，所述第一请求信息用于请求所述终端设备在所述第一时间段内采集数据，并上报采集的数据的特征信息和/或特征分布信息。
  20. 根据权利要求19所述的方法，其特征在于，所述第一请求信息包括以下至少一项：
    第一指示信息，所述第一指示信息指示请求所述终端设备上报的信息为所述终端设备采集的数据的特征信息和/或特征分布信息；
    第二指示信息，所述第二指示信息指示第一方式和/或第二方式，所述第一方式用于确定所述终端设备采集的数据的特征信息，所述第二方式用于确定所述终端设备采集的数据的特征分布信息；
    终端配置信息，所述终端配置信息指示所述第一终端设备采集所述P1组数据时应满足的配置条件；
    触发信息，所述触发信息包括触发方式和/或触发配置，所述触发方式包括以下至少一项：事件触发、周期触发，所述触发配置包括以下至少一项：触发事件、触发周期；
    数量信息，所述数量信息用于指示所述P1的数值；
    区域信息，所述区域信息用于指示所述第一区域；
    时间段信息，所述时间段信息用于指示所述第一时间段。
  21. 根据权利要求17至20中任一项所述的方法，其特征在于，在向所述第一终端设备发送第一特征分布信息之前，所述方法还包括：
    接收来自所述第一终端设备的第二请求信息，所述第二请求信息用于请求向所述第一终端设备发送所述第一特征分布信息。
  22. 一种通信方法，其特征在于，所述方法应用于第一服务器，所述方法包括：
    接收来自第一终端设备的第一特征分布信息，所述第一特征分布信息用于指示位于第一区域的终端设备在第一时间段内采集的数据的特征分布，所述第一服务器与所述第一终端设备相关联；
    根据所述第一特征分布信息，生成AI模型；
    向所述第一终端设备或所述第一服务器关联的第二终端设备发送所述AI模型的配置信息。
  23. 根据权利要求22所述的方法，其特征在于，所述方法还包括：
    接收来自所述第一终端设备的P1组数据，所述P1组数据为所述第一终端设备位于第一区域且在第一时间段内采集的数据，P1为正整数；
    其中，所述训练数据包括所述P1组数据。
  24. 根据权利要求22或23所述的方法，其特征在于，所述方法还包括：
    接收来自所述第二终端设备的P2组数据，所述P2组数据为所述第二终端设备位于第一区域且在第一时间段内采集的数据，P2为正整数；
    其中，所述训练数据包括所述P2组数据。
  25. 根据权利要求24所述的方法，其特征在于，所述训练数据包括所述第一服务器预先获取的数据。
  26. 根据权利要求22至25中任一项所述的方法，其特征在于，在根据所述第一特征分布信息，生成AI模型之前，所述方法还包括：
    确定所述第一特征分布信息和第二特征分布信息的差异满足第一条件，所述第二特征分布信息为所述第一服务器预先获取的。
  27. 根据权利要求22至26中任一项所述的方法，其特征在于，在接收来自第一终端设备的第一特征分布信息之前，所述方法还包括：
    向所述第一终端设备发送第三请求信息，所述第三请求信息用于请求向所述第一服务器发送所述第一特征分布信息。
  28. 一种通信装置，其特征在于，所述通信装置包括处理器，所述处理器用于执行如权利要求11至27中任一项所述的方法。
  29. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质存储有计算机程序或指令，当所述计算机程序或指令在计算机上运行时，使得如权利要求11至27中任一项所述的方法被执行。
  30. 一种计算机程序产品，其特征在于，所述计算机程序产品包括：计算机程序或指令，当所述计算机程序或指令在计算机上运行时，使得如权利要求11至27中任一项所述的方法被执行。
PCT/CN2023/105896 2022-07-12 2023-07-05 一种通信方法、装置及系统 WO2024012326A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN202210820351 2022-07-12
CN202210820351.1 2022-07-12
CN202210977688.3 2022-08-15
CN202210977688.3A CN117459961A (zh) 2022-07-12 2022-08-15 一种通信方法、装置及系统

Publications (1)

Publication Number Publication Date
WO2024012326A1 true WO2024012326A1 (zh) 2024-01-18

Family

ID=89535545

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/105896 WO2024012326A1 (zh) 2022-07-12 2023-07-05 一种通信方法、装置及系统

Country Status (1)

Country Link
WO (1) WO2024012326A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112580826A (zh) * 2021-02-05 2021-03-30 支付宝(杭州)信息技术有限公司 业务模型训练方法、装置及系统
WO2022015221A1 (en) * 2020-07-14 2022-01-20 Telefonaktiebolaget Lm Ericsson (Publ) Managing a wireless device that is operable to connect to a communication network
CN114548416A (zh) * 2020-11-26 2022-05-27 华为技术有限公司 数据模型训练方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022015221A1 (en) * 2020-07-14 2022-01-20 Telefonaktiebolaget Lm Ericsson (Publ) Managing a wireless device that is operable to connect to a communication network
CN114548416A (zh) * 2020-11-26 2022-05-27 华为技术有限公司 数据模型训练方法及装置
CN112580826A (zh) * 2021-02-05 2021-03-30 支付宝(杭州)信息技术有限公司 业务模型训练方法、装置及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HUAWEI, HISILICON: "Discussion on general aspects of AI/ML framework", 3GPP DRAFT; R1-2203139, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. RAN WG1, no. e-Meeting; 20220509 - 20220520, 29 April 2022 (2022-04-29), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052143957 *

Similar Documents

Publication Publication Date Title
CN111819872B (zh) 信息传输方法、装置、通信设备及存储介质
US11451452B2 (en) Model update method and apparatus, and system
US11838787B2 (en) Functional architecture and interface for non-real-time ran intelligent controller
WO2019102064A1 (en) Joint beam reporting for wireless networks
CN108738065B (zh) 一种资源信息确定方法及终端设备、网络设备
US20230113050A1 (en) Link reachability determining method and apparatus
WO2023143267A1 (zh) 一种模型配置方法及装置
US11929938B2 (en) Evaluating overall network resource congestion before scaling a network slice
CN117459961A (zh) 一种通信方法、装置及系统
WO2024012331A1 (zh) 一种确定人工智能ai模型的方法及装置
CN112088555A (zh) 用于无线通信系统中的资源分配的协调器网络节点和接入网络节点
WO2024012326A1 (zh) 一种通信方法、装置及系统
CN117716674A (zh) 用于ai-ml模型训练的基于网络资源模型的解决方案
WO2023069534A1 (en) Using ai-based models for network energy savings
WO2023160459A1 (zh) 一种人工智能算法模型获取方法及装置
WO2019095794A1 (zh) 一种传输信道质量信息的方法和装置
US20240080788A1 (en) Information transmission method, lightweight processing method, and related communication apparatus
WO2024099175A1 (zh) 一种算法管理方法和装置
WO2024041117A1 (zh) 一种计算任务的分割方法及相关装置
WO2024067248A1 (zh) 一种获取训练数据集的方法和装置
WO2023186048A1 (zh) 一种ai服务信息获取方法、装置及系统
WO2022062908A1 (zh) 一种配置切换配置信息的方法及装置
US20230351207A1 (en) Model data sending method and apparatus
WO2024093739A1 (zh) 一种通信方法及装置
US20230262502A1 (en) System and method for mdas assisted gst configuration

Legal Events

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

Ref document number: 23838818

Country of ref document: EP

Kind code of ref document: A1