WO2021212347A1 - 一种通信方法及相关设备 - Google Patents

一种通信方法及相关设备 Download PDF

Info

Publication number
WO2021212347A1
WO2021212347A1 PCT/CN2020/085998 CN2020085998W WO2021212347A1 WO 2021212347 A1 WO2021212347 A1 WO 2021212347A1 CN 2020085998 W CN2020085998 W CN 2020085998W WO 2021212347 A1 WO2021212347 A1 WO 2021212347A1
Authority
WO
WIPO (PCT)
Prior art keywords
terminal
indicator
efficiency
rate
network device
Prior art date
Application number
PCT/CN2020/085998
Other languages
English (en)
French (fr)
Inventor
沈嘉
Original Assignee
Oppo广东移动通信有限公司
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
Application filed by Oppo广东移动通信有限公司 filed Critical Oppo广东移动通信有限公司
Priority to EP20932266.8A priority Critical patent/EP4135442A4/en
Priority to PCT/CN2020/085998 priority patent/WO2021212347A1/zh
Priority to CN202080099536.7A priority patent/CN115362735A/zh
Publication of WO2021212347A1 publication Critical patent/WO2021212347A1/zh
Priority to US18/047,625 priority patent/US20230066109A1/en

Links

Images

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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/098—Distributed learning, e.g. federated learning
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W72/00—Local resource management
    • H04W72/20—Control channels or signalling for resource management
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W72/00—Local resource management
    • H04W72/50—Allocation or scheduling criteria for wireless resources
    • H04W72/53—Allocation or scheduling criteria for wireless resources based on regulatory allocation policies
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]

Definitions

  • This application relates to the field of wireless network communication technology, and in particular to a communication method and related equipment.
  • AI Artificial Intelligence
  • the embodiments of the present application provide a communication method and related equipment, which can indicate to the network equipment an indicator that associates the current wireless channel condition of the terminal and the AI parameter combination at the same time, so as to ensure the reliability of resource allocation.
  • the first aspect of the embodiments of the present application provides a communication method, including:
  • the terminal sends a first indicator to the network device, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first artificial intelligence AI parameter.
  • the combination of the first indicator and the first resource parameter forms a first correspondence.
  • the second aspect of the embodiments of the present application provides another communication method, including:
  • the network device receives the first indicator sent by the terminal, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first artificial intelligence AI parameter, so The combination of the first indicator and the first resource parameter forms a first correspondence.
  • a third aspect of the embodiments of the present application provides a communication device, including:
  • the sending unit is configured to send the first indicator to the network device by the terminal, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first artificial intelligence An AI parameter, where the first indicator and the first resource parameter are combined into a first correspondence.
  • a fourth aspect of the embodiments of the present application provides a communication device, including:
  • the receiving unit is configured to receive a first indicator sent by a terminal by a network device, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first manual resource parameter combination. Intelligent AI parameters, the first indicator and the first resource parameter are combined into a first correspondence.
  • the fifth aspect of the embodiments of the present application provides a terminal, including a processor, a receiver, a transmitter, and a memory.
  • the processor, the receiver, the transmitter, and the memory are connected to each other, wherein the memory is used to store a computer program
  • the computer program includes program instructions, and the processor is configured to invoke the program instructions to execute the method of the first aspect described above.
  • the sixth aspect of the embodiments of the present application provides a network device, including a processor, a receiver, a transmitter, and a memory.
  • the processor, the receiver, the transmitter, and the memory are connected to each other, wherein the memory is used to store a computer A program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method of the second aspect described above.
  • a seventh aspect of the embodiments of the present application provides a computer-readable storage medium, wherein the computer storage medium stores a computer program, and the computer program includes program instructions that, when executed by a processor, cause The processor executes the method of the first aspect described above.
  • An eighth aspect of the embodiments of the present application provides a computer-readable storage medium, wherein the computer storage medium stores a computer program, and the computer program includes program instructions that, when executed by a processor, cause The processor executes the method of the second aspect described above.
  • the terminal can report the current resource parameters of the terminal to the network device.
  • the resource parameters can be a combination of wireless channel conditions and AI parameters.
  • the terminal reporting the current resource parameters of the terminal to the network device is based on the terminal sending the first
  • the first indicator is used to indicate the combination of the terminal’s wireless channel conditions and AI parameters, so as to indicate to the network device that the current wireless channel conditions of the terminal and the combination of AI parameters are associated at the same time to ensure the resource allocation. reliability.
  • FIG. 1 is a schematic diagram of the architecture of a communication system provided by an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of a communication method provided by an embodiment of this application.
  • FIG. 3a is a schematic structural diagram of an AI operation uninstallation provided by an embodiment of this application.
  • FIG. 3b is a schematic structural diagram of an AI operation segmentation provided by an embodiment of this application.
  • FIG. 4 is a schematic structural diagram of the difference between the required terminal calculation amount and the available computing capacity of the terminal for different segmentation points according to an embodiment of the application;
  • FIG. 5 is a schematic flowchart of another communication method provided by an embodiment of this application.
  • FIG. 6 is a schematic structural diagram of a communication device provided by an embodiment of this application.
  • FIG. 7 is a schematic structural diagram of another communication device provided by an embodiment of this application.
  • FIG. 8 is a schematic structural diagram of a terminal provided by an embodiment of this application.
  • FIG. 9 is a schematic structural diagram of a network device provided by an embodiment of this application.
  • FIG. 1 is a schematic diagram of a communication architecture provided by an embodiment of the present application.
  • the communication system includes a terminal and a network device.
  • the communication system may include a network device 101 and one or more terminals 102.
  • the terminal 102 can communicate with the network device 101, and the terminal 102 and the terminal 102 can also communicate with each other.
  • the communication system can be Global System For Mobile Communication (GSM), Code Division Multiple Access (CDMA) system, Wideband Code Division Multiple Access (WCDMA) system, global Worldwide Interoperability For Microwave Access (WiMAX) systems, Long Term Evolution (LTE) systems, 5G communication systems (such as New Radio (NR)), communication systems that integrate multiple communication technologies ( For example, a communication system that integrates LTE technology and NR technology), or a subsequent evolution communication system.
  • GSM Global System For Mobile Communication
  • CDMA Code Division Multiple Access
  • WCDMA Wideband Code Division Multiple Access
  • WiMAX global Worldwide Interoperability For Microwave Access
  • LTE Long Term Evolution
  • 5G communication systems such as New Radio (NR)
  • NR New Radio
  • FIG. 1 The form and quantity of network devices and terminals shown in FIG. 1 are only for example, and do not constitute a limitation to the embodiments of the present application.
  • the terminals 102 may be dispersed in the entire communication system, and each terminal 102 may be stationary or mobile.
  • the terminal 102 can also be referred to by those skilled in the art as a mobile station, a subscriber station, a mobile unit, a subscriber unit, a wireless unit, a remote unit, a mobile device, a user equipment, a wireless device, a wireless communication device, a remote device, a mobile subscriber station, and a wireless communication device.
  • the terminal 102 may be a cellular phone, a personal digital assistant (PDA), a wireless modem, a wireless communication device, a handheld device, a tablet computer, a laptop computer, a cordless phone, and a wireless local loop (Wireless Local Loop, WLL) Stand and wait.
  • the terminal 102 can communicate with the network device 101 in the communication system.
  • the network device 101 may perform wireless communication with the terminal 102 via one or more antennas.
  • the network device 101 can provide communication coverage for its geographic area.
  • the network device 101 may include different types such as macro base stations, micro base stations, relay stations, and access points.
  • the base station may be referred to by those skilled in the art as a base station transceiver, a wireless base station, an access point, a wireless transceiver, a basic service set (Basic Service Set, BSS), and an extended service set (Extended Service Set, ESS). ), NodeB, evolved NodeB (eNB or eNodeB), or some other appropriate terms.
  • the base station is called gNB.
  • the network device can allocate resources to the terminal, and the terminal can communicate based on the allocated resources.
  • the terminal may send information indicating communication resource parameters and/or AI parameters (such as parameter information itself, or an indicator indicating corresponding parameter information, etc.) to the network device, and the network device may receive the information, and the network device may receive the information based on the received
  • the information allocated to the terminal to improve the reliability of resource allocation.
  • the terminal may periodically report the Channel Quality Indicator (CQI) to the network device, and the terminal may use the physical uplink control channel (PUCCH) or the physical uplink shared channel (Physical Uplink Control Channel, PUCCH) periodically.
  • the CQI is sent to the network device on the Shared Channel (PUSCH).
  • the network equipment adjusts the modulation mode, coding rate, and spectrum efficiency of the terminal according to the CQI report.
  • the CQI is used to reflect the channel quality of the Physical Downlink Shared Channel (PDSCH).
  • PDSCH Physical Downlink Shared Channel
  • Table 1 the channel quality of the PDSCH is represented by 0-15. 0 means the channel quality is the worst, and 15 means the channel quality is the best.
  • the terminal selects one of the 16 CQIs according to the result of the downlink channel estimation, and reports the CQI index to the network device.
  • the network equipment comprehensively considers various factors, and may consider the resource allocation problem among multiple terminals, and allocates the final resource parameters for the terminal, such as the modulation mode, coding rate and spectrum efficiency actually used by the terminal as shown in Table 2.
  • the terminal can report the current wireless channel condition of the terminal and the combination of AI parameters to the network device, for example, by reporting the indicator corresponding to the combination of the current wireless channel condition of the terminal and the AI parameter, that is, The indicator is used to indicate the combination of the current wireless channel conditions and AI parameters of the terminal, so that the network device can allocate resources according to the indicator, and allocate the best combination of wireless communication resource parameters and AI parameters to the terminal to realize wireless communication resource parameters The best match with the AI parameter configuration, to avoid the mismatch between the two, resulting in AI task quality degradation and resource waste.
  • the trigger condition for the terminal to send the indicator to the network device may be that the terminal sends the indicator to the network device when the terminal needs to perform an AI task, or the terminal periodically sends the indicator to the network device, etc.
  • the application examples are not limited.
  • the terminal can send an indicator to the network device when starting the AI task.
  • the terminal can send an indicator to the network device at a preset time interval.
  • the application of AI in the 5G system can be the following three application scenarios: segmented AI operation in the 5G system, download of the AI model in the 5G system, and training of the AI model in the 5G system.
  • the terminal and network equipment can cooperate to complete AI inference operations.
  • the terminal mainly runs relatively low-complexity calculations that are sensitive to delay and privacy protection
  • the network device mainly runs time synchronization. Delay and privacy-insensitive calculations of relatively high complexity, that is, the terminal first completes a part of the AI operation, reports the intermediate data to the network device, and then the network device completes the rest of the AI operation.
  • the terminal can download the required model from the network device for use.
  • the terminal first uses the global model provided by the mobile communication network for training, and then reports the trained local gradient of the terminal to the network device, and then the network device merges the local model of the terminal to form a more optimized global model.
  • FIG. 2 is a schematic flowchart of a communication method provided by an embodiment of the present application.
  • the method of this embodiment is applied to the above-mentioned communication system and includes the following steps:
  • the terminal sends the first indicator to the network device.
  • the terminal may send a first indicator to the network device, where the first indicator is used to identify a first resource parameter combination, the first resource parameter combination includes the first communication resource parameter and the first artificial intelligence AI parameter, and the first The indicator and the first resource parameter are combined into a first correspondence.
  • the first communication resource parameter may indicate the current wireless channel condition of the terminal, and specifically may be at least one of a modulation method, coding rate, and spectrum efficiency
  • the first AI parameter may indicate information related to the AI model of the terminal.
  • Parameters which can specifically be the computing power level of the terminal running AI operation, the AI model used by the terminal, the split point of the terminal running AI model, the steps of the terminal running AI operation, and the amount of training data required by the terminal to complete a round of AI training
  • the terminal uses at least one of the reporting cycles of the training result of the AI model.
  • the first indicator may be a digital number, a bit value, or other indication information, which is not limited in this application.
  • the terminal may obtain the first resource parameter combination, for example, obtain the first communication resource parameter according to channel estimation of the current wireless channel, and obtain the first AI parameter according to its own available computing power. Then, the first indicator is determined according to the first resource parameter combination.
  • the terminal may determine the first indicator according to the first resource parameter combination and the first correspondence, where the first correspondence is the correspondence between the first indicator and the first resource parameter combination, and the first resource parameter The combination may be the first communication resource parameter and the first AI parameter.
  • the first correspondence relationship that is, the correspondence relationship between the first indicator and the first resource parameter combination may be defined by a table or in other ways.
  • the first correspondence can be predefined, or can be configured by system information, or can be configured by radio resource control (Radio Resource Control, RRC) signaling, or can also be configured by Media Access Control Control Element (MAC CE) ) Configuration, this application is not limited.
  • RRC Radio Resource Control
  • MAC CE Media Access Control Control Element
  • the terminal may select a suitable combination of the first communication resource parameter and the first AI parameter according to the channel estimation of the current wireless channel and its own available computing power resources, and send the corresponding first indicator to Network equipment.
  • the combination of the first communication resource parameter and the first AI parameter may be a combination of any one of the first communication resource parameters and any one of the first AI parameters, and may be any one of the first communication resource parameters and the first AI parameter.
  • a combination of any number of AI parameters can be a combination of any number of the first communication resource parameters and any one of the first AI parameters, or can be a combination of any number of the first communication resource parameters and the first AI parameter.
  • a combination of any number of AI parameters can be a combination of any number of AI parameters.
  • the terminal may indicate according to downlink control information (DCI) or wireless Network Temporary Identity Radio Network Tempory Identity (RNTI) or DCI format, and determine the first correspondence from the multiple correspondences between the first indicator and the first resource parameter combination.
  • DCI downlink control information
  • RNTI wireless Network Temporary Identity Radio Network Tempory Identity
  • the network device may send DCI indication information to the terminal, the DCI indication information may be used to indicate the correspondence between the first indicator and the first resource parameter combination, and the terminal may receive the DCI indication information sent by the network device, and then Determine the first correspondence relationship according to the DCI indication information.
  • the terminal may determine the first correspondence relationship according to RNTI information, and the RNTI information may be used to indicate the correspondence relationship between the first indicator and the first resource parameter combination.
  • the network device may send DCI format information to the terminal, the DCI format information may be used to schedule the correspondence between the first indicator and the first resource parameter combination, and the terminal may receive the DCI format information sent by the network device, and then The first correspondence is determined according to the DCI format information.
  • the correspondences may be multiple correspondences between the first indicator and the first resource parameter combination, and the correspondences are all expressed in tables, such as Table 3 and Table 5. Comparing these two tables, it can be seen that Table 3 The first four columns in Table 5 all indicate the first indicator, modulation method, coding rate, and spectrum efficiency.
  • the fifth column in Table 3 indicates the computing power level of the terminal running AI operations, and the fifth column in Table 5 indicates the terminal operation.
  • the division point of the AI model can be based on the difference in the fifth column, that is, the difference in the first AI parameter, using the DCI indicator or RNTI or DCI format to determine the currently required table, that is, the currently required first indicator and the first indicator.
  • a correspondence between resource parameter combinations can be based on the difference in the fifth column, that is, the difference in the first AI parameter, using the DCI indicator or RNTI or DCI format to determine the currently required table, that is, the currently required first indicator and the first indicator.
  • the terminal may send the first indicator to the network device through PUCCH; or, it may send the first indicator to the network device through PUSCH; or, it may also send the first indicator to the network device through MAC CE.
  • An indicator may be used to indicate whether the first indicator is a packet is accessed or not.
  • the network device receives the first indicator sent by the terminal.
  • the network device may receive the first indicator sent by the terminal.
  • the network device may determine the first resource parameter combination of the terminal according to the correspondence between the first indicator and the first resource parameter combination, that is, the terminal Current wireless channel conditions and AI parameters.
  • the network device may indicate or indicate according to Downlink Control Information (DCI).
  • DCI Downlink Control Information
  • RNTI Radio Network Temporary Identifier Radio Network Tempory Identity
  • the first correspondence is determined from multiple correspondences between the first indicator and the first resource parameter combination.
  • the network device may determine the first correspondence relationship according to the DCI indication information, and the DCI indication information may be used to indicate the correspondence relationship between the first indicator and the first resource parameter combination.
  • the terminal may send RNTI information to the network device, the RNTI information may be used to indicate the correspondence between the first indicator and the first resource parameter combination, and the network device may receive the RNTI information sent by the terminal, and then according to the RNTI The information determines the first correspondence.
  • the network device may determine the first correspondence relationship according to the DCI format information, and the DCI format information may be used to schedule the correspondence relationship between the first indicator and the first resource parameter combination.
  • the terminal and the network device determine the downlink control information (Downlink Control Information, DCI) indication or wireless network used in the first correspondence from the multiple correspondences between the first indicator and the first resource parameter combination.
  • Downlink Control Information Downlink Control Information, DCI
  • RNTI Radio Network Temporary Identity
  • DCI format may be the same or different, which is not limited in this application.
  • the network device may receive the first indicator sent by the terminal through PUCCH; or, may receive the first indicator sent by the terminal through PUSCH; or, may also receive the first indicator sent by the terminal through MAC CE symbol.
  • the terminal sends a first indicator to the network device, and the network device receives the first indicator sent by the terminal, where the first indicator is used to identify a first resource parameter combination, and the first resource
  • the parameter combination includes a first communication resource parameter and a first artificial intelligence AI parameter, and the first indicator and the first resource parameter are combined into a first corresponding relationship, so that the network device indicates that the current wireless channel condition of the terminal is associated with the terminal at the same time.
  • AI parameter combination indicator to ensure the reliability of resource allocation.
  • FIG. 5 is a schematic flowchart of another communication method provided by an embodiment of the present application.
  • the method of this embodiment is applied to the above-mentioned communication system and includes the following steps:
  • the terminal sends a first indicator to the network device.
  • the network device receives the first indicator sent by the terminal.
  • steps 501 to 502 For the specific implementation of steps 501 to 502, reference may be made to the related descriptions in steps 201 to 202 in the above communication method, which will not be repeated here.
  • the network device sends the second indicator to the terminal.
  • the network device may send a second indicator to the terminal, where the second indicator is used to identify a second resource parameter combination, and the second resource parameter combination includes a second communication resource parameter and a second AI parameter, so The combination of the second indicator and the second resource parameter forms a second correspondence.
  • the second communication resource parameter may indicate the wireless channel resource scheduled by the network device to the terminal, and specifically may be at least one of modulation method, coding rate, and spectrum efficiency, and the second AI parameter may indicate that the network device sends the terminal to the terminal.
  • the parameters related to the scheduled terminal to run the AI model can be specifically the computing power level of the terminal running AI operation, the AI model used by the terminal, the split point of the terminal running AI model, the steps of the terminal running AI operation, and the terminal is completing a round of AI At least one of the amount of training data required in the training and the reporting period of the training result of the terminal using the AI model.
  • the second indicator may be a digital number, a bit value, or other indication information, which is not limited in this application.
  • the network device may determine the current first resource parameter combination of the terminal according to the first indicator sent by the terminal, and then determine the second resource parameter combination suitable for the terminal according to the first resource parameter combination.
  • the network device may determine the second indicator according to the second resource parameter combination and the second correspondence, where the second correspondence may be the correspondence between the second indicator and the second resource parameter combination, and the second The resource parameter combination may be the second communication resource parameter and the second AI parameter.
  • the second correspondence relationship that is, the correspondence relationship between the second indicator and the second resource parameter combination may be defined by a table or in other ways. For example, as shown in Table 4, a table is shown, and the table represents The corresponding relationship between the second indicator and the second resource parameter combination.
  • the first column represents the second indicator, and the last four columns represent the second resource parameter combination. It can be seen from the table that each second indicator The symbols have corresponding second resource parameter combinations.
  • the second correspondence can be predefined, or can be configured by system information, or can be configured by radio resource control (Radio Resource Control, RRC) signaling, or can also be configured by Media Access Control Control Element (MAC CE) ) Configuration, this application is not limited.
  • RRC Radio Resource Control
  • MAC CE Media Access Control Element
  • the network device may use the DCI indicator or RNTI or DCI format from the second indicator.
  • the second correspondence is determined from the multiple correspondences between the symbol and the second resource parameter combination.
  • the network device may determine the second correspondence relationship according to the DCI indication information, and the DCI indication information may be used to indicate the correspondence relationship between the second indicator and the second resource parameter combination.
  • the terminal may send RNTI information to the network device, the RNTI information may be used to indicate the correspondence between the second indicator and the second resource parameter combination, and the network device may receive the RNTI information sent by the terminal, and then according to the RNTI The information determines the second correspondence.
  • the network device may determine the second correspondence relationship according to the DCI format information, and the DCI format information may be used to schedule the correspondence relationship between the second indicator and the second resource parameter combination.
  • the correspondences may be multiple correspondences between the second indicator and the second resource parameter combination, and the correspondences are all expressed in tables, such as Table 4 and Table 6. Comparing these two tables, you can see Table 4
  • the first four columns in Table 6 all indicate the second indicator, modulation method, coding rate, and spectrum efficiency.
  • the fifth column in Table 4 indicates the computing power level of the AI operation of the terminal, and the fifth column in Table 6 indicates the terminal operation.
  • the division point of the AI model can be based on the difference in the fifth column, that is, the difference in the second AI parameter, using the DCI indicator or RNTI or DCI format to determine the currently required table, that is, the currently required second indicator and the first 2. Correspondence between resource parameter combinations.
  • the network device may send the second indicator to the terminal through the Physical Downlink Control Channel (PDCCH); alternatively, it may send the second indicator to the terminal through the PDSCH; or, it may also send the second indicator through the PDSCH.
  • the MAC CE sends the second indicator to the terminal.
  • the terminal receives the second indicator sent by the network device.
  • the terminal can receive the second indicator sent by the network device, and determine the second indicator of the terminal according to the correspondence between the second indicator and the second resource parameter combination.
  • the resource parameter combination that is, the terminal can use the second communication resource parameter and the second AI parameter scheduled by the network device.
  • the terminal may indicate according to downlink control information (DCI) or wireless Network Temporary Identity Radio Network Temporary Identity (RNTI) or DCI format, and determine the second correspondence from the multiple correspondences between the second indicator and the second resource parameter combination.
  • DCI downlink control information
  • RNTI wireless Network Temporary Identity Radio Network Temporary Identity
  • the network device may send DCI indication information to the terminal, the DCI indication information may be used for the correspondence between the second indicator and the second resource parameter combination, and the terminal may receive the DCI indication information sent by the network device, and then according to The DCI second indicator determines the second correspondence.
  • the terminal may determine the second correspondence relationship according to RNTI information, and the RNTI information may be used to indicate the correspondence relationship between the second indicator and the second resource parameter combination.
  • the network device may send DCI format information to the terminal, and the DCI format information may be used to schedule the correspondence between the second indicator and the second resource parameter combination, and the terminal may receive the DCI format information sent by the network device, and then The second correspondence is determined according to the DCI format information.
  • the terminal and the network device determine the downlink control information (Downlink Control Information, DCI) indication or wireless network used for determining the second correspondence from the multiple correspondences between the second indicator and the second resource parameter combination.
  • Downlink Control Information Downlink Control Information
  • RNTI Temporary Identifier Radio Network Temporary Identity
  • DCI format may be the same or different, which is not limited in this application.
  • the available computing power of the terminal may affect the quality of transmission of a certain AI task between the terminal and the network device.
  • the terminal can report the current wireless channel conditions and available computing power to the network device.
  • Table 3 shows the correspondence between the first indicator and the first resource parameter combination, where the first resource parameter combination includes the modulation method, coding rate, and frequency spectrum in the first communication resource parameter.
  • the computing power level can be formed into several levels based on pre-agreed rules, corresponding to a certain computing power level.
  • Level 1 there are two levels of terminal computing power: Level 1 and Level 2. Different first indicators in Table 3 can be mapped to Level 1 or Level 2. Specifically, the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency, and available computing power from Table 3 according to the channel estimation of the current wireless channel and its own available computing power, and then assign its corresponding first indicator Send to the network device.
  • the network device can learn the current first communication resource parameter of the terminal and the computing power level of the terminal running AI operation according to the first indicator sent by the terminal, and then according to the first communication resource parameter and the computing power level of the terminal running AI operation Comprehensively determine the best second communication resource parameter and the computing power level of the terminal running AI operation, and then send the second communication resource parameter and the second indicator corresponding to the computing power level of the terminal running AI operation to the terminal, and the terminal receives the first
  • the second communication resource parameter and available computing power scheduled by the network device can be used to complete the AI task.
  • Table 4 shows the correspondence between the second indicator and the second resource parameter combination, where the second resource parameter combination includes the modulation method, coding rate, and frequency spectrum in the second communication resource parameter.
  • the corresponding second AI operation can be determined according to Table 4. indicator.
  • First indicator Modulation method Encoding rate Spectral efficiency Computing power level 0 QPSK Rate 0 Efficiency 0 Level 1 1 QPSK Rate 1 Efficiency 1 Level 1 2 QPSK Rate 2 Efficiency 2 Level 1 3 QPSK Rate 3 Efficiency 3 Level 1
  • Second indicator Modulation method Encoding rate Spectral efficiency Computing power level 0 2 Rate 0 Efficiency 0 Level 1 1 2 Rate 1 Efficiency 1 Level 1 2 2 Rate 2 Efficiency 2 Level 1 3 2 Rate 3 Efficiency 3 Level 1 4 2 Rate 4 Efficiency 4 Level 1 5 2 Rate 5 Efficiency 5 Level 1
  • Table 3 there is no one-to-one correspondence between Table 3 and Table 4.
  • Table 3 and Table 4 are the representation methods of the first correspondence and the second correspondence, respectively. The same is true for the table shown below.
  • the terminal runs relatively low-complexity calculations that are sensitive to delay and privacy protection, and the network device runs relatively insensitive to delay and privacy.
  • High-complexity calculations are shown in Figures 3a and 3b.
  • Figure 4 shows a typical convolutional neural network (Convolutional Neural Networks, CNN) structure. It can be seen from Figure 4 that there are some optional split points in the CNN, and the terminal only executes a part of the layer. The calculation (the calculation amount is the calculation amount level 1 and 2 respectively) until the split point, and then the intermediate data amount is transmitted to the network device, and the network device performs the calculation of the remaining layers.
  • CNN convolutional Neural Networks
  • the selection of the split point can affect the wireless communication resources that the network device needs to allocate to the terminal.
  • the terminal can report current wireless channel conditions and split points to the network device. For example, as shown in Table 5, Table 5 shows the correspondence between the first indicator and the first resource parameter combination, where the first resource parameter combination includes the modulation method, coding rate, and frequency spectrum in the first communication resource parameter.
  • the efficiency and the split point of the terminal running the AI model in the first AI parameter as shown in Table 5, there are two choices for the split point of the AI model: Split point 1 or Split point 2.
  • the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency, and split point from Table 5 according to the channel estimation of the current wireless channel and its own available computing power resources, and then send its corresponding first indicator To network equipment.
  • the network device can learn the current first communication resource parameter of the terminal and the split point of the terminal running AI model according to the first indicator sent by the terminal, and then comprehensively determine it according to the first communication resource parameter and the split point of the terminal running AI model
  • the optimal second communication resource parameter and the split point of the terminal running AI model, and then the second indicator corresponding to the second communication resource parameter and the split point of the terminal running AI model is sent to the terminal.
  • the second communication resource parameter scheduled by the network device and the split point of the terminal running AI model can be used to complete the AI task. For example, as shown in Table 6, Table 6 shows the correspondence between the second indicator and the second resource parameter combination.
  • the second resource parameter combination includes the modulation method, coding rate, and spectral efficiency in the second communication resource parameter and the split point of the terminal operating AI model in the second AI parameter
  • the network device determines the best second After the communication resource parameters and the split point of the terminal running the AI model, the corresponding second indicator can be determined according to Table 6.
  • Second indicator Modulation method Encoding rate Spectral efficiency Split point 0 2 Rate 0 Efficiency 0 Split point 1 1 2 Rate 1 Efficiency 1 Split point 1 2 2 Rate 2 Efficiency 2 Split point 1 3 2 Rate 3 Efficiency 3 Split point 1 4 2 Rate 4 Efficiency 4 Split point 1 5 2 Rate 5 Efficiency 5 Split point 1 6 2 Rate 6 Efficiency 6 Split point 1 7 4 Rate 7 Efficiency 7 Split point 1 8 4 Rate 8 Efficiency 8 Split point 1 9 4 Rate 9 Efficiency 9 Split point 1 10 6 Rate 10 Efficiency 10 Split point 1 11 6 Rate 11 Efficiency 11 Split point 1 12 6 Rate 12 Efficiency 12 Split point 1 13 6 Rate 13 Efficiency 13 Split point 1 14 6 Rate 14 Efficiency 14 Split point 1 15 6 Rate 15 Efficiency 15 Split point 1 16 2 Rate 0 Efficiency 0 Split point 2 17 2 Rate 1 Efficiency 1 Split point 2 18 2 Rate 2 Efficiency 2 Split point 2 19 2 Rate 3 Efficiency 3 Split point 2 20 2 Rate 4 Efficiency 4 Split point 2 twenty one 2 Rate 5 Efficiency 5 Split point 2 twenty two 2 Rate 6 Efficiency 6 Split point 2
  • the AI operation segmentation method may be to segment different steps or different parts of an AI process to the terminal and network equipment for operation, and different segmentation methods output
  • the amount of intermediate data is also very different.
  • Table 7 shows the correspondence between the first indicator and the first resource parameter combination, where the first resource parameter combination is the modulation method, coding rate, and frequency spectrum in the first communication resource parameter.
  • the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency and procedure from Table 7 according to the channel estimation of the current wireless channel and its own available computing power resources, and send its corresponding first indicator to Network equipment. Further, the network device determines the second communication resource parameter suitable for the terminal and the step for the terminal to run the AI operation according to the first indicator reported by the terminal, and sends the corresponding second indicator to the terminal, as shown in Table 8. After the terminal receives the second indicator, it can use the second communication resource parameter scheduled by the network device and the steps for the terminal to execute the AI operation to complete the AI task.
  • Step 1-4 64QAM Rate 14 Efficiency 14 Step 1-2 15 64QAM Rate 15 Efficiency 15 Step 1-2 16 QPSK Rate 0 Efficiency 0 Step 1-4 17 QPSK Rate 1 Efficiency 1 Step 1-4 18 QPSK Rate 2 Efficiency 2 Step 1-4 19 QPSK Rate 3 Efficiency 3 Step 1-4 20 QPSK Rate 4 Efficiency 4 Step 1-4 twenty one QPSK Rate 5 Efficiency 5 Step 1-4 twenty two QPSK Rate 6 Efficiency 6 Step 1-4 twenty three 16QAM Rate 7 Efficiency 7 Step 1-4 twenty four 16QAM Rate 8 Efficiency 8 Step 1-4 25 16QAM Rate 9 Efficiency 9 Step 1-4 26 64QAM Rate 10 Efficiency 10 Step 1-4 27 64QAM Rate 11 Efficiency 11 Step 1-4 28 64QAM Rate 12 Efficiency 12 Step 1-4 29 64QAM Rate 13 Efficiency 13 Step 1-4 30 64QAM Rate 14 Efficiency 14 Step 1-4 31 64QAM Rate 15 Efficiency 15 Step 1-4
  • Second indicator Modulation method Encoding rate Spectral efficiency step 0 2 Rate 0 Efficiency 0 Step 1-2 1 2 Rate 1 Efficiency 1 Step 1-2 2 2 Rate 2 Efficiency 2 Step 1-2 3 2 Rate 3 Efficiency 3 Step 1-2 4 2 Rate 4 Efficiency 4 Step 1-2 5 2 Rate 5 Efficiency 5 Step 1-2 6 2 Rate 6 Efficiency 6 Step 1-2 7 4 Rate 7 Efficiency 7 Step 1-2 8 4 Rate 8 Efficiency 8 Step 1-2 9 4 Rate 9 Efficiency 9 Step 1-2 10 6 Rate 10 Efficiency 10 Step 1-2 11 6 Rate 11 Efficiency 11 Step 1-2 12 6 Rate 12 Efficiency 12 Step 1-2 13 6 Rate 13 Efficiency 13 Step 1-2 14 6 Rate 14 Efficiency 14 Step 1-2 15 6 Rate 15 Efficiency 15 Step 1-2
  • Step 1-4 16 2 Rate 0 Efficiency 0 Step 1-4 17 2 Rate 1 Efficiency 1 Step 1-4 18 2 Rate 2 Efficiency 2 Step 1-4 19 2 Rate 3 Efficiency 3 Step 1-4 20 2 Rate 4 Efficiency 4 Step 1-4 twenty one 2 Rate 5 Efficiency 5 Step 1-4 twenty two 2 Rate 6 Efficiency 6 Step 1-4 twenty three 4 Rate 7 Efficiency 7 Step 1-4 twenty four 4 Rate 8 Efficiency 8 Step 1-4 25 4 Rate 9 Efficiency 9 Step 1-4 26 6 Rate 10 Efficiency 10 Step 1-4 27 6 Rate 11 Efficiency 11 Step 1-4 28 6 Rate 12 Efficiency 12 Step 1-4 29 6 Rate 13 Efficiency 13 Step 1-4 30 6 Rate 14 Efficiency 14 Step 1-4 31 6 Rate 15 Efficiency 15 Step 1-4
  • the first AI parameter is the AI model used by the terminal
  • Table 9 shows the correspondence between the first indicator and the first resource parameter combination, where the first The resource parameter combination is the combination of the modulation method, coding rate and spectrum efficiency in the first communication resource parameter and the AI model used by the terminal in the first AI parameter.
  • the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency, and AI model suitable for use by the terminal from Table 9 according to the channel estimation of the current wireless channel and its own available computing resources, and assign it to the first The indicator is sent to the network device.
  • the network device determines the second communication resource parameter suitable for the terminal and the AI model used by the terminal according to the first indicator reported by the terminal, and sends the corresponding second indicator to the terminal, as shown in Table 10. After receiving the second indicator, the terminal can use the second communication resource parameter and the AI model scheduled by the network device to complete the AI task.
  • Second indicator Modulation method Encoding rate Spectral efficiency Model 0 QPSK Rate 0 Efficiency 0 Model #1 1 QPSK Rate 1 Efficiency 1 Model #1 2 QPSK Rate 2 Efficiency 2 Model #1 3 QPSK Rate 3 Efficiency 3 Model #1 4 QPSK Rate 4 Efficiency 4 Model #1 5 QPSK Rate 5 Efficiency 5 Model #1 6 QPSK Rate 6 Efficiency 6 Model #1 7 16QAM Rate 7 Efficiency 7 Model #1 8 16QAM Rate 8 Efficiency 8 Model #1
  • Second indicator Modulation method Encoding rate Spectral efficiency Model 0 2 Rate 0 Efficiency 0 Model #1 1 2 Rate 1 Efficiency 1 Model #1 2 2 Rate 2 Efficiency 2 Model #1 3 2 Rate 3 Efficiency 3 Model #1 4 2 Rate 4 Efficiency 4 Model #1 5 2 Rate 5 Efficiency 5 Model #1 6 2 Rate 6 Efficiency 6 Model #1 7 4 Rate 7 Efficiency 7 Model #1 8 4 Rate 8 Efficiency 8 Model #1 9 4 Rate 9 Efficiency 9 Model #1 10 6 Rate 10 Efficiency 10 Model #1
  • the first AI parameter is the amount of training data required by the terminal to complete a round of AI training
  • Table 11 shows the relationship between the first indicator and the first resource parameter combination
  • the first resource parameter combination is the modulation method, coding rate, and spectral efficiency in the first communication resource parameter and the amount of training data required by the terminal in the first AI parameter to complete a round of AI training Combinations, as shown in Table 11, there are two types of AI/ML training data that the terminal can complete in one round of training: Data size 1 and Data size 2.
  • the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency, and appropriate AI training data amount from Table 11 according to the channel estimation of the current wireless channel and its own available computing power resources, and assign it to the first
  • the indicator is sent to the network device.
  • the network device determines the second communication resource parameter and the amount of training data suitable for the terminal according to the first indicator reported by the terminal, and sends the corresponding second indicator to the terminal, as shown in Table 12.
  • the terminal can use the first communication resource parameter and the amount of training data scheduled by the network device to complete the AI task.
  • First indicator Modulation method Encoding rate Spectral efficiency Training data 0 QPSK Rate 0 Efficiency 0 Data size 1 1 QPSK Rate 1 Efficiency 1 Data size 1 2 QPSK Rate 2 Efficiency 2 Data size 1
  • Second indicator Modulation method Encoding rate Spectral efficiency Training data 0 2 Rate 0 Efficiency 0 Data size 1 1 2 Rate 1 Efficiency 1 Data size 1 2 2 Rate 2 Efficiency 2 Data size 1 3 2 Rate 3 Efficiency 3 Data size 1 4 2 Rate 4 Efficiency 4 Data size 1
  • the first AI parameter for the reporting period of the training result of the terminal using the AI model, for example, as shown in Table 13, Table 13 shows the correspondence between the first indicator and the first resource parameter combination, where , The first resource parameter combination is the combination of the modulation method, coding rate and spectrum efficiency in the first communication resource parameter and the training result reporting period in the first AI parameter.
  • the terminal can complete one round of training
  • the terminal can select the appropriate modulation mode, coding rate, spectrum efficiency, and training result reporting period based on the channel estimation of the current wireless channel and its own available computing resources, and send its corresponding first indicator to the network device .
  • the network device determines the second communication resource parameter and the training result reporting period suitable for the terminal according to the first indicator reported by the terminal, and sends the corresponding second indicator to the terminal, as shown in Table 14. After receiving the second indicator, the terminal can use the first communication resource parameter and the training result reporting period scheduled by the network device to complete the AI task.
  • Second indicator Modulation method Encoding rate Spectral efficiency Training result reporting cycle 0 2 Rate 0 Efficiency 0 Period 1 1 2 Rate 1 Efficiency 1 Period 1 2 2 Rate 2 Efficiency 2 Period 1 3 2 Rate 3 Efficiency 3 Period 1 4 2 Rate 4 Efficiency 4 Period 1 5 2 Rate 5 Efficiency 5 Period 1 6 2 Rate 6 Efficiency 6 Period 1 7 4 Rate 7 Efficiency 7 Period 1 8 4 Rate 8 Efficiency 8 Period 1 9 4 Rate 9 Efficiency 9 Period 1 10 6 Rate 10 Efficiency 10 Period 1 11 6 Rate 11 Efficiency 11 Period 1 12 6 Rate 12 Efficiency 12 Period 1 13 6 Rate 13 Efficiency 13 Period 1 14 6 Rate 14 Efficiency 14 Period 1 15 6 Rate 15 Efficiency 15 Period 1 16 2 Rate 0 Efficiency 0 Period 2 17 2 Rate 1 Efficiency 1 Period 2 18 2 Rate 2 Efficiency 2 Period 2 19 2 Rate 3 Efficiency 3 Period 2 20 2 Rate 4 Efficiency 4 Period 2 twenty one 2 Rate 5 Efficiency 5 Period 2 twenty two 2 Rate 6 Efficiency 6 Period 2 twenty three 4 Rate 7 Efficiency 7 Period 2 twenty four 4 Rate 8 Efficiency 8 Period 2 25 4 Rate 9 Efficiency 9 Period 2 26 6 Rate 10 Efficiency 10 Period 2 27 6 Rate 11 Efficiency 11 Period 2 28 6 Rate 12 Efficiency
  • the terminal may select a suitable modulation method, coding rate, and computing power level for AI operation of the terminal based on the channel estimation of the current wireless channel and its own available computing power resources, and send the corresponding first indicator to the network Device, further, the network device determines the appropriate modulation mode, coding rate, and computing power level of AI operation of the terminal according to the first indicator reported by the terminal, or the terminal’s modulation mode, coding rate, The terminal runs the computing power level of the AI operation and the AI model that the terminal is suitable for use, and sends the corresponding second indicator to the terminal. After receiving the second indicator, the terminal can use the first communication resource parameters and parameters scheduled by the network device. AI tasks are completed in the reporting cycle of training results.
  • the terminal may receive the second indicator sent by the network device through the PDCCH; or, it may receive the second indicator sent by the PDSCH network device; or, it may also receive the second indicator sent by the network device through the MAC CE. Two indicators.
  • the terminal sends a first indicator to the network device, and the network device receives the first indicator sent by the terminal, where the first indicator is used to identify the first resource parameter combination, and the first resource parameter combination includes the first resource parameter combination.
  • a communication resource parameter and a first artificial intelligence AI parameter, the first indicator and the first resource parameter are combined into a first corresponding relationship, and further, the network device sends a second indicator to the terminal, and the terminal receives the second instruction sent by the network device
  • the second indicator is used to identify the second resource parameter combination, the second resource parameter combination includes the second communication resource parameter and the second AI parameter, and the second indicator and the second resource parameter combination form a second correspondence relationship, thereby
  • the network equipment can schedule the best combination of wireless communication resource parameters and AI parameters for the terminal according to the wireless channel conditions and AI parameters reported by the terminal, so as to achieve the best match between the wireless communication resource parameters and the AI parameter configuration, and avoid disagreement between the two. AI service quality degradation and resource waste caused by matching.
  • FIG. 6 is a schematic structural diagram of a communication device provided by an embodiment of the present application, which is applied to a terminal, and the device includes:
  • the sending unit 601 is configured to send a first indicator to the network device by the terminal, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first manual resource parameter combination. Intelligent AI parameters, the first indicator and the first resource parameter are combined into a first correspondence.
  • the device further includes a receiving unit 602, specifically configured to:
  • the first AI parameter and/or the second AI parameter includes at least one of the following: the computing power level at which the terminal runs AI operations, the AI model used by the terminal, and the The segmentation point of the terminal running the AI model, the steps of the terminal running the AI operation, the amount of training data required by the terminal to complete a round of AI training, and the reporting period of the training result of the terminal using the AI model.
  • the first communication resource parameter and/or the second communication resource parameter includes at least one of the following: modulation method, coding rate, and spectrum efficiency.
  • the first corresponding relationship is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration;
  • the second correspondence is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration.
  • the device further includes a determining unit 603, specifically configured to:
  • the terminal determines the first correspondence from a plurality of correspondences between the first indicator and the first resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format; and,
  • the terminal determines the second correspondence from multiple correspondences between the second indicator and the second resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format.
  • the sending unit 601 is specifically configured to:
  • the terminal sends the first indicator to the network device through the physical uplink control channel PUCCH;
  • the terminal sends the first indicator to the network device through the physical uplink shared channel PUSCH;
  • the terminal sends the first indicator to the network device through the media access control unit MAC CE.
  • the receiving unit 602 is specifically configured to:
  • the terminal receives the second indicator sent by the network device through the physical downlink shared channel PDSCH;
  • the terminal receives the second indicator sent by the network device through a media access control unit MAC CE.
  • the division of units in the embodiments of the present application is illustrative, and is only a logical function division, and there may be other division methods in actual implementation.
  • the functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.
  • FIG. 7 is a schematic structural diagram of a communication device provided by an embodiment of the present application. Applied to network equipment, the device includes:
  • the receiving unit 701 is configured to receive a first indicator sent by a terminal by a network device, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first resource parameter combination. Artificial intelligence AI parameters, the first indicator and the first resource parameter are combined into a first correspondence.
  • the device further includes a sending unit 702, specifically configured to:
  • the network device sends a second indicator to the terminal, where the second indicator is used to identify a second resource parameter combination, and the second resource parameter combination includes a second communication resource parameter and a second AI parameter, so The combination of the second indicator and the second resource parameter forms a second correspondence.
  • the first AI parameter and/or the second AI parameter includes at least one of the following: the computing power level at which the terminal runs AI operations, the AI model used by the terminal, and the The segmentation point of the terminal running the AI model, the steps of the terminal running the AI operation, the amount of training data required by the terminal to complete a round of AI training, and the reporting period of the training result of the terminal using the AI model.
  • the first communication resource parameter and/or the second communication resource parameter include at least one of the following: modulation method, coding rate, and spectrum efficiency.
  • the first corresponding relationship is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration;
  • the second correspondence is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration.
  • the device further includes a determining unit 703, specifically configured to:
  • the terminal determines the first correspondence from a plurality of correspondences between the first indicator and the first resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format; and,
  • the terminal determines the second correspondence from multiple correspondences between the second indicator and the second resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format.
  • the receiving unit 701 is specifically configured to:
  • the network device receives the first indicator sent by the terminal through the physical uplink control channel PUCCH;
  • the network device receives the first indicator sent by the terminal through the physical uplink shared channel PUSCH;
  • the network device receives the first indicator sent by the terminal through a media access control unit MAC CE.
  • the sending unit 702 is specifically configured to:
  • the network device sends the second indicator to the terminal through the physical downlink control channel PDCCH;
  • the network device sends the second indicator to the terminal through the physical downlink shared channel PDSCH;
  • the network device sends the second indicator to the terminal through a media access control unit MAC CE.
  • the division of units in the embodiments of the present application is illustrative, and is only a logical function division, and there may be other division methods in actual implementation.
  • the functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.
  • FIG. 8 is a schematic structural diagram of a terminal provided by an embodiment of the application.
  • the terminal shown in FIG. 8 may include: a processor 801, a receiver 802, a transmitter 803, and a memory 804.
  • the processor 801, The receiver 802, the transmitter 803, and the memory 804 are respectively connected to a bus, where the memory 804 is used to store a computer program, and the computer program includes program instructions, and the processor 801 is used to execute the program instructions stored in the memory 804.
  • the processor 801 executes the following operations by running the executable program code in the memory 804:
  • the terminal sends a first indicator to the network device through the transmitter 803, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first artificial intelligence AI Parameters, the first indicator and the first resource parameter are combined into a first correspondence.
  • the processor 801 is further configured to:
  • the terminal receives the second indicator sent by the network device through the receiver 802, where the second indicator is used to identify a second resource parameter combination, and the second resource parameter combination includes a second communication resource parameter and a second communication resource parameter.
  • Two AI parameters, where the second indicator and the second resource parameter are combined into a second correspondence.
  • the first AI parameter and/or the second AI parameter includes at least one of the following: the computing power level at which the terminal runs AI operations, the AI model used by the terminal, and the The segmentation point of the terminal running the AI model, the steps of the terminal running the AI operation, the amount of training data required by the terminal to complete a round of AI training, and the reporting period of the training result of the terminal using the AI model.
  • the first communication resource parameter and/or the second communication resource parameter includes at least one of the following: modulation method, coding rate, and spectrum efficiency.
  • the first corresponding relationship is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration;
  • the second correspondence is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration.
  • the processor 801 is further configured to:
  • the terminal determines the first correspondence from a plurality of correspondences between the first indicator and the first resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format; and,
  • the terminal determines the second correspondence from multiple correspondences between the second indicator and the second resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format.
  • the processor 801 is specifically configured to:
  • the terminal sends the first indicator to the network device through the transmitter 803 through the physical uplink control channel PUCCH;
  • the terminal sends the first indicator to the network device through the transmitter 803 through the physical uplink shared channel PUSCH;
  • the terminal sends the first indicator to the network device through the transmitter 803 through the media access control unit MAC CE.
  • the processor 801 is specifically configured to:
  • the terminal receives the second indicator sent by the network device through the physical downlink control channel PDCCH through the receiver 802;
  • the terminal receives the second indicator sent by the network device through the receiver 802 through the physical downlink shared channel PDSCH;
  • the terminal receives the second indicator sent by the network device through the receiver 802 through the media access control unit MAC CE.
  • FIG. 9 is a schematic structural diagram of a network device provided by an embodiment of the application.
  • the network device shown in FIG. 9 may include: a processor 901, a receiver 902, a transmitter 903, and a memory 904.
  • the 901, the receiver 902, the transmitter 903, and the memory 904 are respectively connected to a bus, where the memory 904 is used to store a computer program, and the computer program includes program instructions, and the processor 901 is used to execute the program instructions stored in the memory 904.
  • the processor 901 executes the following operations by running the executable program code in the memory 904:
  • the network device receives the first indicator sent by the terminal through the receiver 902, where the first indicator is used to identify a first resource parameter combination, and the first resource parameter combination includes a first communication resource parameter and a first artificial intelligence.
  • An AI parameter where the first indicator and the first resource parameter are combined into a first correspondence.
  • the processor 901 is further configured to:
  • the network device sends a second indicator to the terminal through the transmitter 903, where the second indicator is used to identify a second resource parameter combination, and the second resource parameter combination includes a second communication resource parameter and a second resource parameter combination.
  • AI parameter, the second indicator and the second resource parameter are combined into a second correspondence.
  • the first AI parameter and/or the second AI parameter includes at least one of the following: the computing power level at which the terminal runs AI operations, the AI model used by the terminal, and the The segmentation point of the terminal running the AI model, the steps of the terminal running the AI operation, the amount of training data required by the terminal to complete a round of AI training, and the reporting period of the training result of the terminal using the AI model.
  • the first communication resource parameter and/or the second communication resource parameter include at least one of the following: modulation method, coding rate, and spectrum efficiency.
  • the first corresponding relationship is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration;
  • the second correspondence is obtained by predefined or radio resource control RRC signaling configuration or system information configuration or media access control unit MAC CE configuration.
  • the processor 901 is further configured to:
  • the terminal determines the first correspondence from a plurality of correspondences between the first indicator and the first resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format; and,
  • the terminal determines the second correspondence from multiple correspondences between the second indicator and the second resource parameter combination according to the downlink control information DCI indication or the radio network temporary identifier RNTI or the DCI format.
  • the processor 901 is specifically configured to:
  • the network device receives the first indicator sent by the terminal through the receiver 902 through the physical uplink control channel PUCCH;
  • the network device receives the first indicator sent by the terminal through the receiver 902 through the physical uplink shared channel PUSCH;
  • the network device receives the first indicator sent by the terminal through the receiver 902 through the media access control unit MAC CE.
  • the processor 901 is specifically configured to:
  • the network device sends the second indicator to the terminal through the receiver 903 through the physical downlink control channel PDCCH;
  • the network device sends the second indicator to the terminal through the receiver 903 through the physical downlink shared channel PDSCH;
  • the network device sends the second indicator to the terminal through the receiver 903 through the media access control unit MAC CE.
  • An embodiment of the present application also provides a computer storage medium, the computer storage medium stores program instructions, and the program execution may include part or all of the steps of the communication method in the embodiment corresponding to FIG. 2 or FIG. 5.
  • the embodiments of the present application also provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the method embodiments described above. Part or all of the steps described in the terminal or network device.
  • the computer program product may be a software installation package.
  • the steps of the method or algorithm described in combination with the disclosure of the present application may be implemented in a hardware manner, or may be implemented in a manner in which a processor executes software instructions.
  • Software instructions can be composed of corresponding software modules, which can be stored in random access memory (Random Access Memory, RAM), flash memory, read-only memory (Read Only Memory, ROM), and erasable programmable read-only memory ( Erasable Programmable ROM (EPROM), Electrically Erasable Programmable Read-Only Memory (Electrically EPROM, EEPROM), register, hard disk, mobile hard disk, CD-ROM or any other form of storage medium known in the art.
  • An exemplary storage medium is coupled to the processor, so that the processor can read information from the storage medium and write information to the storage medium.
  • the storage medium may also be an integral part of the processor.
  • the processor and the storage medium may be located in the ASIC.
  • the ASIC may be located in a communication device such as a terminal.
  • the processor and the storage medium may also exist as discrete components in the communication device.
  • the program can be stored in a computer readable storage medium.
  • the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

Abstract

本申请提供一种通信方法及相关设备，所述方法包括：终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系，从而向网络设备指示同时关联终端当前的无线信道条件和AI参数组合的指示符，以确保资源分配的可靠性。

Description

一种通信方法及相关设备 技术领域
本申请涉及无线网络通信技术领域，尤其涉及一种通信方法及相关设备。
背景技术
随着无线网络通信技术的发展，人工智能(Artifical Intelligence，AI)正在移动通信终端中承担越来越多的重要任务。针对AI场景，如何进行资源指示以提升资源分配的可靠性成为急需解决的问题。
发明内容
本申请实施例提供了一种通信方法及相关设备，能够向网络设备指示同时关联终端当前的无线信道条件和AI参数组合的指示符，以确保资源分配的可靠性。
本申请实施例第一方面提供了一种通信方法，包括：
终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
本申请实施例第二方面提供了另一种通信方法，包括：
网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
本申请实施例第三方面提供了一种通信装置，包括：
发送单元，用于终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
本申请实施例第四方面提供了一种通信装置，包括：
接收单元，用于网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
本申请实施例第五方面提供了一种终端，包括处理器、接收器、发射器和存储器，所述处理器、接收器、发射器和存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行上述第一方面的方法。
本申请实施例第六方面提供了一种网络设备，包括处理器、接收器、发射器和存储器，所述处理器、接收器、发射器和存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行上述第二方面的方法。
本申请实施例第七方面提供了一种计算机可读存储介质，其特征在于，所述计算机存储介质存储有计算机程序，所述计算机程序包括程序指令，所述程序指令当被处理器执行时使所述处理器执行上述第一方面的方法。
本申请实施例第八方面提供了一种计算机可读存储介质，其特征在于，所述计算机存储介质存储有计算机程序，所述计算机程序包括程序指令，所述程序指令当被处理器执行时使所述处理器执行上述第二方面的方法。
通过实施本申请实施例，终端可以向网络设备上报终端当前的资源参数，资源参数可以是无线信道条件和AI参数的组合，终端向网络设备上报终端当前的资源参数是根据终端向网络设备发送第一指示符来实现，第一指示符用于指示终端的无线信道条件和AI参数的组合，从而向网络设备指示同时关联终端当前的无线信道条件和AI参数组合的指示符，以确保资源分配的可靠性。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍。
图1是本申请实施例提供的一种通信系统的架构示意图；
图2为本申请实施例提供的一种通信方法的流程示意图；
图3a为本申请实施例提供的一种AI操作卸载的结构示意图；
图3b为本申请实施例提供的一种AI操作分割的结构示意图；
图4为本申请实施例提供的一种不同分割点的所需终端计算量与终端可用计算能力的差别的结构示意图；
图5为本申请实施例提供的另一种通信方法的流程示意图；
图6为本申请实施例提供的一种通信装置的结构示意图；
图7为本申请实施例提供的另一种通信装置的结构示意图；
图8为本申请实施例提供的一种终端的结构示意图；
图9为本申请实施例提供的一种网络设备的结构示意图。
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清楚、详细的描述。
请参见图1，图1是本申请实施例提供的一种通信架构示意图。所述通信系统包括终端和网络设备，如图1所示，该通信系统可以包括一个网络设备101、以及一个或者多个终端102。终端102可以与网络设备101进行通信，终端102与终端102也可以相互进行通信。该通信系统可以是全球移动通信系统(Global System For Mobile Communication，GSM)、码分多址(Code Division Multiple Access，CDMA)系统、宽带码分多址(Wideband Code Division Multiple Access，WCDMA)系统、全球微波互联接入(Worldwide Interoperability For Microwave Access，WiMAX)系统、长期演进(Long Term Evolution，LTE)系统、5G通信系统(例如新空口(New Radio，NR))、多种通信技术融合的通信系统(例如LTE技术和NR技术融合的通信系统)、或者后续演进通信系统。
图1中所示的网络设备和终端的形态和数量仅用于举例，并不构成对本申请实施例的限定。
其中，终端102可以散布于整个通信系统中，并且每个终端102可以是静止的或者移动的。终端102还可以被本领域技术人员称为移动站、用户站、移动单元、用户单元、无线单元、远程单元、移动设备、用户设备、无线设备、无线通信设备、远程设备、移动用户站、接入终端、移动终端、无线终端、远程终端、手持设备、用户代理、移动客户端、客户端或者一些其它适当的术语。终端102可以是蜂窝电话、个人数字助理(Personal Digital Assistant，PDA)、无线调制解调器、无线通信设备、手持设备、平板电脑、膝上型计算机、无绳电话、无线本地环路(Wireless Local Loop，WLL)站等。终端102能够与通信系统中的网络设备101进行通信。
其中，网络设备101可以经由一个或多个天线与终端102进行无线通信。网络设备101可以为其所在地理区域提供通信覆盖。所述网络设备101可以包括宏基站，微基站，中继站，接入点等不同类型。在一些实施例中，基站可以被本领域技术人员称为基站收发机、无线基站、接入点、无线收发机、基本服务集(Basic Service Set，BSS)、扩展服务集(Extended Service Set，ESS)、节点B(NodeB)、演进的节点B(evolved NodeB，eNB或eNodeB)或者其它一些适当的术语。示例性地，在5G系统中，基站被称为gNB。
网络设备可以给终端分配资源，进而终端可以基于分配的资源进行通信。可选的，终端可以向网络设备发送指示通信资源参数和/或AI参数的信息(比如参数信息本身，或者指示对应参数信息的指示符等等)，网络设备可接收该信息，并可基于接收的信息为终端分配资源，以提升资源分配的可靠性。
例如，在一些场景中，终端可周期性向网络设备上报信道质量指示符(Channel Quality Indicator，CQI)，终端可以在在物理上行控制信道(Physical Uplink Control Channel，PUCCH)或者物理上行共享信道(Physical Uplink Shared Channel，PUSCH)上发送CQI给网络设备。而网络设备则根据CQI的上报来调整终端的调制方式、编码码率和频谱效率。CQI是用来反映物理下行共享信道(Physical Downlink Shared Channel，PDSCH)的信道质量，例如表1所示，用0～15来表示PDSCH的信道质量。0表示信道质量最差，15表示信道质量最好。终端根据对下行信道估计的结果在16个CQI中选择一个，将CQI索引上报给网络设备。网络设备综合考虑各方面因素，可能考虑到多个终端之间的资源调配问题，为终端分配最终使用的资源参数，如表2所示的终端实际使用的调制方式、编码码率和频谱效率。
表1：
Figure PCTCN2020085998-appb-000001
表2：
Figure PCTCN2020085998-appb-000002
又如，在一些场景中，终端可向网络设备上报终端当前的无线信道条件和AI参数的组合，比如通过上报终端当前的无线信道条件和AI参数的组合对应的指示符来实现，也即，该指示符用于指示终端当前的无线信道条件和AI参数的组合，以便于网络设备根据该指示符进行资源分配，为终端分配最佳的无线通信资源参数和AI参数组合，实现无线通信资源参数和AI参数配置的最佳匹配，避免因为两者之间的不匹配，而造成AI任务质量下降和资源浪费。
可选的，终端向网络设备发送指示符的触发条件可以是当终端需要进行一项AI任务时终端向网络设备发送指示符，或者是终端周期性的向网络设备发送指示符，等等，本申请实施例不作限定。例如，终端可在启动AI任务时，向网络设备发送指示符。又如终端可按照预设时间间隔向网络设备发送指示符。
5G系统中AI的应用可以是如下所述三大应用场景：5G系统中的分割式AI操作、5G系统中的AI模型的下载和5G系统中的AI模型的训练。
例如，对于在5G系统中的分割式AI操作，终端与网络设备可以合作完成AI推理操作，其中，终端主要运行对时延和隐私保护敏感的相对低复杂度的计算，网络设备主要运行对时延和隐私不敏感的相对高复杂度的计算，即终端先完成一部分AI操作，将中间数据上报给网络设备，再由网络设备完成剩下的AI操作。
例如，对于在5G系统中的AI模型的下载，在终端自身存储中不具备所需模型得情况下，终端可以从网络设备下载所需模型使用。
例如，对于在5G系统中的AI模型的训练，在AI模型的训练过程中采用基于移动通信网络的分布式学习和联邦学习。在训练过程中，首先终端利用移动通信网络提供的全局模型进行训练，再将终端训练后的局部梯度上报给网络设备，再由网络设备对终端的局部模型进行合并，形成更优化的全局模型。
在本申请实施例中提供了一种通信方法及相关设备，有助于确保资源的可靠分配。以下分别详细说明。
请参见图2，图2是本申请实施例提供的一种通信方法的流程示意图。本实施例的方法应用于上述的通信系统，包括以下步骤：
201：终端向网络设备发送第一指示符。
具体的，终端可以向网络设备发送第一指示符，其中，第一指示符用于标识第一资源参数组合，第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，第一指示符与第一资源参数组合为第一对应关系。
可选的，第一通信资源参数可以表示终端当前的的无线信道条件，具体的可以是调制方法、编码码率和频谱效率中的至少一种，第一AI参数可以表示终端运行AI模型相关的参数，具体的可以是终端运行AI操作的算力水平、终端使用的AI模型、终端运行AI模型的分割点、终端运行AI操作的步骤、终端在完成一轮AI训练中所需的训练数据量和终端利用AI模型的训练结果上报周期中的至少一种。可选的，第一指示符可以是数字编号，比特值，或者其他指示信息，本申请不做限定。
在一些可行的实施方式中，终端可以获取该第一资源参数组合，比如根据对当前无线信道的信道估计得到第一通信资源参数，以及根据自身可用算力资源得到第一AI参数。进而根据该第一资源参数组合确定该第一指示符。可选的，终端可以根据第一资源参数组合和第一对应关系确定第一指示符， 其中，第一对应关系为第一指示符与第一资源参数组合之间的对应关系，第一资源参数组合可以是第一通信资源参数和第一AI参数。可选的，第一对应关系，即第一指示符与第一资源参数组合之间的对应关系可以由表格定义或者由其他方式定义，例如表3所示，示出了表格，该表格表示了第一指示符与第一资源参数组合之间的对应关系，表3中第一列表示第一指示符，后四列表示第一资源参数组合，从表中可以看出，每一个第一指示符都有对应的第一资源参数组合。第一对应关系可以预定义，或者可以由系统信息配置，或者可以由无线资源控制(Radio Resource Control，RRC)信令配置，或者也可以由媒体接入控制单元(Media Access Control Control Element，MAC CE)配置，本申请不做限定。
在一些可行的实施方式中，终端可以根据对当前无线信道的信道估计和自身可用算力资源，选择适合的第一通信资源参数和第一AI参数的组合，将对应的第一指示符发送给网络设备。例如，第一通信资源参数和第一AI参数的组合可以是第一通信资源参数中的任意一个和第一AI参数中的任意一个的组合，可以是第一通信资源参数中的任意一个和第一AI参数中的任意多个的组合，可以是第一通信资源参数中的任意多个和第一AI参数中的任意一个的组合，也可以是第一通信资源参数中的任意多个和第一AI参数中的任意多个的组合。
在一些可行的实施方式中，第一指示符与第一资源参数组合之间的对应关系可能存在多个，在这种情况下，终端可以根据下行控制信息(Downlink Control Information，DCI)指示或无线网络临时标识无线网络临时标识(Radio Network Tempory Identity，RNTI)或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定第一对应关系。
示例性的，网络设备可向终端发送DCI指示信息，该DCI指示信息可用于指示第一指示符与第一资源参数组合之间的对应关系，终端可接收网络设备发送的该DCI指示信息，进而根据该DCI指示信息确定第一对应关系。
示例性的，终端可以根据RNTI信息确定第一对应关系，该RNTI信息可用于指示第一指示符与第一资源参数组合之间的对应关系。
示例性的，网络设备可向终端发送DCI格式信息，该DCI格式信息可用于调度第一指示符与第一资源参数组合之间的对应关系，终端可接收网络设备发送的该DCI格式信息，进而根据该DCI格式信息确定第一对应关系。
示例性的，第一指示符与第一资源参数组合之间的对应关系可能存在多个，并且对应关系均用表格表示，例如表3与表5，比较这两个表，可以看出表3与表5中前四列均表示第一指示符、调制方法、编码码率和频谱效率，而表3中第五列表示终端运行AI操作的算力水平，表5中第五列表示终端运行AI模型的分割点，可以根据第五列的不同，即第一AI参数的不同，利用DCI指示或RNTI或DCI格式来确定当前所需的表格，也就是当前所需要的第一指示符与第一资源参数组合之间的对应关系。
在一些可行的实施方式中，终端可以通过PUCCH，向网络设备发送第一指示符；或者，可以通过PUSCH，向网络设备发送第一指示符；或者，也可以通过MAC CE，向网络设备发送第一指示符。
202：网络设备接收终端发送的第一指示符。
具体的，在终端向网络设备发送第一指示符之后，网络设备就可以接收终端发送的第一指示符。
在一些可行的实施方式中，网络设备在接收终端发送的第一指示符之后，就可以根据第一指示符与第一资源参数组合之间的对应关系确定终端的第一资源参数组合，即终端当前的无线信道条件和AI参数。
在一些可行的实施方式中，第一指示符与第一资源参数组合之间的对应关系可能存在多个，在这种情况下，网络设备可以根据下行控制信息(Downlink Control Information，DCI)指示或无线网络临时标识无线网络临时标识(Radio Network Tempory Identity，RNTI)或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定第一对应关系。
示例性的，网络设备可以根据DCI指示信息确定第一对应关系，该DCI指示信息可用于指示第一指示符与第一资源参数组合之间的对应关系。
示例性的，终端可向网络设备发送RNTI信息，该RNTI信息可用于指示第一指示符与第一资源参数组合之间的对应关系，网络设备可接收终端发送的该RNTI信息，进而根据该RNTI信息确定第一对应关系。
示例性的，网络设备可以根据DCI格式信息确定第一对应关系，该DCI格式信息可用于调度第一指示符与第一资源参数组合之间的对应关系。
需要说明的是，终端和网络设备从第一指示符与第一资源参数组合之间的多个对应关系中确定第一对应关系所利用的下行控制信息(Downlink Control Information，DCI)指示或无线网络临时标识无线网络临时标识(Radio Network Tempory Identity，RNTI)或DCI格式可以是相同的，也可以是不同的，本申请不做限定。
在一些可行的实施方式中，网络设备可以通过PUCCH接收终端发送的第一指示符；或者，可以通过PUSCH接收终端发送的第一指示符；或者，也可以通过MAC CE接收终端发送的第一指示符。
在本申请实施例中，终端向网络设备发送第一指示符，网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系，从而使网络设备指示同时关联终端当前的无线信道条件和AI参数组合的指示符，以确保资源分配的可靠性。
请参见图5，图5是本申请实施例提供的另一种通信方法的流程示意图。本实施例的方法应用于上述的通信系统，包括以下步骤：
501：终端向网络设备发送第一指示符。
502：网络设备接收终端发送的第一指示符。
其中，步骤501～502的具体实现方式可以参见上述通信方法中步骤201～202中的相关描述，此处不再赘述。
303：网络设备向终端发送第二指示符。
具体的，网络设备可以向终端发送第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
可选的，第二通信资源参数可以表示网络设备向终端调度的无线信道资源，具体的可以是调制方法、编码码率和频谱效率中的至少一种，第二AI参数可以表示网络设备向终端调度的终端运行AI模型相关的参数，具体的可以是终端运行AI操作的算力水平、终端使用的AI模型、终端运行AI模型的分割点、终端运行AI操作的步骤、终端在完成一轮AI训练中所需的训练数据量和终端利用AI模 型的训练结果上报周期中的至少一种。可选的，第二指示符可以是数字编号，比特值，或者其他指示信息，本申请不做限定。
在一些可行的实施方式中，网络设备可以根据终端发送的第一指示符，确定终端当前的第一资源参数组合，再根据第一资源参数组合，确定适合终端的第二资源参数组合。可选的，网络设备可以根据第二资源参数组合和第二对应关系确定第二指示符，其中，第二对应关系可以为第二指示符与第二资源参数组合之间的对应关系，第二资源参数组合可以是第二通信资源参数和第二AI参数。可选的，第二对应关系，即第二指示符与第二资源参数组合之间的对应关系可以由表格定义或者由其他方式定义，例如表4所示，示出了表格，该表格表示了第二指示符与第二资源参数组合之间的对应关系，表4中第一列表示第二指示符，后四列表示第二资源参数组合，从表中可以看出，每一个第二指示符都有对应的第二资源参数组合。第二对应关系可以预定义，或者可以由系统信息配置，或者可以由无线资源控制(Radio Resource Control，RRC)信令配置，或者也可以由媒体接入控制单元(Media Access Control Control Element，MAC CE)配置，本申请不做限定。
在一些可行的实施方式中，第二指示符与第二资源参数组合之间的对应关系可能存在多个，在这种情况下，网络设备可以根据DCI指示或RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定第二对应关系。
示例性的，网络设备可以根据DCI指示信息确定第二对应关系，该DCI指示信息可用于指示第二指示符与第二资源参数组合之间的对应关系。
示例性的，终端可向网络设备发送RNTI信息，该RNTI信息可用于指示第二指示符与第二资源参数组合之间的对应关系，网络设备可接收终端发送的该RNTI信息，进而根据该RNTI信息确定第二对应关系。
示例性的，网络设备可以根据DCI格式信息确定第二对应关系，该DCI格式信息可用于调度第二指示符与第二资源参数组合之间的对应关系。
示例性的，第二指示符与第二资源参数组合之间的对应关系可能存在多个，并且对应关系均用表格表示，例如表4与表6，比较这两个表，可以看出表4与表6中前四列均表示第二指示符、调制方法、编码码率和频谱效率，而表4中第五列表示终端运行AI操作的算力水平，表6中第五列表示终端运行AI模型的分割点，可以根据第五列的不同，即第二AI参数的不同，利用DCI指示或RNTI或DCI格式来确定当前所需的表格，也就是当前所需要的第二指示符与第二资源参数组合之间的对应关系。
在一些可行的实施方式中，网络设备可以通过物理下行控制信道(Physical Downlink Control Channel，PDCCH)向终端发送第二指示符；或者，可以通过PDSCH向终端发送第二指示符；或者，也可以通过MAC CE向终端发送第二指示符。
504：终端接收网络设备发送的第二指示符。
具体的，在网络设备发送第二指示符给终端之后，终端就可以接收网络设备发送的第二指示符，并根据第二指示符与第二资源参数组合之间的对应关系确定终端的第二资源参数组合，也就是说，终端可以使用网络设备调度的第二通信资源参数和第二AI参数。
在一些可行的实施方式中，第二指示符与第二资源参数组合之间的对应关系可能存在多个，在这种情况下，终端可以根据下行控制信息(Downlink Control Information，DCI)指示或无线网络临时标 识无线网络临时标识(Radio Network Tempory Identity，RNTI)或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定第二对应关系。
示例性的，网络设备可向终端发送DCI指示信息，该DCI指示信息可用于第二指示符与第二资源参数组合之间的对应关系，终端可接收网络设备发送的该DCI指示信息，进而根据该DCI第二指示符确定第二对应关系。
示例性的，终端可以根据RNTI信息确定第二对应关系，该RNTI信息可用于指示第二指示符与第二资源参数组合之间的对应关系。
示例性的，网络设备可向终端发送DCI格式信息，该DCI格式信息可用于调度第二指示符与第二资源参数组合之间的对应关系，终端可接收网络设备发送的该DCI格式信息，进而根据该DCI格式信息确定第二对应关系。
需要说明的是，终端和网络设备从第二指示符与第二资源参数组合之间的多个对应关系中确定第二对应关系所利用的下行控制信息(Downlink Control Information，DCI)指示或无线网络临时标识无线网络临时标识(Radio Network Tempory Identity，RNTI)或DCI格式可以是相同的，也可以是不同的，本申请不做限定。
示例性的，在第一AI参数为终端运行AI操作的算力水平的情况下，终端的可用算力可以影响到终端和网络设备之间就某项AI任务传输的质量。为了保证终端在进行AI任务时，终端能够使用无线通信资源配置和AI参数配置，终端可以将当前的无线信道条件和可用算力上报给网络设备。例如表3所示，表3表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合包含第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的终端运行AI操作的算力水平。算力水平可以基于预先约定的规则形成几个等级(Level)，分别对应一定的计算能力水平。如表3所示的终端算力能力有两个等级：Level 1和Level 2。表3中不同的第一指示符可以映射到Level 1或Level 2。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，从表3中选择适合的调制方式、编码码率、频谱效率和可用算力，再将其对应的第一指示符发送给网络设备。进一步的，网络设备可以根据终端发送的第一指示符，了解终端当前的第一通信资源参数和终端运行AI操作的算力水平，再根据第一通信资源参数和终端运行AI操作的算力水平综合确定最佳的第二通信资源参数和终端运行AI操作的算力水平，再把第二通信资源参数和终端运行AI操作的算力水平对应的第二指示符发送给终端，终端接收到第二指示符后，就可以使用网络设备调度的第二通信资源参数和可用算力完成AI任务。例如表4所示，表4表示的是第二指示符与第二资源参数组合之间的对应关系，其中，第二资源参数组合包含第二通信资源参数中的调制方法、编码码率和频谱效率和第二AI参数中的终端运行AI操作的算力水平，在网络设备确定最佳的第二通信资源参数和终端运行AI操作的算力水平之后，就可以根据表4确定对应的第二指示符。
表3：
第一指示符 调制方法 编码码率 频谱效率 算力水平
0 QPSK Rate 0 Efficiency 0 Level 1
1 QPSK Rate 1 Efficiency 1 Level 1
2 QPSK Rate 2 Efficiency 2 Level 1
3 QPSK Rate 3 Efficiency 3 Level 1
4 QPSK Rate 4 Efficiency 4 Level 1
5 QPSK Rate 5 Efficiency 5 Level 1
6 QPSK Rate 6 Efficiency 6 Level 1
7 16QAM Rate 7 Efficiency 7 Level 1
8 16QAM Rate 8 Efficiency 8 Level 1
9 16QAM Rate 9 Efficiency 9 Level 1
10 64QAM Rate 10 Efficiency 10 Level 1
11 64QAM Rate 11 Efficiency 11 Level 1
12 64QAM Rate 12 Efficiency 12 Level 1
13 64QAM Rate 13 Efficiency 13 Level 1
14 64QAM Rate 14 Efficiency 14 Level 1
15 64QAM Rate 15 Efficiency 15 Level 1
16 QPSK Rate 0 Efficiency 0 Level 2
17 QPSK Rate 1 Efficiency 1 Level 2
18 QPSK Rate 2 Efficiency 2 Level 2
19 QPSK Rate 3 Efficiency 3 Level 2
20 QPSK Rate 4 Efficiency 4 Level 2
21 QPSK Rate 5 Efficiency 5 Level 2
22 QPSK Rate 6 Efficiency 6 Level 2
23 16QAM Rate 7 Efficiency 7 Level 2
24 16QAM Rate 8 Efficiency 8 Level 2
25 16QAM Rate 9 Efficiency 9 Level 2
26 64QAM Rate 10 Efficiency 10 Level 2
27 64QAM Rate 11 Efficiency 11 Level 2
28 64QAM Rate 12 Efficiency 12 Level 2
29 64QAM Rate 13 Efficiency 13 Level 2
30 64QAM Rate 14 Efficiency 14 Level 2
31 64QAM Rate 15 Efficiency 15 Level 2
表4：
第二指示符 调制方法 编码码率 频谱效率 算力水平
0 2 Rate 0 Efficiency 0 Level 1
1 2 Rate 1 Efficiency 1 Level 1
2 2 Rate 2 Efficiency 2 Level 1
3 2 Rate 3 Efficiency 3 Level 1
4 2 Rate 4 Efficiency 4 Level 1
5 2 Rate 5 Efficiency 5 Level 1
6 2 Rate 6 Efficiency 6 Level 1
7 4 Rate 7 Efficiency 7 Level 1
8 4 Rate 8 Efficiency 8 Level 1
9 4 Rate 9 Efficiency 9 Level 1
10 6 Rate 10 Efficiency 10 Level 1
11 6 Rate 11 Efficiency 11 Level 1
12 6 Rate 12 Efficiency 12 Level 1
13 6 Rate 13 Efficiency 13 Level 1
14 6 Rate 14 Efficiency 14 Level 1
15 6 Rate 15 Efficiency 15 Level 1
16 2 Rate 0 Efficiency 0 Level 2
17 2 Rate 1 Efficiency 1 Level 2
18 2 Rate 2 Efficiency 2 Level 2
19 2 Rate 3 Efficiency 3 Level 2
20 2 Rate 4 Efficiency 4 Level 2
21 2 Rate 5 Efficiency 5 Level 2
22 2 Rate 6 Efficiency 6 Level 2
23 4 Rate 7 Efficiency 7 Level 2
24 4 Rate 8 Efficiency 8 Level 2
25 4 Rate 9 Efficiency 9 Level 2
26 6 Rate 10 Efficiency 10 Level 2
27 6 Rate 11 Efficiency 11 Level 2
28 6 Rate 12 Efficiency 12 Level 2
29 6 Rate 13 Efficiency 13 Level 2
30 6 Rate 14 Efficiency 14 Level 2
31 6 Rate 15 Efficiency 15 Level 2
需要说明的是，表3与表4中没有一一对应关系，如表3第二行与表4第二行，表3与表4分别是第一对应关系和第二对应关系的表示方法，下面所示的表格也是如此。
示例性的，在第一AI参数为终端运行AI模型的分割点的情况下，终端运行对时延和隐私保护敏感的相对低复杂度的计算，网络设备运行对时延和隐私不敏感的相对高复杂度的计算，如图3a与3b所示。如图4所示是一个典型的卷积神经网络(Convolutional Neural Networks，CNN)结构，从图4中可以看出，在CNN中具有一些可选的分割点(Split point)，终端只执行一部分层的计算(计算量分别为计算量水平1、2)，直至分割点，然后将中间数据量传输给网络设备，由网络设备执行剩下层的计算。同时，可以看出，终端在分割点2下需要执行的计算量大于分割点1，并且终端在分割点1下需要上传的中间数据量明显大于分割点2。因此，分割点的选择可以影响到网络设备需要分配给终端的无线通信资源。为了保证终端在进行AI任务时，终端能够使用无线通信资源配置和AI参数配置， 终端可以将当前的无线信道条件和分割点上报给网络设备。例如表5所示，表5表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合包含第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的终端运行AI模型的分割点，如表5所示的AI模型分割点有两种选择：Split point 1或Split point 2。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，从表5中选择适合的调制方式、编码码率、频谱效率和分割点，再将其对应的第一指示符发送给网络设备。进一步的，网络设备可以根据终端发送的第一指示符，了解终端当前的第一通信资源参数和终端运行AI模型的分割点，再根据第一通信资源参数和终端运行AI模型的分割点综合确定最佳的第二通信资源参数和终端运行AI模型的分割点，再把第二通信资源参数和终端运行AI模型的分割点对应的第二指示符发送给终端，终端接收到第二指示符后，就可以使用网络设备调度的第二通信资源参数和终端运行AI模型的分割点完成AI任务，例如表6所示，表6表示的是第二指示符与第二资源参数组合之间的对应关系，其中，第二资源参数组合包含第二通信资源参数中的调制方法、编码码率和频谱效率和第二AI参数中的终端运行AI模型的分割点，在网络设备确定最佳的第二通信资源参数和终端运行AI模型的分割点之后，就可以根据表6确定对应的第二指示符。
表5：
第一指示符 调制方法 编码码率 频谱效率 分割点
0 QPSK Rate 0 Efficiency 0 Split point 1
1 QPSK Rate 1 Efficiency 1 Split point 1
2 QPSK Rate 2 Efficiency 2 Split point 1
3 QPSK Rate 3 Efficiency 3 Split point 1
4 QPSK Rate 4 Efficiency 4 Split point 1
5 QPSK Rate 5 Efficiency 5 Split point 1
6 QPSK Rate 6 Efficiency 6 Split point 1
7 16QAM Rate 7 Efficiency 7 Split point 1
8 16QAM Rate 8 Efficiency 8 Split point 1
9 16QAM Rate 9 Efficiency 9 Split point 1
10 64QAM Rate 10 Efficiency 10 Split point 1
11 64QAM Rate 11 Efficiency 11 Split point 1
12 64QAM Rate 12 Efficiency 12 Split point 1
13 64QAM Rate 13 Efficiency 13 Split point 1
14 64QAM Rate 14 Efficiency 14 Split point 1
15 64QAM Rate 15 Efficiency 15 Split point 1
16 QPSK Rate 0 Efficiency 0 Split point 2
17 QPSK Rate 1 Efficiency 1 Split point 2
18 QPSK Rate 2 Efficiency 2 Split point 2
19 QPSK Rate 3 Efficiency 3 Split point 2
20 QPSK Rate 4 Efficiency 4 Split point 2
21 QPSK Rate 5 Efficiency 5 Split point 2
22 QPSK Rate 6 Efficiency 6 Split point 2
23 16QAM Rate 7 Efficiency 7 Split point 2
24 16QAM Rate 8 Efficiency 8 Split point 2
25 16QAM Rate 9 Efficiency 9 Split point 2
26 64QAM Rate 10 Efficiency 10 Split point 2
27 64QAM Rate 11 Efficiency 11 Split point 2
28 64QAM Rate 12 Efficiency 12 Split point 2
29 64QAM Rate 13 Efficiency 13 Split point 2
30 64QAM Rate 14 Efficiency 14 Split point 2
31 64QAM Rate 15 Efficiency 15 Split point 2
表6：
第二指示符 调制方法 编码码率 频谱效率 分割点
0 2 Rate 0 Efficiency 0 Split point 1
1 2 Rate 1 Efficiency 1 Split point 1
2 2 Rate 2 Efficiency 2 Split point 1
3 2 Rate 3 Efficiency 3 Split point 1
4 2 Rate 4 Efficiency 4 Split point 1
5 2 Rate 5 Efficiency 5 Split point 1
6 2 Rate 6 Efficiency 6 Split point 1
7 4 Rate 7 Efficiency 7 Split point 1
8 4 Rate 8 Efficiency 8 Split point 1
9 4 Rate 9 Efficiency 9 Split point 1
10 6 Rate 10 Efficiency 10 Split point 1
11 6 Rate 11 Efficiency 11 Split point 1
12 6 Rate 12 Efficiency 12 Split point 1
13 6 Rate 13 Efficiency 13 Split point 1
14 6 Rate 14 Efficiency 14 Split point 1
15 6 Rate 15 Efficiency 15 Split point 1
16 2 Rate 0 Efficiency 0 Split point 2
17 2 Rate 1 Efficiency 1 Split point 2
18 2 Rate 2 Efficiency 2 Split point 2
19 2 Rate 3 Efficiency 3 Split point 2
20 2 Rate 4 Efficiency 4 Split point 2
21 2 Rate 5 Efficiency 5 Split point 2
22 2 Rate 6 Efficiency 6 Split point 2
23 4 Rate 7 Efficiency 7 Split point 2
24 4 Rate 8 Efficiency 8 Split point 2
25 4 Rate 9 Efficiency 9 Split point 2
26 6 Rate 10 Efficiency 10 Split point 2
27 6 Rate 11 Efficiency 11 Split point 2
28 6 Rate 12 Efficiency 12 Split point 2
29 6 Rate 13 Efficiency 13 Split point 2
30 6 Rate 14 Efficiency 14 Split point 2
31 6 Rate 15 Efficiency 15 Split point 2
示例性的，在第一AI参数为终端运行AI操作的步骤的情况下，AI操作分割方法可以是将一个AI流程的不同步骤或不同部分分割给终端和网络设备进行操作，不同的分割方法输出的中间数据量也有很大不同。例如表7所示，表7表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合是第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的终端运行AI操作的步骤的组合，如表7所示终端的AI操作步骤有两种选择：Step 1-2和Step 1-4。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，从表7中选择适合的调制方式、编码码率、频谱效率和步骤，将其对应的第一指示符发送给给网络设备。进一步的，网络设备根据终端上报的第一指示符确定适合终端的第二通信资源参数和终端运行AI操作的步骤，并将对应的第二指示符发送给终端，如表8所示。终端接收到第二指示符后，就可以使用网络设备调度的第二通信资源参数和终端运行AI操作的步骤完成AI任务。
表7：
第一指示符 调制方法 编码码率 频谱效率 步骤
0 QPSK Rate 0 Efficiency 0 Step 1-2
1 QPSK Rate 1 Efficiency 1 Step 1-2
2 QPSK Rate 2 Efficiency 2 Step 1-2
3 QPSK Rate 3 Efficiency 3 Step 1-2
4 QPSK Rate 4 Efficiency 4 Step 1-2
5 QPSK Rate 5 Efficiency 5 Step 1-2
6 QPSK Rate 6 Efficiency 6 Step 1-2
7 16QAM Rate 7 Efficiency 7 Step 1-2
8 16QAM Rate 8 Efficiency 8 Step 1-2
9 16QAM Rate 9 Efficiency 9 Step 1-2
10 64QAM Rate 10 Efficiency 10 Step 1-2
11 64QAM Rate 11 Efficiency 11 Step 1-2
12 64QAM Rate 12 Efficiency 12 Step 1-2
13 64QAM Rate 13 Efficiency 13 Step 1-2
14 64QAM Rate 14 Efficiency 14 Step 1-2
15 64QAM Rate 15 Efficiency 15 Step 1-2
16 QPSK Rate 0 Efficiency 0 Step 1-4
17 QPSK Rate 1 Efficiency 1 Step 1-4
18 QPSK Rate 2 Efficiency 2 Step 1-4
19 QPSK Rate 3 Efficiency 3 Step 1-4
20 QPSK Rate 4 Efficiency 4 Step 1-4
21 QPSK Rate 5 Efficiency 5 Step 1-4
22 QPSK Rate 6 Efficiency 6 Step 1-4
23 16QAM Rate 7 Efficiency 7 Step 1-4
24 16QAM Rate 8 Efficiency 8 Step 1-4
25 16QAM Rate 9 Efficiency 9 Step 1-4
26 64QAM Rate 10 Efficiency 10 Step 1-4
27 64QAM Rate 11 Efficiency 11 Step 1-4
28 64QAM Rate 12 Efficiency 12 Step 1-4
29 64QAM Rate 13 Efficiency 13 Step 1-4
30 64QAM Rate 14 Efficiency 14 Step 1-4
31 64QAM Rate 15 Efficiency 15 Step 1-4
表8：
第二指示符 调制方法 编码码率 频谱效率 步骤
0 2 Rate 0 Efficiency 0 Step 1-2
1 2 Rate 1 Efficiency 1 Step 1-2
2 2 Rate 2 Efficiency 2 Step 1-2
3 2 Rate 3 Efficiency 3 Step 1-2
4 2 Rate 4 Efficiency 4 Step 1-2
5 2 Rate 5 Efficiency 5 Step 1-2
6 2 Rate 6 Efficiency 6 Step 1-2
7 4 Rate 7 Efficiency 7 Step 1-2
8 4 Rate 8 Efficiency 8 Step 1-2
9 4 Rate 9 Efficiency 9 Step 1-2
10 6 Rate 10 Efficiency 10 Step 1-2
11 6 Rate 11 Efficiency 11 Step 1-2
12 6 Rate 12 Efficiency 12 Step 1-2
13 6 Rate 13 Efficiency 13 Step 1-2
14 6 Rate 14 Efficiency 14 Step 1-2
15 6 Rate 15 Efficiency 15 Step 1-2
16 2 Rate 0 Efficiency 0 Step 1-4
17 2 Rate 1 Efficiency 1 Step 1-4
18 2 Rate 2 Efficiency 2 Step 1-4
19 2 Rate 3 Efficiency 3 Step 1-4
20 2 Rate 4 Efficiency 4 Step 1-4
21 2 Rate 5 Efficiency 5 Step 1-4
22 2 Rate 6 Efficiency 6 Step 1-4
23 4 Rate 7 Efficiency 7 Step 1-4
24 4 Rate 8 Efficiency 8 Step 1-4
25 4 Rate 9 Efficiency 9 Step 1-4
26 6 Rate 10 Efficiency 10 Step 1-4
27 6 Rate 11 Efficiency 11 Step 1-4
28 6 Rate 12 Efficiency 12 Step 1-4
29 6 Rate 13 Efficiency 13 Step 1-4
30 6 Rate 14 Efficiency 14 Step 1-4
31 6 Rate 15 Efficiency 15 Step 1-4
示例性的，在第一AI参数为终端使用的AI模型的情况下，例如表9所示，表9表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合是第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的终端使用的AI模型的组合，如表7所示终端可用的AI模型有两个：Model#1和Model#2。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，从表9中选择适合的调制方式、编码码率、频谱效率和终端适合使用的AI模型，将其对应的第一指示符发送给给网络设备。进一步的，网络设备根据终端上报的第一指示符确定适合终端的第二通信资源参数和终端使用的AI模型，并将对应的第二指示符发送给终端，如表10所示。终端接收到第二指示符后，就可以使用网络设备调度的第二通信资源参数和AI模型完成AI任务。
表9：
第一指示符 调制方法 编码码率 频谱效率 模型
0 QPSK Rate 0 Efficiency 0 Model #1
1 QPSK Rate 1 Efficiency 1 Model #1
2 QPSK Rate 2 Efficiency 2 Model #1
3 QPSK Rate 3 Efficiency 3 Model #1
4 QPSK Rate 4 Efficiency 4 Model #1
5 QPSK Rate 5 Efficiency 5 Model #1
6 QPSK Rate 6 Efficiency 6 Model #1
7 16QAM Rate 7 Efficiency 7 Model #1
8 16QAM Rate 8 Efficiency 8 Model #1
9 16QAM Rate 9 Efficiency 9 Model #1
10 64QAM Rate 10 Efficiency 10 Model #1
11 64QAM Rate 11 Efficiency 11 Model #1
12 64QAM Rate 12 Efficiency 12 Model #1
13 64QAM Rate 13 Efficiency 13 Model #1
14 64QAM Rate 14 Efficiency 14 Model #1
15 64QAM Rate 15 Efficiency 15 Model #1
16 QPSK Rate 0 Efficiency 0 Model #2
17 QPSK Rate 1 Efficiency 1 Model #2
18 QPSK Rate 2 Efficiency 2 Model #2
19 QPSK Rate 3 Efficiency 3 Model #2
20 QPSK Rate 4 Efficiency 4 Model #2
21 QPSK Rate 5 Efficiency 5 Model #2
22 QPSK Rate 6 Efficiency 6 Model #2
23 16QAM Rate 7 Efficiency 7 Model #2
24 16QAM Rate 8 Efficiency 8 Model #2
25 16QAM Rate 9 Efficiency 9 Model #2
26 64QAM Rate 10 Efficiency 10 Model #2
27 64QAM Rate 11 Efficiency 11 Model #2
28 64QAM Rate 12 Efficiency 12 Model #2
29 64QAM Rate 13 Efficiency 13 Model #2
30 64QAM Rate 14 Efficiency 14 Model #2
31 64QAM Rate 15 Efficiency 15 Model #2
表10：
第二指示符 调制方法 编码码率 频谱效率 模型
0 2 Rate 0 Efficiency 0 Model #1
1 2 Rate 1 Efficiency 1 Model #1
2 2 Rate 2 Efficiency 2 Model #1
3 2 Rate 3 Efficiency 3 Model #1
4 2 Rate 4 Efficiency 4 Model #1
5 2 Rate 5 Efficiency 5 Model #1
6 2 Rate 6 Efficiency 6 Model #1
7 4 Rate 7 Efficiency 7 Model #1
8 4 Rate 8 Efficiency 8 Model #1
9 4 Rate 9 Efficiency 9 Model #1
10 6 Rate 10 Efficiency 10 Model #1
11 6 Rate 11 Efficiency 11 Model #1
12 6 Rate 12 Efficiency 12 Model #1
13 6 Rate 13 Efficiency 13 Model #1
14 6 Rate 14 Efficiency 14 Model #1
15 6 Rate 15 Efficiency 15 Model #1
16 2 Rate 0 Efficiency 0 Model #2
17 2 Rate 1 Efficiency 1 Model #2
18 2 Rate 2 Efficiency 2 Model #2
19 2 Rate 3 Efficiency 3 Model #2
20 2 Rate 4 Efficiency 4 Model #2
21 2 Rate 5 Efficiency 5 Model #2
22 2 Rate 6 Efficiency 6 Model #2
23 4 Rate 7 Efficiency 7 Model #2
24 4 Rate 8 Efficiency 8 Model #2
25 4 Rate 9 Efficiency 9 Model #2
26 6 Rate 10 Efficiency 10 Model #2
27 6 Rate 11 Efficiency 11 Model #2
28 6 Rate 12 Efficiency 12 Model #2
29 6 Rate 13 Efficiency 13 Model #2
30 6 Rate 14 Efficiency 14 Model #2
31 6 Rate 15 Efficiency 15 Model #2
示例性的，在第一AI参数为终端在完成一轮AI训练中所需的训练数据量情况下，例如表11所示，表11表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合是第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的终端在完成一轮AI训练中所需的训练数据量的组合，如表11所示终端在一轮训练中可完成的AI/ML训练数据量有两种：Data size 1和Data size 2。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，从表11中选择适合的调制方式、编码码率、频谱效率和适合的AI训练数据量，将其对应的第一指示符发送给给网络设备。进一步的，网络设备根据终端上报的第一指示符确定适合终端的第二通信资源参数和训练数据量，并将对应的第二指示符发送给终端，如表12所示。终端接收到第二指示符后，就可以使用网络设备调度的第一通信资源参数和训练数据量完成AI任务。
表11：
第一指示符 调制方法 编码码率 频谱效率 训练数据
0 QPSK Rate 0 Efficiency 0 Data size 1
1 QPSK Rate 1 Efficiency 1 Data size 1
2 QPSK Rate 2 Efficiency 2 Data size 1
3 QPSK Rate 3 Efficiency 3 Data size 1
4 QPSK Rate 4 Efficiency 4 Data size 1
5 QPSK Rate 5 Efficiency 5 Data size 1
6 QPSK Rate 6 Efficiency 6 Data size 1
7 16QAM Rate 7 Efficiency 7 Data size 1
8 16QAM Rate 8 Efficiency 8 Data size 1
9 16QAM Rate 9 Efficiency 9 Data size 1
10 64QAM Rate 10 Efficiency 10 Data size 1
11 64QAM Rate 11 Efficiency 11 Data size 1
12 64QAM Rate 12 Efficiency 12 Data size 1
13 64QAM Rate 13 Efficiency 13 Data size 1
14 64QAM Rate 14 Efficiency 14 Data size 1
15 64QAM Rate 15 Efficiency 15 Data size 1
16 QPSK Rate 0 Efficiency 0 Data size 2
17 QPSK Rate 1 Efficiency 1 Data size 2
18 QPSK Rate 2 Efficiency 2 Data size 2
19 QPSK Rate 3 Efficiency 3 Data size 2
20 QPSK Rate 4 Efficiency 4 Data size 2
21 QPSK Rate 5 Efficiency 5 Data size 2
22 QPSK Rate 6 Efficiency 6 Data size 2
23 16QAM Rate 7 Efficiency 7 Data size 2
24 16QAM Rate 8 Efficiency 8 Data size 2
25 16QAM Rate 9 Efficiency 9 Data size 2
26 64QAM Rate 10 Efficiency 10 Data size 2
27 64QAM Rate 11 Efficiency 11 Data size 2
28 64QAM Rate 12 Efficiency 12 Data size 2
29 64QAM Rate 13 Efficiency 13 Data size 2
30 64QAM Rate 14 Efficiency 14 Data size 2
31 64QAM Rate 15 Efficiency 15 Data size 2
表12：
第二指示符 调制方法 编码码率 频谱效率 训练数据
0 2 Rate 0 Efficiency 0 Data size 1
1 2 Rate 1 Efficiency 1 Data size 1
2 2 Rate 2 Efficiency 2 Data size 1
3 2 Rate 3 Efficiency 3 Data size 1
4 2 Rate 4 Efficiency 4 Data size 1
5 2 Rate 5 Efficiency 5 Data size 1
6 2 Rate 6 Efficiency 6 Data size 1
7 4 Rate 7 Efficiency 7 Data size 1
8 4 Rate 8 Efficiency 8 Data size 1
9 4 Rate 9 Efficiency 9 Data size 1
10 6 Rate 10 Efficiency 10 Data size 1
11 6 Rate 11 Efficiency 11 Data size 1
12 6 Rate 12 Efficiency 12 Data size 1
13 6 Rate 13 Efficiency 13 Data size 1
14 6 Rate 14 Efficiency 14 Data size 1
15 6 Rate 15 Efficiency 15 Data size 1
16 2 Rate 0 Efficiency 0 Data size 2
17 2 Rate 1 Efficiency 1 Data size 2
18 2 Rate 2 Efficiency 2 Data size 2
19 2 Rate 3 Efficiency 3 Data size 2
20 2 Rate 4 Efficiency 4 Data size 2
21 2 Rate 5 Efficiency 5 Data size 2
22 2 Rate 6 Efficiency 6 Data size 2
23 4 Rate 7 Efficiency 7 Data size 2
24 4 Rate 8 Efficiency 8 Data size 2
25 4 Rate 9 Efficiency 9 Data size 2
26 6 Rate 10 Efficiency 10 Data size 2
27 6 Rate 11 Efficiency 11 Data size 2
28 6 Rate 12 Efficiency 12 Data size 2
29 6 Rate 13 Efficiency 13 Data size 2
30 6 Rate 14 Efficiency 14 Data size 2
31 6 Rate 15 Efficiency 15 Data size 2
示例性的，对于第一AI参数为终端利用AI模型的训练结果上报周期而言，例如表13所示，表13表示的是第一指示符与第一资源参数组合之间的对应关系，其中，第一资源参数组合是第一通信资源参数中的调制方法、编码码率和频谱效率和第一AI参数中的训练结果上报周期的组合，如表13所示终端在一轮训练中可完成的AI模型训练结果上报周期有两种：Period 1和Period 2。具体的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，选择适合的调制方式、编码码率、频谱效率和训练结果上报周期，将其对应的第一指示符发送给网络设备。进一步的，网络设备根据终端上报的第一指示符确定适合终端的第二通信资源参数和训练结果上报周期，并将对应的第二指示符发送给终端，如表14所示。终端接收到第二指示符后，就可以使用网络设备调度的第一通信资源参数和训练 结果上报周期完成AI任务。
表13：
第一指示符 调制方法 编码码率 频谱效率 训练结果上报周期
0 QPSK Rate 0 Efficiency 0 Period 1
1 QPSK Rate 1 Efficiency 1 Period 1
2 QPSK Rate 2 Efficiency 2 Period 1
3 QPSK Rate 3 Efficiency 3 Period 1
4 QPSK Rate 4 Efficiency 4 Period 1
5 QPSK Rate 5 Efficiency 5 Period 1
6 QPSK Rate 6 Efficiency 6 Period 1
7 16QAM Rate 7 Efficiency 7 Period 1
8 16QAM Rate 8 Efficiency 8 Period 1
9 16QAM Rate 9 Efficiency 9 Period 1
10 64QAM Rate 10 Efficiency 10 Period 1
11 64QAM Rate 11 Efficiency 11 Period 1
12 64QAM Rate 12 Efficiency 12 Period 1
13 64QAM Rate 13 Efficiency 13 Period 1
14 64QAM Rate 14 Efficiency 14 Period 1
15 64QAM Rate 15 Efficiency 15 Period 1
16 QPSK Rate 0 Efficiency 0 Period 2
17 QPSK Rate 1 Efficiency 1 Period 2
18 QPSK Rate 2 Efficiency 2 Period 2
19 QPSK Rate 3 Efficiency 3 Period 2
20 QPSK Rate 4 Efficiency 4 Period 2
21 QPSK Rate 5 Efficiency 5 Period 2
22 QPSK Rate 6 Efficiency 6 Period 2
23 16QAM Rate 7 Efficiency 7 Period 2
24 16QAM Rate 8 Efficiency 8 Period 2
25 16QAM Rate 9 Efficiency 9 Period 2
26 64QAM Rate 10 Efficiency 10 Period 2
27 64QAM Rate 11 Efficiency 11 Period 2
28 64QAM Rate 12 Efficiency 12 Period 2
29 64QAM Rate 13 Efficiency 13 Period 2
30 64QAM Rate 14 Efficiency 14 Period 2
31 64QAM Rate 15 Efficiency 15 Period 2
表14：
第二指示符 调制方法 编码码率 频谱效率 训练结果上报周期
0 2 Rate 0 Efficiency 0 Period 1
1 2 Rate 1 Efficiency 1 Period 1
2 2 Rate 2 Efficiency 2 Period 1
3 2 Rate 3 Efficiency 3 Period 1
4 2 Rate 4 Efficiency 4 Period 1
5 2 Rate 5 Efficiency 5 Period 1
6 2 Rate 6 Efficiency 6 Period 1
7 4 Rate 7 Efficiency 7 Period 1
8 4 Rate 8 Efficiency 8 Period 1
9 4 Rate 9 Efficiency 9 Period 1
10 6 Rate 10 Efficiency 10 Period 1
11 6 Rate 11 Efficiency 11 Period 1
12 6 Rate 12 Efficiency 12 Period 1
13 6 Rate 13 Efficiency 13 Period 1
14 6 Rate 14 Efficiency 14 Period 1
15 6 Rate 15 Efficiency 15 Period 1
16 2 Rate 0 Efficiency 0 Period 2
17 2 Rate 1 Efficiency 1 Period 2
18 2 Rate 2 Efficiency 2 Period 2
19 2 Rate 3 Efficiency 3 Period 2
20 2 Rate 4 Efficiency 4 Period 2
21 2 Rate 5 Efficiency 5 Period 2
22 2 Rate 6 Efficiency 6 Period 2
23 4 Rate 7 Efficiency 7 Period 2
24 4 Rate 8 Efficiency 8 Period 2
25 4 Rate 9 Efficiency 9 Period 2
26 6 Rate 10 Efficiency 10 Period 2
27 6 Rate 11 Efficiency 11 Period 2
28 6 Rate 12 Efficiency 12 Period 2
29 6 Rate 13 Efficiency 13 Period 2
30 6 Rate 14 Efficiency 14 Period 2
31 6 Rate 15 Efficiency 15 Period 2
示例性的，终端可以根据对当前无线信道的信道估计和自身可用算力资源，选择适合的调制方式、编码码率和终端运行AI操作的算力水平，将相应的第一指示符发送给网络设备，进一步的，网络设 备根据终端上报的第一指示符确定适合的可以是终端的调制方式、编码码率、终端运行AI操作的算力水平，也可以是终端的调制方式、编码码率、终端运行AI操作的算力水平和终端适合使用的AI模型，并将相应的第二指示符发送给终端，终端接收到第二指示符后，就可以使用网络设备调度的第一通信资源参数和训练结果上报周期完成AI任务。
在一些可行的实施方式中，终端可以通过PDCCH接收网络设备发送的第二指示符；或者，可以接收通过PDSCH网络设备发送的第二指示符；或者，也可以通过MAC CE接收网络设备发送的第二指示符。
在本申请实施例中，终端向网络设备发送第一指示符，网络设备接收终端发送的第一指示符，其中，第一指示符用于标识第一资源参数组合，第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，第一指示符与第一资源参数组合为第一对应关系，进一步的，网络设备向终端发送第二指示符，终端接收网络设备发送的第二指示符，其中，二指示符用于标识第二资源参数组合，第二资源参数组合包括第二通信资源参数和第二AI参数，第二指示符与第二资源参数组合为第二对应关系，从而使网络设备可以根据终端上报的无线信道条件和AI参数，为终端调度最佳的无线通信资源参数和AI参数组合，实现无线通信资源参数和AI参数配置的最佳匹配，避免两者之间不匹配造成的AI业务质量下降和资源浪费。
参见图6，图6是本申请实施例提供的一种通信装置的结构示意图，应用于终端，该装置包括：
发送单元601，用于终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
在一些可行的实施方式中，所述装置还包括接收单元602，具体用于：
所述终端接收所述网络设备发送的第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
在一些可行的实施方式中，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
在一些可行的实施方式中，所述第一通信资源参数和/或所述第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
在一些可行的实施方式中，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
在一些可行的实施方式中，所述装置还包括确定单元603，具体用于：
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
在一些可行的实施方式中，发送单元601，具体用于：
所述终端通过物理上行控制信道PUCCH，向网络设备发送第一指示符；
或，所述终端通过物理上行共享信道PUSCH，向网络设备发送第一指示符；
或，所述终端通过媒体接入控制单元MAC CE，向网络设备发送第一指示符。
在一些可行的实施方式中，接收单元602，具体用于：
所述终端通过物理下行控制信道PDCCH接收所述网络设备发送的第二指示符；
或，终端通过物理下行共享信道PDSCH接收所述网络设备发送的第二指示符；
或，终端通过媒体接入控制单元MAC CE接收所述网络设备发送的第二指示符。
可以理解，本申请实施例中对单元的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。本申请实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。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。
参见图7，图7是本申请实施例提供的一种通信装置的结构示意图。应用于网络设备，该装置包括：
接收单元701，用于网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
在一些可行的实施方式中，所述装置还包括发送单元702，具体用于：
所述网络设备向所述终端发送第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
在一些可行的实施方式中，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
在一些可行的实施方式中，所述第一通信资源参数和/或第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
在一些可行的实施方式中，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
在一些可行的实施方式中，所述装置还包括确定单元703，具体用于：
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
在一些可行的实施方式中，所述接收单元701，具体用于：
所述网络设备通过物理上行控制信道PUCCH接收终端发送的第一指示符；
或，所述网络设备通过物理上行共享信道PUSCH接收终端发送的第一指示符；
或，所述网络设备通过媒体接入控制单元MAC CE接收终端发送的第一指示符。
在一些可行的实施方式中，所述发送单元702，具体用于：
所述网络设备通过物理下行控制信道PDCCH，向所述终端发送第二指示符；
或，所述网络设备通过物理下行共享信道PDSCH，向所述终端发送第二指示符；
或，所述网络设备通过媒体接入控制单元MAC CE，向所述终端发送第二指示符。
可以理解，本申请实施例中对单元的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。本申请实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。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。
参见图8，图8为本申请实施例提供的一种终端的结构示意图，其中，图8所示的终端可以包括：处理器801、接收器802、发射器803以及存储器804，处理器801、接收器802、发射器803以及存储器804分别连接总线，其中，存储器804中用于存储计算机程序，所述计算机程序包括程序指令，处理器801用于执行存储器804存储的程序指令。
在本申请实施例中，处理器801通过运行存储器804中的可执行程序代码，执行如下操作：
终端向网络设备通过发射器803发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
在一些可行的实施方式中，所述处理器801，还用于：
所述终端通过接收器802接收所述网络设备发送的第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
在一些可行的实施方式中，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
在一些可行的实施方式中，所述第一通信资源参数和/或所述第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
在一些可行的实施方式中，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
在一些可行的实施方式中，所述处理器801，还用于：
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
在一些可行的实施方式中，所述处理器801，具体用于：
所述终端通过物理上行控制信道PUCCH，向网络设备通过发射器803发送第一指示符；
或，所述终端通过物理上行共享信道PUSCH，向网络设备通过发射器803发送第一指示符；
或，所述终端通过媒体接入控制单元MAC CE，向网络设备通过发射器803发送第一指示符。
在一些可行的实施方式中，所述处理器801，具体用于：
所述终端通过物理下行控制信道PDCCH，通过接收器802接收所述网络设备发送的第二指示符；
或，终端通过物理下行共享信道PDSCH，通过接收器802接收所述网络设备发送的第二指示符；
或，终端通过媒体接入控制单元MAC CE，通过接收器802接收所述网络设备发送的第二指示符。
参见图9，图9为本申请实施例提供的一种网络设备的结构示意图，其中，图9所示的网络设备可以包括：处理器901、接收器902、发射器903以及存储器904，处理器901、接收器902、发射器903以及存储器904分别连接总线，其中，存储器904中用于存储计算机程序，所述计算机程序包括程序指令，处理器901用于执行存储器904存储的程序指令。
在本申请实施例中，处理器901通过运行存储器904中的可执行程序代码，执行如下操作：
网络设备通过接收器902接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
在一些可行的实施方式中，所述处理器901，还用于：
所述网络设备向所述终端通过发射器903发送第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
在一些可行的实施方式中，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
在一些可行的实施方式中，所述第一通信资源参数和/或第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
在一些可行的实施方式中，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
在一些可行的实施方式中，所述处理器901，还用于：
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
在一些可行的实施方式中，所述处理器901，具体用于：
所述网络设备通过物理上行控制信道PUCCH，通过接收器902接收终端发送的第一指示符；
或，所述网络设备通过物理上行共享信道PUSCH，通过接收器902接收终端发送的第一指示符；
或，所述网络设备通过媒体接入控制单元MAC CE，通过接收器902接收终端发送的第一指示符。
在一些可行的实施方式中，所述处理器901，具体用于：
所述网络设备通过物理下行控制信道PDCCH，向所述终端通过接收器903发送第二指示符；
或，所述网络设备通过物理下行共享信道PDSCH，向所述终端通过接收器903发送第二指示符；
或，所述网络设备通过媒体接入控制单元MAC CE，向所述终端通过接收器903发送第二指示符。
本申请实施例还提供了一种计算机存储介质，该计算机存储介质中存储有程序指令，所述程序执行时可包括如图2或者图5对应实施例中的通信方法的部分或全部步骤。
本申请实施例还提供了一种计算机程序产品，其中，所述计算机程序产品包括存储了计算机程序的非瞬时性计算机可读存储介质，所述计算机程序可操作来使计算机执行如上述方法实施例中终端或网络设备所描述的部分或全部步骤。该计算机程序产品可以为一个软件安装包。
结合本申请公开内容所描述的方法或者算法的步骤可以硬件的方式来实现，也可以是由处理器执行软件指令的方式来实现。软件指令可以由相应的软件模块组成，软件模块可以被存放于随机存取存储器(Random Access Memory，RAM)、闪存、只读存储器(Read Only Memory，ROM)、可擦除可编程只读存储器(Erasable Programmable ROM，EPROM)、电可擦可编程只读存储器(Electrically EPROM，EEPROM)、寄存器、硬盘、移动硬盘、只读光盘(CD-ROM)或者本领域熟知的任何其它形式的存储介质中。一种示例性的存储介质耦合至处理器，从而使处理器能够从该存储介质读取信息，且可向该存储介质写入信息。当然，存储介质也可以是处理器的组成部分。处理器和存储介质可以位于ASIC中。另外，该ASIC可以位于通信装置如终端中。当然，处理器和存储介质也可以作为分立组件存在于通信装置中。
可以理解，本文中涉及的第一、第二、各种数字编号仅为描述方便进行的区分，并不用来限制本申请实施例的范围。
可以理解，在本申请的各种实施例中，上述各过程的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请实施例的实施过程构成任何限定。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，可通过计算机程序来指令相关的硬件来完成，该的程序可存储于计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。而前述的存储介质包括：ROM、RAM、磁碟或者光盘等各种可存储程序代码的介质。

Claims (22)

  1. 一种通信方法，其特征在于，包括：
    终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
  2. 根据权利要求1所述的方法，其特征在于，在终端向网络设备发送第一指示符之后，还包括：
    所述终端接收所述网络设备发送的第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
  3. 根据权利要求2所述的方法，其特征在于，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
  4. 根据权利要求2所述的方法，其特征在于，所述第一通信资源参数和/或所述第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
  5. 根据权利要求1-4任一项所述的方法，其特征在于，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
    所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
  6. 根据权利要求1-4任一项所述的方法，其特征在于，还包括：
    所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
    所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
  7. 根据权利要求1-6任一项所述的方法，其特征在于，所述终端向网络设备发送第一指示符，包括：
    所述终端通过物理上行控制信道PUCCH，向网络设备发送第一指示符；
    或，所述终端通过物理上行共享信道PUSCH，向网络设备发送第一指示符；
    或，所述终端通过媒体接入控制单元MAC CE，向网络设备发送第一指示符。
  8. 根据权利要求2-6任一项所述的方法，其特征在于，所述终端接收所述网络设备发送的第二指示符，包括：
    所述终端通过物理下行控制信道PDCCH接收所述网络设备发送的第二指示符；
    或，终端通过物理下行共享信道PDSCH接收所述网络设备发送的第二指示符；
    或，终端通过媒体接入控制单元MAC CE接收所述网络设备发送的第二指示符。
  9. 一种通信方法，其特征在于，包括：
    网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源 参数组合为第一对应关系。
  10. 根据权利要求9所述的方法，其特征在于，在网络设备接收终端发送的第一指示符之后，还包括：
    所述网络设备向所述终端发送第二指示符，其中，所述二指示符用于标识第二资源参数组合，所述第二资源参数组合包括第二通信资源参数和第二AI参数，所述第二指示符与所述第二资源参数组合为第二对应关系。
  11. 根据权利要求10所述的方法，其特征在于，所述第一AI参数和/或所述第二AI参数包括以下至少一种：所述终端运行AI操作的算力水平、所述终端使用的AI模型、所述终端运行AI模型的分割点、所述终端运行AI操作的步骤、所述终端在完成一轮AI训练中所需的训练数据量和所述终端利用AI模型的训练结果上报周期。
  12. 根据权利要求10所述的方法，其特征在于，所述第一通信资源参数和/或第二通信资源参数包括以下至少一种：调制方法、编码码率和频谱效率。
  13. 根据权利要求9-12任一项所述的方法，其特征在于，所述第一对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到；以及，
    所述第二对应关系由预定义或无线资源控制RRC信令配置或系统信息配置或媒体接入控制单元MAC CE配置得到。
  14. 根据权利要求9-12任一项所述的方法，其特征在于，还包括：
    所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第一指示符与第一资源参数组合之间的多个对应关系中确定所述第一对应关系；以及，
    所述终端根据下行控制信息DCI指示或无线网络临时标识RNTI或DCI格式，从第二指示符与第二资源参数组合之间的多个对应关系中确定所述第二对应关系。
  15. 根据权利要求9-14任一项所述的方法，其特征在于，所述网络设备接收终端发送的第一指示符，包括：
    所述网络设备通过物理上行控制信道PUCCH接收终端发送的第一指示符；
    或，所述网络设备通过物理上行共享信道PUSCH接收终端发送的第一指示符；
    或，所述网络设备通过媒体接入控制单元MAC CE接收终端发送的第一指示符。
  16. 根据权利要求10-14任一项所述的方法，其特征在于，所述网络设备向所述终端发送第二指示符，包括：
    所述网络设备通过物理下行控制信道PDCCH，向所述终端发送第二指示符；
    或，所述网络设备通过物理下行共享信道PDSCH，向所述终端发送第二指示符；
    或，所述网络设备通过媒体接入控制单元MAC CE，向所述终端发送第二指示符。
  17. 一种通信装置，其特征在于，所述装置包括:
    发送单元，用于终端向网络设备发送第一指示符，其中，所述第一指示符用于标识第一资源参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
  18. 一种通信装置，其特征在于，所述装置包括：
    接收单元，用于网络设备接收终端发送的第一指示符，其中，所述第一指示符用于标识第一资源 参数组合，所述第一资源参数组合包括第一通信资源参数和第一人工智能AI参数，所述第一指示符与所述第一资源参数组合为第一对应关系。
  19. 一种终端，其特征在于，包括处理器、接收器、发射器和存储器，所述处理器、接收器、发射器和存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行如权利要求1-12任一项所述的方法。
  20. 一种网络设备，其特征在于，包括处理器、接收器、发射器和存储器，所述处理器、接收器、发射器和存储器相互连接，其中，所述存储器用于存储计算机程序，所述计算机程序包括程序指令，所述处理器被配置用于调用所述程序指令，执行如权利要求13-24任一项所述的方法。
  21. 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，所述计算机程序包括程序指令，所述程序指令当被处理器执行时使所述处理器执行如权利要求1-8任一项所述的方法。
  22. 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，所述计算机程序包括程序指令，所述程序指令当被处理器执行时使所述处理器执行如权利要求9-16任一项所述的方法。
PCT/CN2020/085998 2020-04-21 2020-04-21 一种通信方法及相关设备 WO2021212347A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP20932266.8A EP4135442A4 (en) 2020-04-21 2020-04-21 COMMUNICATION PROCESS AND ASSOCIATED MECHANISM
PCT/CN2020/085998 WO2021212347A1 (zh) 2020-04-21 2020-04-21 一种通信方法及相关设备
CN202080099536.7A CN115362735A (zh) 2020-04-21 2020-04-21 一种通信方法及相关设备
US18/047,625 US20230066109A1 (en) 2020-04-21 2022-10-18 Communication method and related devices

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/085998 WO2021212347A1 (zh) 2020-04-21 2020-04-21 一种通信方法及相关设备

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/047,625 Continuation US20230066109A1 (en) 2020-04-21 2022-10-18 Communication method and related devices

Publications (1)

Publication Number Publication Date
WO2021212347A1 true WO2021212347A1 (zh) 2021-10-28

Family

ID=78271065

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/085998 WO2021212347A1 (zh) 2020-04-21 2020-04-21 一种通信方法及相关设备

Country Status (4)

Country Link
US (1) US20230066109A1 (zh)
EP (1) EP4135442A4 (zh)
CN (1) CN115362735A (zh)
WO (1) WO2021212347A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114172765A (zh) * 2021-12-03 2022-03-11 中国信息通信研究院 一种无线通信人工智能信道估计方法和设备
WO2022233294A1 (zh) * 2021-05-07 2022-11-10 展讯通信（上海）有限公司 通信方法、装置和电子设备
WO2023070684A1 (zh) * 2021-11-01 2023-05-04 Oppo广东移动通信有限公司 无线通信的方法和设备
WO2023202385A1 (zh) * 2022-04-22 2023-10-26 大唐移动通信设备有限公司 信息传输方法、装置及存储介质
WO2024011581A1 (zh) * 2022-07-15 2024-01-18 华为技术有限公司 一种通信方法及装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190138908A1 (en) * 2018-12-28 2019-05-09 Francesc Guim Bernat Artificial intelligence inference architecture with hardware acceleration
CN110908804A (zh) * 2019-11-26 2020-03-24 北京邮电大学 计算资源分配方法、装置及其模型的训练方法、装置
CN111010890A (zh) * 2018-08-06 2020-04-14 Lg电子株式会社 在无线通信系统的coreset中接收信号的方法和使用该方法的设备
CN111034310A (zh) * 2018-08-03 2020-04-17 Lg电子株式会社 发送或接收参考信号的方法及设备

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009021422A1 (fr) * 2007-08-10 2009-02-19 Huawei Technologies Co., Ltd. Procédé et système permettant d'établir une communication à compression d'en-tête, et entité fonctionnelle de politique de compression d'en-tête
CN105992247A (zh) * 2015-01-30 2016-10-05 中兴通讯股份有限公司 无线数据的收发方法及装置
CN108924910B (zh) * 2018-07-25 2021-03-09 Oppo广东移动通信有限公司 Ai模型的更新方法及相关产品
US11736764B2 (en) * 2019-09-13 2023-08-22 Intel Corporation Artificial intelligence inference on protected media content in a vision processing unit

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111034310A (zh) * 2018-08-03 2020-04-17 Lg电子株式会社 发送或接收参考信号的方法及设备
CN111010890A (zh) * 2018-08-06 2020-04-14 Lg电子株式会社 在无线通信系统的coreset中接收信号的方法和使用该方法的设备
US20190138908A1 (en) * 2018-12-28 2019-05-09 Francesc Guim Bernat Artificial intelligence inference architecture with hardware acceleration
CN110908804A (zh) * 2019-11-26 2020-03-24 北京邮电大学 计算资源分配方法、装置及其模型的训练方法、装置

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
OPPO; CMCC; CHINA TELECOM; CHINA UNICOM; QUALCOMM: "SID Proposal: Study on AI/ML Model Transfer in 5GS", 3GPP DRAFT; S1-193479, vol. SA WG1, 25 November 2019 (2019-11-25), Reno, USA, pages 1 - 3, XP051831222 *
See also references of EP4135442A4 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022233294A1 (zh) * 2021-05-07 2022-11-10 展讯通信（上海）有限公司 通信方法、装置和电子设备
WO2023070684A1 (zh) * 2021-11-01 2023-05-04 Oppo广东移动通信有限公司 无线通信的方法和设备
CN114172765A (zh) * 2021-12-03 2022-03-11 中国信息通信研究院 一种无线通信人工智能信道估计方法和设备
WO2023097869A1 (zh) * 2021-12-03 2023-06-08 中国信息通信研究院 一种无线通信人工智能信道估计方法和设备
CN114172765B (zh) * 2021-12-03 2023-07-07 中国信息通信研究院 一种无线通信人工智能信道估计方法和设备
WO2023202385A1 (zh) * 2022-04-22 2023-10-26 大唐移动通信设备有限公司 信息传输方法、装置及存储介质
WO2024011581A1 (zh) * 2022-07-15 2024-01-18 华为技术有限公司 一种通信方法及装置

Also Published As

Publication number Publication date
EP4135442A4 (en) 2023-04-26
EP4135442A1 (en) 2023-02-15
US20230066109A1 (en) 2023-03-02
CN115362735A (zh) 2022-11-18

Similar Documents

Publication Publication Date Title
WO2021212347A1 (zh) 一种通信方法及相关设备
CN109802758A (zh) 管理在新无线电中的搜索空间之间的控制信道盲搜索
US11356990B2 (en) Method and device for resource allocation
CN102368871B (zh) 一种pdcch资源的配置应用方法及装置
US20230231672A1 (en) Method and apparatus for obtaining resource indication value
RU2735383C1 (ru) Способ передачи данных, оконечное устройство и сетевое устройство
US11695500B2 (en) Communication method, terminal device, and network device
CN103493402B (zh) 用来控制用于固定设备的寻呼的技术
CN109831827B (zh) 数据传输方法、终端和基站
CN112714486A (zh) Pdcch的检测方法、drx配置方法、终端、基站
CN105519222A (zh) 数据传输方法和设备
US20190349978A1 (en) Physical Resource Block Scaling For Data Channel With HARQ Process
TW201818706A (zh) 通訊方法、終端和網路設備
CN111294960A (zh) 识别下行控制信息的方法及设备
WO2020051845A1 (zh) Rs集合的配置方法、装置、设备及存储介质
US11705981B2 (en) Communication method, MCS receiving method, MCS notification method, and device
CN111885716A (zh) 一种资源配置的方法及装置
CN112752259B (zh) 一种终端类型的确定方法、网络设备及终端
CN111132343A (zh) 上行控制信道资源确定方法、终端和网络侧设备
CN108401294B (zh) 控制信息的发送和接收方法、接入网设备和终端设备
US20230020414A1 (en) Terminal and communication method
WO2019197027A1 (en) Methods and apparatuses for selective modulation and coding scheme (mcs) exclusion
US20230041809A1 (en) Terminal and communication method
CA3060623A1 (en) Wireless communication method, terminal device, and network device
EP3855788A1 (en) Method, device and system for determining channel quality information

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: 20932266

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020932266

Country of ref document: EP

Effective date: 20221111