WO2023179577A1 - 一种通信方法及相关装置 - Google Patents

一种通信方法及相关装置 Download PDF

Info

Publication number
WO2023179577A1
WO2023179577A1 PCT/CN2023/082648 CN2023082648W WO2023179577A1 WO 2023179577 A1 WO2023179577 A1 WO 2023179577A1 CN 2023082648 W CN2023082648 W CN 2023082648W WO 2023179577 A1 WO2023179577 A1 WO 2023179577A1
Authority
WO
WIPO (PCT)
Prior art keywords
entity
processing
data
parameters
output
Prior art date
Application number
PCT/CN2023/082648
Other languages
English (en)
French (fr)
Inventor
马江镭
葛屹群
童文
李榕
王坚
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2023179577A1 publication Critical patent/WO2023179577A1/zh

Links

Classifications

    • 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
    • H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
    • H04W16/22—Traffic simulation tools or models

Definitions

  • the present application relates to the field of communication technology, and in particular, to a communication method and related devices.
  • AI Artificial intelligence
  • the network layer such as network optimization, mobility management, resource allocation, etc.
  • the physical layer such as channel coding and decoding, channel prediction, receiver, etc.
  • the current combination of AI and communication mainly involves the design of AI models or training methods for given system parameters in specific scenarios.
  • scenarios such as channel conditions
  • system parameters and other environments change, it needs to be re-designed. Designing corresponding models and retraining them is difficult to deploy and use in general communication systems.
  • Embodiments of the present application provide a communication method and related devices that can adapt to changes in the environment and support effective data transmission and reception.
  • embodiments of the present application provide a communication method for a first processing entity of a first communication device.
  • the method includes: the first processing entity sends a first registration request to the second entity, and the first registration request It includes a first identifier indicating one or more of the model information or data information of the first processing entity, and the model parameters of the second entity are adjusted according to the real-time environment parameters; the first processing entity receives the second The first registration response sent by the entity, the first registration response indicates the input parameters of the second entity; the first processing entity sends a first response confirmation, the first response confirmation indicates the output parameters of the first processing entity Whether it matches the input parameters of the second entity.
  • the second entity may also be called an adaptive entity.
  • the first processing entity registers with the second entity to determine that the output parameters of the first processing entity match the input parameters of the second entity, wherein the model parameters of the second entity can be adjusted according to the real-time environment parameters, which can Adapt to changes in the environment and support effective data sending and receiving.
  • the first response confirmation indicates that the match is successful
  • the method further includes: the first processing entity performs the first processing on the first input data to obtain the first output data, and the first output data is the output of the first processing entity. ; The first processing entity transmits the first output data to the second entity through the first interface.
  • the first processing entity when the output parameter of the first processing entity successfully matches the input parameter of the second entity, the first processing entity transmits the first output data to the second entity, thereby achieving effective data transmission.
  • the first processing entity includes an input adaptation module, a data space conversion module and an output adaptation module;
  • the input adaptation module and the output adaptation module of the first processing entity respectively perform dimension transformation and other processing on the input data and output data of the first processing entity to adapt to the pre-processing module of the first processing entity.
  • the output dimension is the input dimension of the data space transformation module, and the output dimension of the first processing entity is adapted to the output dimension of the subsequent processing module (such as the second entity) of the first processing entity.
  • the first processing includes one or more of quantization, source coding, channel coding, modulation, and MIMO precoding, or is combined with quantization, source coding, channel coding, modulation, and MIMO precoding.
  • quantization source coding, channel coding, modulation, and MIMO precoding.
  • MIMO precoding One or more equivalent treatments in the encoding.
  • one or more operations performed in traditional data transmission and reception can be implemented through the first processing entity.
  • the first processing includes: encoding and modulation, the first input data is a bit stream, and the first output data is a modulated symbol.
  • the first processing includes quantization, source coding, channel coding and modulation, the first input data is a symbol sequence, and the first output data is a modulated symbol sequence.
  • the first processing entity that implements different functions can be registered in the matching second entity to achieve effective data sending and receiving.
  • the first interface defines the interaction process and/or signaling between the first processing entity and the second entity.
  • the first interface defines a process for transmitting one or more of the registration request, the registration response, or the response confirmation.
  • the first response confirmation indicates that the output parameter of the first processing entity successfully matches the input parameter of the second entity.
  • the method further includes: the first processing entity sends a second registration request to the fourth entity, and the second The registration request includes a first identifier, the first identifier indicates one or more of the model information or data information of the first processing entity, and the model parameters of the fourth entity can be adjusted according to the real-time environment parameters; the first processing entity receives the fourth entity A second registration response is sent, the second registration response indicates the input parameters of the fourth entity; the first processing entity sends a second response confirmation, the second response confirmation indicates whether the output parameters of the first processing entity match the input parameters of the fourth entity .
  • the first processing entity may initiate registration with the fourth entity to ensure that the first processing entity registers with the adaptive entity that matches it.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • This implementation can enable the second entity to determine whether the first processing entity is consistent with the corresponding model information of the second entity based on the model information corresponding to the first identification, thereby determining the first registration response, thereby enabling the first processing entity to register to the appropriate On the second entity.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • This implementation can enable the second entity to determine whether the first processing entity is consistent with the corresponding data information of the second entity based on the data information corresponding to the first identification, thereby determining the first registration response, and thereby enabling the first processing entity to register to the appropriate On the second entity.
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity
  • the input parameters of the second entity include the third One or more of the type, dimension, or precision of the input data for the second entity.
  • the first processing entity and the second entity belong to the same communication device.
  • embodiments of the present application provide a communication method for a second entity of a second communication device.
  • the method includes: the second entity receives a first registration request sent by the first processing entity, and the first registration request including a first identifier indicating one or more of the model information or data information of the first processing entity; the model parameters of the second entity are adjusted according to the real-time environment parameters; the second entity reports to the first processing entity A first registration response is sent, the first registration response indicates the input parameters of the second entity; the input parameters of the second entity are used to determine whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the second entity receives the registration request of the first processing entity and feeds back the input parameters of the second entity to the first processing entity, so that the first processing entity determines the output parameters of the first processing entity and the input parameters of the second entity.
  • Matching in which the model parameters of the second entity can be adjusted according to real-time environment parameters, can adapt to changes in the environment, and support effective data transmission and reception.
  • the second entity receives a first response confirmation indicating that the output parameter of the first processing entity matches the input parameter of the second entity successfully, and the method further includes: second The entity receives the first output data through the first interface, and the first output data is the output data of the first processing entity; the second entity performs a second process on the first output data to obtain the second output data; the second entity outputs the second output data. Output Data.
  • the second entity outputting the second output data includes: the second entity sending the second output data to the third communication device.
  • the second entity performs a second process on the first output data to obtain the second output data, including: the second entity maps the first output data to the second output data.
  • the mapping method can be adjusted according to real-time environmental parameters.
  • mapping method from the first output data to the second output data according to real-time environmental parameters, it can adapt to the changing transmission environment and support effective data transmission and reception.
  • the first interface defines the interaction process and/or signaling between the first processing entity and the second entity.
  • the first interface defines a process of transmitting one or more of a registration request, a registration response, or a response confirmation.
  • the second entity while sending the first registration response, the second entity starts the first response confirmation timing, and when the response confirmation timing expires, sends the first response confirmation timeout indication to the first control entity.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity
  • the input parameters of the second entity include the third One or more of the type, dimension, or precision of the input data for the second entity.
  • the first processing entity and the second entity belong to the same communication device.
  • embodiments of the present application provide a communication method for a third processing entity of a third communication device.
  • the method includes: the third processing entity sends a second registration request to the second entity, and the second registration request including a second identifier indicating one or more of the model information or data information of the third processing entity, and the model parameters of the second entity are adjusted according to the real-time environment parameters; the third processing entity receives the second A second registration response sent by the entity, the second registration response indicating the output parameters of the second entity; the third processing entity sends a second response confirmation, the second response confirmation The confirmation indicates whether the input parameters of the third processing entity match the output parameters of the second entity.
  • the second entity may also be called an adaptive entity.
  • the third processing entity registers with the second entity to determine that the input parameters of the third processing entity match the output parameters of the second entity, wherein the model parameters of the second entity can be adjusted according to the real-time environment parameters, which can Adapt to changes in the environment and support effective data sending and receiving.
  • the second response confirmation indicates that the input parameters of the third processing entity match the output parameters of the second entity.
  • the method further includes: the third processing entity obtains the second output data transmitted by the second entity, and The second output data is subjected to a third process to obtain third data.
  • the third processing entity when the input parameters of the third processing entity successfully match the output parameters of the second entity, the third processing entity receives the second output data transmitted by the second entity, and performs the third processing on the second output data to obtain The third data realizes effective data transmission.
  • the third processing entity includes an input adaptation module, a data space conversion module and an output adaptation module;
  • the input configuration module is used to adapt the dimensions of the second output data and the input dimensions of the data space conversion module;
  • data The spatial transformation module is used to perform third processing on the second output data to obtain second intermediate output data;
  • the output adaptation module is used to perform dimension transformation on the second intermediate output data to obtain third data.
  • the input adaptation module and the output adaptation module of the third processing entity respectively perform dimension transformation and other processing on the input data and output data of the third processing entity to adapt to the pre-sequence processing module (such as the output dimension of the second entity) and the input dimension of the data space transformation module, and adapt the output dimension of the third processing entity to the output dimension of the subsequent processing module of the third processing entity.
  • the third processing is the inverse operation of the first processing in the first aspect, and the third processing includes one or more of MIMO decoding, demodulation, channel decoding, source decoding, or One or more equivalent processes of MIMO decoding, demodulation, channel decoding, and source decoding.
  • one or more operations performed in traditional data transmission and reception can be implemented through the third processing entity.
  • the third processing includes: demodulation and channel decoding, the second output data is a modulated symbol, and the third data is a bit stream.
  • the third processing includes demodulation, channel decoding, and source decoding
  • the second output data is a modulated symbol sequence
  • the third data is a symbol sequence
  • the third processing entity that implements different functions can be registered in the matching second entity to achieve effective data sending and receiving.
  • the second interface defines the interaction process and/or signaling between the third processing entity and the second entity.
  • the second interface defines a process for transmitting one or more of the registration request, the registration response, or the response confirmation.
  • the second response confirmation indicates that the input parameter of the third processing entity fails to match the output parameter of the second entity.
  • the method further includes: the third processing entity sends a second registration request to the fourth entity, and the second The registration request includes a second identification.
  • the second identification indicates one or more of the model information or data information of the third processing entity.
  • the model parameters of the fourth entity can be adjusted according to the real-time environment parameters; the third processing entity receives the fourth entity.
  • the second registration response is sent, and the second registration response indicates the input parameters of the fourth entity; the third processing entity sends a second response confirmation, and the second response confirmation indicates whether the input parameters of the third processing entity match the input parameters of the fourth entity. .
  • the third processing entity when the input parameter of the third processing entity fails to match the output parameter of the second entity, the third processing entity The entity can initiate registration with the fourth entity to ensure that the third processing entity registers with the adaptive entity that matches it.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • This implementation can enable the second entity to determine whether the third processing entity is consistent with the corresponding model information of the second entity based on the model information corresponding to the second identification, thereby determining the second registration response, and thereby causing the third processing entity to register to the appropriate On the second entity.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • This implementation can enable the second entity to determine whether the third processing entity is consistent with the corresponding data information of the second entity based on the data information corresponding to the second identification, thereby determining the second registration response, and thereby allowing the third processing entity to register to the appropriate On the second entity.
  • the input parameters of the third processing entity include one or more of the type, dimension or precision of the input data of the third processing entity
  • the output parameters of the second entity include the third processing entity.
  • embodiments of the present application provide a communication method for a second entity of a fourth communication device.
  • the method includes: the second entity receives a second registration request sent by a third processing entity, and the second registration request including a second identifier indicating one or more of the model information or data information of the third processing entity, and the model parameters of the second entity are adjusted according to the real-time environment parameters; the second entity reports to the third processing entity A second registration response is sent, the second registration response indicates the output parameters of the second entity; the output parameters of the second entity are used to determine whether the input parameters of the third processing entity match the output parameters of the second entity.
  • the second entity receives the registration request of the third processing entity and feeds back the output parameters of the second entity to the third processing entity, so that the third processing entity determines the input parameters of the third processing entity and the output parameters of the second entity.
  • Matching in which the model parameters of the second entity can be adjusted according to real-time environment parameters, can adapt to changes in the environment, and support effective data transmission and reception.
  • the second entity receives a second response confirmation indicating that the input parameters of the third processing entity and the output parameters of the second entity are successfully matched.
  • the method further includes: second The entity performs a second process on the first output data to obtain second output data, and the first output data is the output data of the first processing entity; the second entity transmits the second output data to the third processing entity through the second interface.
  • the second entity performs a second process on the first output data to obtain the second output data, including: the second entity maps the first output data to the second output data.
  • the mapping method can be adjusted according to real-time environmental parameters.
  • mapping method from the first output data to the second output data according to real-time environmental parameters, it can adapt to the changing transmission environment and support effective data transmission and reception.
  • the second interface defines the interaction process and/or signaling between the third processing entity and the second entity.
  • the second interface defines a process of transmitting one or more of a registration request, a registration response, or a response confirmation.
  • the second entity while sending the second registration response, the second entity starts a second response confirmation timing, and when the response confirmation timing expires, sends a second response confirmation timeout indication to the third control entity.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the data information includes one or more of the following: processing data type, processing data dimension or processing physical data accuracy.
  • the input parameters of the third processing entity include one or more of the type, dimension or precision of the input data of the third processing entity
  • the output parameters of the second entity include the third processing entity.
  • embodiments of the present application further provide a communication device, which can be used for the first processing entity of the first aspect.
  • the communication device can be a terminal or a network device, or can be a device in a terminal or a network device. (For example, a chip, or a chip system, or a circuit), or a device that can be used in conjunction with a terminal or network device.
  • the communication device may include modules or units that perform one-to-one correspondence with the methods/operations/steps/actions described in the first aspect.
  • the modules or units may be hardware circuits, software, or It can be implemented by hardware circuit combined with software.
  • the communication device may include a processing unit and a transceiver unit.
  • the processing unit is used to call the transceiver unit to perform reception and/or transmission functions.
  • the transceiver unit is configured to send a first registration request to the second entity, where the first registration request includes a first identifier indicating one or more items of model information or data information of the communication device, wherein, The model parameters of the second entity are adjusted according to the real-time environment parameters; the transceiver unit is also used to receive a first registration response sent by the second entity, the first registration response indicating the input parameters of the second entity; the transceiver unit is also used to send the first A first response acknowledgment indicating whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the processing unit is further configured to perform first processing on the first input data to obtain first output data; and the transceiver unit is further configured to transmit the first output data to the second entity through the first interface.
  • the transceiver unit includes an input adaptation module and an output adaptation module
  • the processing unit includes a data space conversion module
  • the input configuration module is used to adapt the dimensions of the first input data to the input of the data space conversion module.
  • Dimension the data space conversion module is used to perform the first processing on the first input data to obtain the first intermediate output data
  • the output adaptation module is used to perform dimension transformation on the first intermediate output data to obtain the first output data.
  • the first processing includes one or more of quantization, source coding, channel coding, modulation, and MIMO precoding, or is combined with one or more of quantization, source coding, channel coding, modulation, and MIMO precoding.
  • quantization source coding, channel coding, modulation, and MIMO precoding
  • the first response confirmation indicates that the output parameters of the communication device successfully match the input parameters of the second entity
  • the transceiver unit is also configured to send a second registration request to the fourth entity
  • the second registration request includes the first Identification
  • the first identification indicates one or more of the model information or data information of the first processing entity
  • the model parameters of the fourth entity can be adjusted according to the real-time environment parameters
  • the transceiver unit is also used to receive the second information sent by the fourth entity.
  • Registration response indicates the input parameters of the fourth entity
  • the transceiver unit is also configured to send a second response confirmation
  • the second response confirmation indicates whether the output parameters of the first processing entity match the input parameters of the fourth entity.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity
  • the input parameters of the second entity include the third One or more of the type, dimension, or precision of the input data for the second entity.
  • the communication device and the second entity belong to the same communication device.
  • the processing unit is a processor
  • the transceiver unit is a transceiver
  • embodiments of the present application further provide a communication device, which can be used for the second entity of the second aspect.
  • the communication device can be a terminal or a network device, or can also be a device in a network device (for example, chip, or chip system, or circuit), or a device that can be used with network equipment.
  • the communication device may include modules or units that perform one-to-one correspondence with the methods/operations/steps/actions described in the second aspect.
  • the modules or units may be hardware circuits, software, or It can be implemented by hardware circuit combined with software.
  • the communication device may include a processing unit and a transceiver unit.
  • the processing unit is used to call the transceiver unit to perform reception and/or transmission functions.
  • the transceiver unit is configured to receive a first registration request sent by a first processing entity.
  • the first registration request includes a first identification, and the first identification indicates one or more of the model information or data information of the first processing entity. item, the model parameters of the communication device are adjusted according to real-time environment parameters; the transceiver unit is used to send a first registration response to the first processing entity, and the first registration response indicates the input parameters of the communication device; the input parameters of the communication device are used to Determine whether the output parameters of the first processing entity match the input parameters of the communication device.
  • the transceiver unit is configured to receive a first response confirmation indicating that the output parameter of the first processing entity matches the input parameter of the communication device successfully, and the transceiver unit is also configured to pass the first response confirmation.
  • the interface receives the first output data, which is the output data of the first processing entity; the processing unit is used to perform a second process on the first output data to obtain the second output data; the transceiver unit is also used to output the second output data.
  • the transceiver unit is specifically configured to send the second output data to the third communication device.
  • the processing unit is specifically configured to map the first output data to the second output data.
  • the mapping method can be adjusted according to real-time environmental parameters.
  • the processing unit is also configured to start the first response confirmation timing while the transceiver unit sends the first registration response; the transceiver unit is also configured to send the first response confirmation timing to the first control entity when the response confirmation timing expires.
  • First response acknowledgment timeout indication is also configured to start the first response confirmation timing while the transceiver unit sends the first registration response; the transceiver unit is also configured to send the first response confirmation timing to the first control entity when the response confirmation timing expires.
  • First response acknowledgment timeout indication is also configured to start the first response confirmation timing while the transceiver unit sends the first registration response; the transceiver unit is also configured to send the first response confirmation timing to the first control entity when the response confirmation timing expires.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity
  • the input parameters of the second entity include the third One or more of the type, dimension, or precision of the input data for the second entity.
  • the communication device and the first processing entity belong to the same communication device.
  • embodiments of the present application further provide a communication device, which can be used for the third processing entity of the third aspect.
  • the communication device can be a terminal or a network device, or can be a device in a terminal or a network device. (For example, a chip, or a chip system, or a circuit), or a device that can be used in conjunction with a terminal or network device.
  • the communication device may include modules or units that perform one-to-one correspondence with the methods/operations/steps/actions described in the third aspect.
  • the modules or units may be hardware circuits, software, or It can be implemented by hardware circuit combined with software.
  • the communication device may include a processing unit and a transceiver unit.
  • the processing unit is used to call the transceiver unit to perform reception and/or transmission functions.
  • the transceiver unit is used to send a second registration request to the second entity, and the second The registration request includes a second identification, the second identification indicates one or more of the model information or data information of the communication device, and the model parameters of the second entity are adjusted according to the real-time environment parameters;
  • the transceiver unit is also used to receive the second A second registration response sent by the entity, the second registration response indicating the output parameters of the second entity;
  • the transceiver unit is also used to send a second response confirmation, the second response confirmation indicating whether the input parameters of the communication device are consistent with the second entity The output parameters match.
  • the second response confirmation indicates that the input parameters of the third processing entity match the output parameters of the second entity.
  • the transceiver unit is also used to obtain the second output data transmitted by the second entity.
  • the processing unit also uses Perform third processing on the second output data to obtain third data.
  • the transceiver unit includes an input adaptation module and an output adaptation module
  • the processing unit includes a data space conversion module
  • the input configuration module is used to adapt the dimensions of the second output data and the input dimensions of the data space conversion module
  • the data space conversion module is used to perform third processing on the second output data to obtain second intermediate output data
  • the output adaptation module is used to perform dimension transformation on the second intermediate output data to obtain third data.
  • the third processing is the inverse operation of the first processing in the first aspect, and the third processing includes one or more of MIMO decoding, demodulation, channel decoding, source decoding, or One or more equivalent processes of MIMO decoding, demodulation, channel decoding, and source decoding.
  • the second response confirmation indicates that the input parameters of the third processing entity and the output parameters of the second entity fail to match
  • the transceiver unit is also configured to send a second registration request to the fourth entity, and the second registration request includes the third Two identifiers, the second identifier indicates one or more of the model information or data information of the third processing entity, the model parameters of the fourth entity can be adjusted according to the real-time environment parameters; the transceiver unit is also used to receive the third entity sent by the fourth entity.
  • Two registration responses indicates the input parameters of the fourth entity; the transceiver unit is also configured to send a second response confirmation, the second response confirmation indicates whether the input parameters of the third processing entity match the input parameters of the fourth entity.
  • embodiments of the present application further provide a communication device, which can be used in the second entity of the fourth aspect.
  • the communication device can be a terminal or network equipment, or can be a device in a terminal or network equipment (for example, a chip, or a chip system, or a circuit), or a device that can be used in conjunction with a terminal or network device.
  • the communication device may include modules or units that perform one-to-one correspondence with the methods/operations/steps/actions described in the fourth aspect.
  • the modules or units may be hardware circuits, software, or It can be implemented by hardware circuit combined with software.
  • the communication device may include a processing unit and a transceiver unit.
  • the processing unit is used to call the transceiver unit to perform reception and/or transmission functions.
  • the transceiver unit is configured to receive a second registration request sent by the third processing entity.
  • the second registration request includes a second identification, and the second identification indicates one or more of the model information or data information of the third processing entity.
  • the model parameters of the communication device are adjusted according to the real-time environment parameters; the transceiver unit is used to send a second registration response to the third processing entity, and the second registration response indicates the output parameters of the communication device; the output parameters of the communication device are used to Determine whether the input parameters of the third processing entity match the output parameters of the second entity.
  • the transceiver unit is configured to receive a second response confirmation, which indicates that the input parameter of the third processing entity matches the output parameter of the communication device successfully.
  • the processing unit is used to perform second processing on the first output data to obtain second output data, where the first output data is the output data of the first processing entity; the transceiver unit is also used to output the second output data.
  • the transceiver unit is also used to output the second output data.
  • the processing unit is specifically configured to map the first output data to the second output data.
  • the mapping method can be adjusted according to real-time environmental parameters.
  • the processing unit is also configured to start the second response confirmation timing while the transceiver unit sends the second registration response.
  • the transceiver unit is also configured to send a second response confirmation timeout indication to the third control entity when the response confirmation timer expires.
  • the model information includes one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the data information includes one or more of the following: processing data type, processing data dimension, or processing data precision.
  • the input parameters of the third processing entity include one or more of the type, dimension or precision of the input data of the third processing entity
  • the output parameters of the second entity include the third processing entity.
  • embodiments of the present application further provide a communication device, including a processor, for implementing the method of the first aspect and various possible implementations thereof.
  • the processor implements the above method through a logic circuit; in another possible implementation, the processor implements the above method by executing instructions.
  • the processor is configured to output a first registration request, the first registration request including a first identification indicating one or more items of model information or data information of the communication device, wherein the second entity's The model parameters are adjusted according to the real-time environment parameters; the processor is also used to receive a first registration response sent by the second entity, the first registration response indicating the input parameters of the second entity; the processor is also used to send a first response confirmation, the The first response confirmation indicates whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the processing unit is further configured to perform a first process on the first input data to obtain the first output data, and transmit the first output data to the second entity through the first interface.
  • embodiments of the present application further provide a communication device, including a processor, for implementing the method of the second aspect and various possible implementations thereof.
  • the processor implements the above method through a logic circuit; in another possible implementation, the processor implements the above method by executing instructions.
  • the processor is configured to receive a first registration request sent by a first processing entity, where the first registration request includes a first identifier, and the first identifier indicates one or more items of model information or data information of the first processing entity.
  • the model parameters of the communication device are adjusted according to the real-time environment parameters; the processor outputs a first registration response, and the first registration response indicates the input parameters of the communication device; the input parameters of the communication device are used to determine the output parameters of the first processing entity Whether it matches the input parameters of the communication device.
  • embodiments of the present application further provide a communication device, including a processor, for implementing the method of the third aspect and various possible implementations thereof.
  • the processor implements the above method through a logic circuit; in another possible implementation, the processor implements the above method by executing instructions.
  • the communication device further includes a memory for storing the instruction.
  • the processor is configured to output a second registration request.
  • the second registration request includes a second identification indicating one or more of the model information or data information of the communication device.
  • the model parameters of the second entity are based on Adjust real-time environment parameters; the processor is also configured to receive a second registration response sent by the second entity, the second registration response indicating the output parameters of the second entity; the processor is also configured to output a second response confirmation, the second The response confirmation indicates whether the input parameters of the communication device match the output parameters of the second entity.
  • embodiments of the present application further provide a communication device, including a processor, for implementing the method of the fourth aspect and various possible implementations thereof.
  • the processor implements the above method through a logic circuit; in another possible implementation, the processor implements the above method by executing instructions.
  • the communication device also includes a device for storing memory for this instruction.
  • the processor is configured to receive a second registration request sent by the third processing entity, the second registration request includes a second identification, and the second identification indicates one or more of the model information or data information of the third processing entity.
  • the model parameters of the communication device are adjusted according to the real-time environment parameters; the processor is also used to output a second registration response, the second registration response indicates the output parameters of the communication device; the output parameters of the communication device are used to determine the third Process whether the input parameters of the entity match the output parameters of the communication device.
  • embodiments of the present application further provide a communication device, including a processor, configured to execute a computer program (or computer-executable instruction) stored in a memory.
  • a computer program or computer-executable instruction
  • the computer program or computer-executable instruction
  • the device When the computer program (or computer-executable instruction) is executed , causing the device to perform the method in the first aspect and each possible implementation of the first aspect, or causing the device to perform the method in the second aspect and each possible implementation of the second aspect, or causing the device to perform the third aspect
  • processor and memory are integrated;
  • the above-mentioned memory is located outside the communication device.
  • the communication device also includes a communication interface, which is used for the communication device to communicate with other devices, such as sending or receiving data and/or signals.
  • the communication interface may be a transceiver, a circuit, a bus, a module, or other types of communication interfaces.
  • embodiments of the present application further provide a communication device for performing the method in the above first aspect and its various possible implementations.
  • embodiments of the present application further provide a communication device for performing the method in the above-mentioned second aspect and its various possible implementations, and/or the method in the fourth aspect and its various possible implementations. method.
  • embodiments of the present application further provide a communication device for performing the method in the above third aspect and its various possible implementations.
  • embodiments of the present application further provide a computer-readable storage medium that stores a computer program (or computer-executable instructions), wherein the computer program (or computer-executable instructions) is When the processor is executed, the first aspect and any possible implementation thereof, the second aspect and any possible implementation thereof, the third aspect and any possible implementation thereof, or the fourth aspect and any possible implementation thereof, are caused.
  • One possible implementation is that some or all of the steps of the method described are performed.
  • embodiments of the present application also provide a computer program product including computer-executable instructions.
  • the computer program product When the computer program product is run, the above-mentioned first aspect and any possible implementation thereof, the second aspect Some or all steps of the method described in any possible implementation thereof, the third aspect and any possible implementation thereof, or the fourth aspect and any possible implementation thereof are executed.
  • embodiments of the present application also provide a computer program including computer-executable instructions.
  • the computer program When the computer program is run, the above-mentioned first aspect and any possible implementation thereof, the second aspect and its Some or all steps of the method described in any possible implementation, the third aspect and any possible implementation thereof, or the fourth aspect and any possible implementation thereof are executed.
  • embodiments of the present application further provide a chip system.
  • the chip system includes a processor and may also include a memory, for implementing the above-mentioned first aspect and any possible implementation thereof, the second aspect and any possible implementation thereof, and the second aspect and any possible implementation thereof. a possible realization, The method described in the third aspect and any possible implementation thereof or the fourth aspect and any possible implementation thereof.
  • the chip system can be composed of chips or include chips and other discrete devices.
  • embodiments of the present application further provide a communication system, including the communication device provided by the fifth aspect and its various possible implementations, the communication device provided by the sixth or eighth aspect and its various possible implementations, A seventh aspect and its various possible implementations provide a communication device.
  • Figure 1 is a schematic structural diagram of an autoencoder.
  • Figure 2 is a schematic architectural diagram of a communication system applicable to the embodiment of the present application.
  • Figure 3 is a schematic diagram of an artificial intelligence communication transceiver.
  • Figure 4 is a schematic diagram of the communication architecture provided by the embodiment of the present application.
  • Figure 5 is an interactive schematic diagram of the communication method 500 provided by the embodiment of the present application.
  • Figure 6 is an interactive schematic diagram of the communication method 600 provided by the embodiment of the present application.
  • Figure 7 is an interactive schematic diagram of the communication method 700 provided by the embodiment of the present application.
  • Figure 8 is an interactive schematic diagram of the communication method 800 provided by the embodiment of the present application.
  • Figure 9 is an interactive schematic diagram of the communication method 900 provided by the embodiment of the present application.
  • Figure 10 is an interactive schematic diagram of the communication method 900 provided by the embodiment of the present application.
  • Figure 11 is a schematic structural diagram of a processing entity provided by an embodiment of the present application.
  • Figure 12 is a schematic structural diagram of an adaptive entity provided by an embodiment of the present application.
  • Figure 13 is an example of implementing the method provided by the embodiment of the present application through a neural network and an adaptive iterative algorithm.
  • Figure 14 is a schematic structural diagram of a communication device provided by an embodiment of the present application.
  • Figure 15 is a schematic structural diagram of a communication device provided by an embodiment of the present application.
  • Figure 16 is a schematic structural diagram of a communication device provided by an embodiment of the present application.
  • Embodiments of the present application provide a communication method and related devices that can adapt to changes in the environment and support effective data transmission and reception.
  • FIG. 1 is a schematic structural diagram of an autoencoder.
  • the input data x is mapped/compressed to the variable z in the latent space through the f function (Encoder), and then from the latent space variable z through the g function (Decoder) Recover data
  • the f function and g function can be realized by neural network, then the optimization goal can be Indicates finding the parameters of the f function and g function to minimize the error in recovering the data.
  • the tasks completed by the communication system are very similar to those of the autoencoder.
  • the communication system can be analogized to a distributed autoencoder (DAE).
  • DAE distributed autoencoder
  • the transmitter sends hidden space variables (waveforms) through the channel, and the information is restored at the receiver.
  • waveforms hidden space variables
  • Autoencoders in the computer field assume that the forward and backward propagation between the encoder and the decoder are ideal propagation, and in systems without communication, the channel is variable, so during training and inference The impact of channel transmission needs to be considered during the processing process.
  • any embodiment described as “exemplary” or “for example” or Aspects are not to be construed as preferred or advantageous over other embodiments or designs.
  • the meaning of “plurality” refers to two or more.
  • multiple processing units refers to two or more processing units.
  • the technical solution of this application can be applied to cellular systems related to the 3rd generation partnership project (3GPP), such as 4th generation (4G) communications such as long term evolution (LTE) systems.
  • 3GPP 3rd generation partnership project
  • 4G 4th generation
  • LTE long term evolution
  • System 5th generation communication system
  • NR new radio
  • FIG. 1 is an example of a communication system suitable for embodiments of the present application.
  • a communication system 100 includes at least one terminal device 110 and at least one access network device 120 .
  • the terminal device 110 mentioned in the embodiment of this application may be a device with a wireless transceiver function, and may communicate with the access network device 120 or other terminal devices.
  • the terminal equipment 110 may specifically refer to user equipment (UE), access terminal, subscriber unit, user station, mobile station, remote station, remote terminal, mobile device, user terminal, wireless communication Device, user agent, or user device.
  • UE user equipment
  • the terminal device may also be a satellite phone, a cellular phone, a smartphone, a wireless data card, a wireless modem, a machine type communications device, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (wireless local) loop (WLL) station, personal digital assistant (PDA), handheld device with wireless communication capabilities, computing device or other processing device connected to a wireless modem, vehicle-mounted equipment, communication equipment carried on high-altitude aircraft, wearable Equipment, drones, robots, terminals in device-to-device (D2D) communication, terminals in vehicle outreach (vehicle to everything, V2X), virtual reality (VR) terminal equipment, Augmented reality (AR) terminal equipment, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical, smart grid ), wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, or terminal equipment in communication networks evolved after 5G, etc. , this application is not limited.
  • SIP session
  • the access network device 120 is a device with wireless transceiver functions and is used to communicate with the terminal device 110 .
  • the access network device may be a node in a radio access network (radio access network, RAN), which may be called a RAN node or network device, or may also be called a base station. It can be an evolved base station (evolved Node B, eNB or eNodeB) in LTE; or it can be a base station in a 5G network such as gNodeB (gNB) or a public land mobile network (PLMN) evolved after 5G. Base stations, or non-3rd generation partnership project (3GPP) access equipment, etc.
  • the network equipment in the embodiment of the present application may include various forms of base stations, such as: macro base stations, micro base stations (also called small stations), relay stations, transmission points (transmitting and receiving points, TRPs), and transmission points.
  • base stations such as: macro base stations, micro base stations (also called small stations), relay stations, transmission points (transmitting and receiving points, TRPs), and transmission points.
  • TP mobile switching center and device-to-device
  • D2D vehicle outreach
  • V2X machine-to-machine
  • M2M machine-to-machine
  • base station functions may also include centralized units (CU) and distributed units (DU) in cloud radio access network (C-RAN) systems, non-terrestrial communications Network equipment in a non-terrestrial network (NTN) communication system is not specifically limited in the embodiments of this application.
  • CU centralized units
  • DU distributed units
  • C-RAN cloud radio access network
  • NTN non-terrestrial communications Network equipment in a non-terrestrial network (NTN) communication
  • the communication system shown in Figure 2 may also include core network equipment (not shown in the figure), and the access network equipment 120 can interact with the core network equipment to provide communication services to terminals.
  • core network equipment can provide communication connections, authentication, management, policy control, and bearer of data services for terminals.
  • the device used to implement the function of the terminal device may be a terminal device; it may also be a device that can support the terminal device to implement the function, such as a chip system.
  • the device can be installed in a terminal device or used in conjunction with the terminal device.
  • the chip system may be composed of chips, or may include chips and other discrete devices.
  • the device used to implement the function of the access network device may be the access network device; it may also be a device that can support the access network device to implement the function, such as a chip system.
  • the device can be installed in the access network equipment or used in conjunction with the access network equipment.
  • the terminal equipment may also have AI processing capabilities
  • the access network equipment may also have AI processing capabilities, which is suitable for scenarios where AI and communication are combined.
  • One possible scenario of combining AI and communication is an AI communication transceiver.
  • one or more modules of the transmitter and receiver are implemented by an AI model (such as a neural network), thereby achieving excellent global performance. .
  • the current combination of AI and communication mainly involves the design of AI models or training methods for given system parameters in specific scenarios.
  • scenarios such as channel conditions
  • system parameters and other environments change, it needs to be re-designed. Designing corresponding models and retraining them is difficult to deploy and use in general communication systems.
  • embodiments of the present application provide a communication method, related devices and systems that can adapt to changes in environment such as channel conditions and user needs to achieve effective data transmission and reception.
  • Figure 4 is an architecture diagram of a communication system provided by an embodiment of the present application.
  • the communication system architecture includes a first processing module 410 , a second processing module 420 , a third processing module 430 , a first interface 440 and a second interface 450 .
  • the first processing module 410 obtains and processes data 1 to obtain data 2, and transmits data 2 to the second processing module 420 through the first interface 440;
  • the second processing module 420 obtains and processes data 2 to obtain data 3, and transmits data 3 through
  • the second interface 450 transmits to the third processing module 430, and the third processing module 430 obtains and processes data 2 to obtain data 4.
  • the first processing module 410 can be set at the sending end, and the data 1 can be a bit stream or a symbol.
  • the first processing module 410 can perform source coding, channel coding, modulation, multiple-input multiple-output (multiple-output) on the data 1
  • One or more operations such as input multiple-output (MIMO) precoding, or operations equivalent to one or more of the above operations
  • the third processing module 430 can be set at the receiving end, for example, the third The processing module 430 can perform some or more operations such as demodulation, channel decoding, and source decoding on the acquired data 3, or perform operations equivalent to one or more of the above operations to obtain data 4, where the data 4 can be a bit stream or symbol obtained by recovering data 1.
  • the first processing module 410 and/or the third processing module 430 may be implemented through a neural network model.
  • the first processing module 410 and/or the third processing module 430 can be configured through offline pre-training using massive environmental data (such as channel data), and training may not be performed during actual use or inference.
  • the first interface 440 defines the interaction process and/or signaling between the first processing module 410 and the second processing module 420.
  • the second interface 450 defines the interaction process and/or signaling between the second processing module 420 and the third processing module 430.
  • the first interface 440 and/or the second interface 450 may be logical interfaces or physical interfaces.
  • the second processing module 420 can also be called an adaptive (selfadaptive) module. During the communication process, it can be adjusted based on environmental data. The above “adjustment” can be achieved through neural network training, or it can be solved by analyzing expressions. now.
  • the first processing module 410 and the third processing module 430 can adopt offline training, and the second processing module 420 can be adjusted according to changes in the actual environment, and can be realized by adapting to changes in the environment with lower complexity. Effective data sending and receiving.
  • the first processing module and the third processing module are obtained through offline training of massive environmental data.
  • One or more first processing modules and/or one or more third processing modules may be pre-stored in the communication device.
  • the one or more first processing modules and/or the one or more third processing modules may be downloaded from the model server or provided by a third party.
  • the communication device can select the first processing module or the third processing module according to its own computing power, storage capacity, power consumption, and task type.
  • the second processing module is a module set locally on the communication device that can be adjusted based on real-time environmental data.
  • the input/output parameters or model parameters of the second processing module may change, which may cause the first The processing module and the second processing module do not match, or the third processing module and the second processing module do not match. Therefore, when the communication device selects the first processing module/third processing module and the second processing module, it needs to consider the adaptation problem between them.
  • communication devices at both sending and receiving ends can respectively select a first processing module and a third processing module to jointly implement data sending and receiving. The selection of the first processing module and the selection of the third processing module also need to be adapted to achieve effective communication.
  • embodiments of the present application provide a registration process.
  • the first processing module of the communication device on the sending side can adapt to the second processing module inside the communication device through the registration process, and can also register to the second processing module.
  • the module is adapted to the third processing module at the receiving end, where the second processing module can be adjusted according to the actual transmission environment, so that each processing module completes data processing and adapts to the changing communication environment to achieve effective data sending and receiving.
  • the interaction process and/or signaling between the processing entity and the adaptive entity can be defined through an interface between the processing entity and the adaptive entity, for example, the above-mentioned first interface and second interface.
  • the interaction process can include registration process, data transmission process, etc.
  • Parameters involved in signaling may include identification information, data type, data accuracy, input dimensions, output dimensions, etc. transmitted between the processing entity and the adaptive entity.
  • the values of parameters involved in the signaling defined by the first interface and the second interface may be predefined by standards, or may be semi-statically configured or dynamically configured.
  • the identification information can be related to the task type of the specific implementation.
  • the task type can be indicated through radio resource control (RRC) signaling, for example, a task type or task identification field added in RRC.
  • RRC radio resource control
  • the data type and data precision can be related to the performance requirements of the task.
  • the performance requirements of the task can be indicated through RRC signaling, for example, adding a performance requirement field in RRC.
  • the input dimension and the output dimension can be related to actual transmission resources, transmission strategies, etc., and the transmission resources or transmission strategies can be indicated through downlink control information (DCI) signaling.
  • the transmission strategy can be a modulation and coding scheme (MCS). For example, parameters such as the number of information bits, the number of coded bits, and the number of modulated symbols after modulation are determined based on the transmission resources and MCS level, and these parameters are used to determine the input dimension and/or the output dimension.
  • MCS modulation and coding scheme
  • this embodiment of the present application provides a communication method 500.
  • the method is applicable to the sending end and involves a first control entity, a first processing entity and a second entity.
  • the first control entity is used to control or manage the first processing entity, the second entity and the interaction between them; the first processing entity and the second entity respectively correspond to the first processing module 410 in the communication architecture shown in Figure 4 and a second processing module 420.
  • the model corresponding to the first processing entity is obtained through offline training, and the model corresponding to the second entity can be adjusted according to environmental information.
  • the first control entity, the first processing entity and the second entity may be physically arranged on the same or different communication devices, and the communication device may be the above-mentioned terminal device or access network device.
  • the specific content of the communication method 500 is introduced below.
  • the first control entity sends a first registration instruction to the first processing entity.
  • the first processing entity receives the registration instruction, where the registration instruction includes the identification information of the second entity.
  • the first control entity selects a target processing entity, ie, a first processing entity, from one or more processing entities, and selects a target adaptive entity, ie, a second entity, from one or more adaptive entities. , and sends a registration instruction including the identification information of the second entity to the first processing entity.
  • the first control entity selects the target processing entity and the target adaptive entity according to the task type.
  • This task type can correspond to one or more of: quantization, source coding, channel coding, modulation, and MIMO precoding.
  • the first control entity is a protocol layer entity above the physical layer, which may also be called a high-level entity.
  • the first control entity is a media access control (media access control, MAC) layer entity, and the registration indication is MAC layer signaling.
  • the first processing entity sends a first registration request to the second entity.
  • the second entity receives the first registration request, and the first registration request includes the first identifier.
  • the first processing entity sends a first registration request to the second entity according to the first registration instruction received in S500.
  • the first processing entity selects the second entity according to its corresponding task type and sends the first registration request to it.
  • the first identification indicates one or more items of model information or data information of the first processing entity.
  • Model information can be called a model description and can include one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • the model purpose indicates the task type corresponding to the first processing entity.
  • model uses include: one or more of quantization, source coding, channel coding, modulation, and MIMO precoding.
  • the model type indicates the type of AI model adopted by the first processing entity. Model types can be indicated hierarchically, including, for example, categories and subcategories.
  • major categories may include fully connected neural network (FCN), convolutional neural network (CNN), recurrent neural network (RNN) or Transformer, etc.; among them, CNN's Subcategories can include: CNN based on space utilization, depth-based CNN, multi-path CNN, CNN based on attention mechanism, etc.; subcategories of RNN can include: gated recurrent unit (GRU, gated recurrent unit), long short-term memory (LSTM, long short-term memory), etc. where model size indicates the size of the model.
  • the model size may include the number of neural network layers, neurons in each layer, and one or more parameters of weights and biases. Among them, model accuracy indicates the accuracy of model parameter usage.
  • Module precision can include: binary, integer or floating point type, etc.
  • the floating point type can also include float16, float32, etc.
  • model performance indicates the inference performance of the model.
  • the inference performance of a model can be expressed by the accuracy with which the model can infer on a reference data set.
  • Data information can be called data description.
  • Data information can include one or more of data type, data dimension, and data accuracy.
  • data type indicates the type of data the model can handle.
  • Data types can include: bits, integers, real numbers, complex numbers, etc.
  • data dimension indicates the input dimension and/or output dimension of the model.
  • Data dimensions can include: the number of elements in the input data vector, or the number of elements in the output data vector.
  • data accuracy indicates the accuracy of the data that the model can handle.
  • Data precision can include: binary, integer or floating point type, etc.
  • the floating point type can also include float16, float32, etc.
  • the second entity sends a first registration response to the first processing entity.
  • the first processing entity receives the first registration response, and the first registration response contains the input parameters of the second entity.
  • the second entity determines whether the model information and/or data information of the first processing entity meets the requirements of the second entity based on the first identification in the received first registration request. The second entity determines that the model information and/or data information of the first processing entity meets the requirements of the second entity, and then sends a first registration response to the first processing entity.
  • the second entity may feedback a registration failure indication to the first processing entity; and, optionally, the first processing The entity reports the registration failure indication to the first control entity.
  • the second entity when the second entity determines that the model information and/or data information of the first processing entity does not meet the needs of the second entity, it may not feedback the registration failure indication and implicitly indicate the registration failure.
  • the first processing entity While sending the first registration request, the first processing entity starts the first registration response timing. If the first registration response is received before the timing times out, the timing is cleared; if the timing times out, the first registration response is sent to the first control entity. Response timeout indication.
  • the input parameters of the second entity include one or more of the type, dimension, or precision of the input data of the second entity.
  • the first processing entity sends a first response confirmation according to the received first registration response.
  • the first response confirmation indicates whether the output parameters of the first processing entity match the input parameters of the second entity. .
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity.
  • the first processing entity determines whether the output parameters of the first processing entity match the input parameters of the second entity. For example, whether the type of the output data of the first processing entity matches the type of the input data of the second entity, whether the dimensions of the output data of the first processing entity match the dimensions of the input data of the second entity, and/or the Whether the accuracy of the output data of the first processing entity matches the accuracy of the input data of the second entity.
  • the first processing entity determines that the output parameter of the first processing entity matches the input parameter of the second entity and sends a first response confirmation to the second entity (S503a); if the match fails, it sends a first response confirmation to the first control entity (S503a). S503b).
  • the first processing entity sends a first response confirmation to the second entity.
  • the first response confirmation indicates that the output parameters of the first processing entity match the input parameters of the second entity successfully.
  • the second entity receives the first response confirmation. .
  • the first processing entity after the first processing entity determines that the output parameter of the first processing entity successfully matches the input parameter of the second entity, the first processing entity performs the first processing on the first input data to obtain the first output data, and The first output data is transmitted to the second entity through the first interface.
  • the first processing entity sends a first response confirmation to the first control entity.
  • the first response confirmation indicates that the output parameter of the first processing entity fails to match the input parameter of the second entity.
  • the first control entity receives the first response confirmation. A response is confirmed.
  • the first processing entity determines that the output parameter of the first processing entity fails to match the input parameter of the second entity, and may feed back a parameter matching failure indication to the first processing entity.
  • the first processing entity may not feed back a parameter matching failure indication.
  • the second entity While sending the first registration response, the second entity starts the first response confirmation timing. If the first response confirmation is received before the timing times out, the timing is cleared; if the timing times out, the first response confirmation is sent to the first control entity. Timeout indication. The second entity learns the parameter matching failure in an implicit manner, which can save signaling overhead.
  • the parameters of the model corresponding to the first processing entity are obtained through offline training, and the parameters of the model corresponding to the second entity can be adjusted according to environmental information.
  • the first processing entity does not need to retrain the model of the first processing entity. Instead, through the registration process shown in Figure 5, register to the appropriate second entity (model information and/or data information match, input and output parameters match) for data processing and transmission. This method can adapt to changes in the actual environment through lower complexity and achieve effective data transmission and reception.
  • this embodiment of the present application provides a communication method 600.
  • the method is applicable to the receiving end and involves a third control entity, a third processing entity and a second entity.
  • the third control entity is used to control or manage the third processing entity, the second entity and the interaction between them; the third processing entity and the second entity respectively correspond to the third processing module 430 in the communication architecture shown in Figure 4 and a second processing module 420.
  • the model corresponding to the third processing entity is obtained through offline training, and the model corresponding to the second entity can be adjusted according to environmental information.
  • the third control entity, the third processing entity and the second entity may be physically arranged on the same or different communication equipment, and the communication equipment may be the above-mentioned terminal equipment or access network equipment.
  • the specific content of the communication method 600 is introduced below.
  • the third control entity sends a second registration instruction to the third processing entity.
  • the third processing entity receives the second registration instruction, and the second registration instruction includes the identification information of the second entity.
  • the third control entity selects a third processing entity from one or more processing entities, selects a second entity from one or more adaptive entities, and sends a message including the second entity to the third processing entity. Registration indication of the entity's identifying information.
  • the third control entity selects the third processing entity and the second entity according to the task type.
  • This task type can correspond to one or more of: demodulation, channel decoding, and channel decoding.
  • the third control entity is a protocol layer entity above the physical layer, which may also be called a high-level entity.
  • the third control entity is a MAC layer entity, and the registration indication is MAC layer signaling.
  • the third control entity interacts with the first control entity through the interaction mechanism of the protocol layer corresponding to the third control entity (for example, MAC layer protocol) to ensure that the third control entity
  • the protocol layer corresponding to the third control entity for example, MAC layer protocol
  • the third processing entity sends a second registration request to the second entity.
  • the second entity receives the second registration request, and the second registration request includes the first identifier.
  • the third processing entity sends a second registration request to the second entity according to the second registration instruction received in S600.
  • the third processing entity selects the second entity according to its corresponding task type and sends the second registration request to it.
  • the first identification indicates one or more items of model information or data information of the third processing entity.
  • Model information can be called a model description and can include one or more of the following: model purpose, model type, model size, model accuracy, or model performance.
  • model purposes indicates the task type corresponding to the third processing entity.
  • model uses include: demodulation, channel decoding, and one or more of channel decoding.
  • Data information can be called data description.
  • Data information can include one or more of data type, data dimension, and data accuracy.
  • data dimension indicates the input dimension and/or output dimension of the model.
  • Data dimensions can include: the number of elements in the input data vector, or the number of elements in the output data vector.
  • the second entity sends a second registration response to the third processing entity.
  • the third processing entity receives the second registration response, and the second registration response contains the output parameters of the second entity.
  • the output parameter of the second entity includes one or more of the type, dimension, or precision of the output data of the second entity.
  • the second entity may feedback a registration failure indication to the third processing entity; and, optionally, the third processing The entity reports the registration failure indication to the third control entity.
  • the second entity when it determines that the model information and/or data information of the third processing entity does not meet the needs of the second entity, it may not feedback the registration failure indication and implicitly indicate the registration failure.
  • the third processing entity While sending the second registration request, the third processing entity starts the second registration response timing. If the second registration response is received before the timing times out, the timing is cleared; if the timing times out, the second registration response is sent to the third control entity. Response timeout indication.
  • the third processing entity sends a second response confirmation according to the received second registration response.
  • the second response confirmation indicates whether the output parameters of the third processing entity match the input parameters of the second entity. .
  • the input parameters of the third processing entity include one or more of the type, dimension or precision of the input data of the third processing entity.
  • the third processing entity determines whether the input parameters of the third processing entity match the output parameters of the second entity. For example, whether the type of the input data of the third processing entity matches the type of the output data of the second entity, whether the dimensions of the input data of the third processing entity match the dimensions of the output data of the second entity, and/or the Whether the accuracy of the input data of the third processing entity matches the accuracy of the output data of the second entity.
  • the third processing entity determines that the input parameter of the third processing entity successfully matches the output parameter of the second entity and sends a second response confirmation to the second entity (S603a); if the match fails, it sends a second response confirmation to the third control entity (S603a). S603b).
  • the third processing entity sends a second response confirmation to the second entity.
  • the second response confirmation indicates that the input parameters of the third processing entity and the output parameters of the second entity match successfully.
  • the second entity receives the second response confirmation. .
  • the third processing entity receives the second output data transmitted by the second entity through the second interface, and Perform second processing on the second output data to obtain third data.
  • the third processing entity receives the second output data transmitted by the second entity through the second interface, and Perform second processing on the second output data to obtain third data.
  • the third processing entity sends a second response confirmation to the third control entity.
  • the second response confirmation indicates that the input parameter of the third processing entity fails to match the output parameter of the second entity.
  • the third control entity receives the second response confirmation. Second response confirms.
  • the third processing entity determines that the input parameter of the third processing entity fails to match the output parameter of the second entity, and may feed back a parameter matching failure indication to the second entity.
  • the third processing entity determines that the input parameter of the third processing entity fails to match the output parameter of the second entity, and may not feed back a parameter matching failure indication.
  • the second entity While sending the second registration response, the second entity starts the second response confirmation timing. If the second response confirmation is received before the timing times out, the timing is cleared; if the timing times out, the second response confirmation is sent to the third control entity. Timeout indication. The second entity learns the parameter matching failure in an implicit manner, which can save signaling overhead.
  • the model corresponding to the third processing entity is obtained through offline training, and the model corresponding to the second entity can be adjusted according to environmental information.
  • the third processing entity does not need to retrain the model of the third processing entity, but registers to the appropriate (model information and/or data information matches, input and output parameters match) through the registration process shown in Figure 6 two Entities that perform data processing and transmission. This method can adapt to changes in the actual environment through lower complexity and achieve effective data transmission and reception.
  • registration request, registration response, and response confirmation may cause registration failure, such as registration response timeout, response confirmation timeout, or input and output parameters between two entities. Mismatch will cause the processing entity to fail to register with the second entity. For failure in any registration stage, you can register the process with the second entity by selecting a new first processing entity/third processing entity, or the first processing entity/third processing entity. The third processing entity initiates a new registration process implementation to the new second entity.
  • Figures 7-9 introduce several possible registration failure processes provided by the embodiments of this application.
  • this embodiment of the present application provides a communication method 700.
  • This method describes a method for handling a registration failure after the first processing entity sends a registration request to the second entity.
  • the communication method 700 is described taking the first processing entity to register with the second entity as an example, and the same applies to the third processing entity to register with the second entity.
  • the method involves a first control entity, a first processing entity, a second entity and a fourth entity, wherein the fourth entity is similar to the second entity and its model can be modulated according to environmental information.
  • the specific content of the communication method 700 is introduced below.
  • the first control entity sends a first registration instruction to the first processing entity.
  • the first processing entity receives the registration instruction, and the registration instruction includes the identification information of the second entity.
  • the first processing entity sends a first registration request to the second entity.
  • the second entity receives the first registration request, and the first registration request includes the first identifier.
  • the timing duration can be predefined.
  • step S703 is executed.
  • the first processing entity sends a first registration response timeout indication to the first control entity.
  • the first control entity receives the first registration response timeout indication.
  • the first control entity sends a third registration instruction to the first processing entity, and accordingly, the first processing entity receives the third registration instruction.
  • the third registration indication includes the identification of the fourth entity.
  • the first control entity selects another entity, that is, the fourth entity, from one or more adaptive entities, and sends a message including the identification information of the fourth entity to the first processing entity.
  • Third Registration Instructions After receiving the first registration response timeout indication, the first control entity selects another entity, that is, the fourth entity, from one or more adaptive entities, and sends a message including the identification information of the fourth entity to the first processing entity.
  • the fourth entity may also be an entity that updates the second entity according to the output parameters of the first processing entity.
  • the first processing entity sends a third registration request to the fourth entity.
  • the fourth entity receives the first registration request, and the first registration request includes the first identifier.
  • the specific content of S705 please refer to S501 and will not be repeated here.
  • the first processing entity starts the first registration response timing after sending the first registration request. If the timing expires, a timeout indication is reported to the first control entity, and the first control entity instructs another adaptive entity to the first processing entity. , avoid long waiting for the first processing entity and improve efficiency.
  • this embodiment of the present application provides a communication method 800.
  • This method describes how to handle the situation where registration failure occurs after the second entity sends a registration response to the first processing entity. For example, the subsequent registration failure occurs because the input parameters of the second entity do not match the output parameters of the first processing entity. operate.
  • the communication method 800 is described taking the first processing entity to register with the second entity as an example, and the same applies to the third processing entity to register with the second entity.
  • the method involves a first control entity, a first processing entity, a second entity and a fourth entity, where the fourth entity is similar to the second entity and its model Can be modulated based on environmental information.
  • the first control entity sends a first registration instruction to the first processing entity.
  • the first processing entity receives the registration instruction, where the registration instruction includes the identification information of the second entity.
  • the first processing entity sends a first registration request to the second entity.
  • the second entity receives the first registration request, and the first registration request includes the first identifier.
  • the second entity sends a first registration response to the first processing entity.
  • the first processing entity receives the first registration response, and the first registration response contains the input parameters of the second entity.
  • the first processing entity sends a first response confirmation to the first control entity.
  • the first response confirmation indicates that the output parameter of the first processing entity fails to match the input parameter of the second entity.
  • the first control entity receives the first response confirmation. A response is confirmed.
  • the output parameters of the first processing entity include one or more of the type, dimension, or precision of the output data of the first processing entity.
  • the first processing entity determines whether the output parameters of the first processing entity match the input parameters of the second entity. For example, whether the type of the output data of the first processing entity matches the type of the input data of the second entity, whether the dimensions of the output data of the first processing entity match the dimensions of the input data of the second entity, and/or the Whether the accuracy of the output data of the first processing entity matches the accuracy of the input data of the second entity.
  • the first processing entity determines that the output parameter of the first processing entity fails to match the input parameter of the second entity and then sends a first response confirmation to the first control entity.
  • the first control entity sends a third registration instruction to the first processing entity, and accordingly, the first processing entity receives the third registration instruction.
  • the third registration indication includes the identification of the fourth entity.
  • the first control entity selects another entity, that is, the fourth entity, from one or more adaptive entities, and sends the identification including the fourth entity to the first processing entity.
  • Third registration instructions for information After receiving the first registration response indicating matching failure, the first control entity selects another entity, that is, the fourth entity, from one or more adaptive entities, and sends the identification including the fourth entity to the first processing entity. Third registration instructions for information.
  • the first processing entity sends a third registration request to the fourth entity.
  • the fourth entity receives the first registration request, and the first registration request includes the first identifier.
  • S705 For the specific content of S705, please refer to S501 and will not be repeated here.
  • the first processing entity determines that the output parameter of the first processing entity fails to match the input parameter of the first entity indicated in the first registration response, it sends a first registration response indicating the matching failure to the first control entity,
  • the first control entity indicates to the first processing entity another adaptive entity that can be registered, thereby avoiding multiple registrations with the second entity, thereby improving efficiency.
  • this embodiment of the present application provides a communication method 900.
  • This method describes subsequent operations for registration failure due to timeout of the first registration response.
  • the communication method 900 is described taking the first processing entity to register with the second entity as an example, and the same applies to the third processing entity to register with the second entity.
  • the method involves a first control entity, a first processing entity, a second entity and a fourth processing entity, where the fourth processing entity is similar to the first processing entity, and the model corresponding to the fourth processing entity is obtained through offline training.
  • the first control entity sends a first registration instruction to the first processing entity.
  • the first processing entity receives the registration instruction, where the registration instruction includes the identification information of the second entity.
  • the first processing entity sends a first registration request to the second entity.
  • the second entity receives the first registration request, and the first registration request includes the first identifier.
  • the second entity sends a first registration response to the first processing entity.
  • the first processing entity receives the first registration response, and the first registration response contains the input parameters of the second entity.
  • the timing duration may be predefined.
  • step S703 is executed.
  • the second entity sends the first response confirmation timeout indication to the first control entity.
  • the first control entity receives the first response confirmation timeout indication.
  • the first control entity sends a fourth registration instruction to the fourth processing entity.
  • the fourth processing entity receives the fourth registration instruction, where the fourth registration instruction includes the identity of the second entity.
  • the first control entity selects a target processing entity different from the first processing entity, that is, the fourth processing entity, from one or more processing entities, and sends a message containing A fourth registration indication of the identity of the second entity.
  • the fourth processing entity sends a third registration request to the second entity.
  • the second entity receives the third registration request, where the third registration request includes the third identification.
  • the fourth processing entity sends a third registration request to the second entity according to the received third registration instruction.
  • the third registration request includes a third identifier.
  • the third identification indicates one or more items of model information or data information of the fourth processing entity. For specific content, please refer to the description related to the first identifier in S501 and will not be repeated here.
  • method 900 when the first processing entity does not feed back the first response confirmation to the second entity before the first response confirmation timing times out, the second entity reports the first response timeout confirmation indication to the first control entity, and the first control entity Select another target processing entity to register with the second entity to avoid multiple failures of the first processing entity to register with the second entity, thus improving efficiency.
  • an embodiment of the present application provides a data transmission method 1000.
  • the method is described by taking the first processing entity, the second entity and the third processing entity as an example, where the first processing entity and the second entity can be set at the sending end, and the third processing entity can be set at the receiving end.
  • the output parameters of the first processing entity match the input parameters of the second entity, and the output parameters of the second entity match the input parameters of the third processing entity.
  • the first processing entity performs the first processing on the data
  • the third processing entity performs the second processing on the data.
  • the second processing may be the reverse operation of the first processing.
  • the specific content of the communication method 1000 is introduced below.
  • the first processing entity performs the first processing on the first input data to obtain the first output data.
  • the first processing entity implements the coding and modulation function
  • the first processing includes channel coding and modulation, or equivalent processing of channel coding and modulation
  • the first input data is the source-encoded bit stream
  • the first The output data is a modulated symbol sequence.
  • the structure of the first processing entity is shown in Figure 11, including an input adaptation module, a data space conversion module, an output adaptation module and an optional parameter configuration module.
  • the first input data corresponds to input 1101 in the figure, which may be a bit stream encoded by the source.
  • the input adaptation module is used to embed and extract features of the input 1101 so that the input 1101 adapts to the input dimension of the data space transformation module.
  • the output of the input adaptation module is a bit group, and the bits in each group are The amount of data corresponds to the input dimensions of the data space transformation module.
  • the data space conversion module is used to perform subspace conversion on the output of the input adaptation module to obtain a modulated symbol sequence to implement coding and modulation functions.
  • the parameters of the data space conversion mode can be determined based on one or more of the channel coding rate, modulation order, and number of transmission resources.
  • the parameter configuration module is used to configure parameters of the data space conversion module according to one or more of channel coding rate, modulation order, and number of transmission resources.
  • the output adaptation module is used to adapt the output of the data space conversion module to a specific data type, data dimension and data accuracy to obtain the first output data, that is, output 1102.
  • the input configuration module is used to adapt the dimensions of the first input data to the input dimensions of the data space conversion module; the data space conversion module is used to perform the first processing on the first input data to obtain the first intermediate output data; and output
  • the adaptation module is used to perform dimension transformation on the first intermediate output data to obtain the first output data.
  • the dimension transformation of the first intermediate output data may include a change from a low dimension to a high dimension, for example, by copying the first intermediate output data to achieve dimensionality increase in order to cope with poor channel conditions.
  • the first processing entity implements quantization, source coding, channel coding and modulation functions, and the first processing includes quantization, source coding, channel coding and modulation, or includes quantization, source coding, channel coding and Equivalent processing of modulation, the first input data is a symbol sequence, and the first output data is a modulated symbol sequence.
  • the first input data is sensing data.
  • the structure of the first processing entity is shown in Figure 11, including an input adaptation module, a data space conversion module, an output adaptation module and an optional parameter configuration module.
  • the first input data corresponds to input 1101 in the figure, and may be a symbol sequence.
  • the input adaptation module is used to embed and extract features of the input 1101 so that the input 1101 adapts to the input dimension of the data space transformation module.
  • the output of the input adaptation module is symbol grouping, and the number of symbols in each group corresponds to the input dimension of the data space transformation module.
  • the data space conversion module is used to perform subspace conversion on the output of the input adaptation module to obtain modulated symbols to implement quantization, source coding, channel coding and modulation functions.
  • the parameters of the data space conversion mode can be determined based on one or more of the quantization level, source coding scheme, channel coding rate, modulation order and number of transmission resources.
  • the parameter configuration module is used to configure the parameters of the data space conversion module according to one or more of channel coding rate, modulation order and number of transmission resources.
  • the output adaptation module is used to adapt the output of the data space conversion module to a specific data type, data dimension and data accuracy to obtain the first output data, that is, output 1102.
  • the data space conversion module is a neural network model
  • the first processing entity is obtained through offline training.
  • the first processing entity transmits the first output data through the first interface, and correspondingly, the second entity receives the first output data through the first interface.
  • the second entity performs the second processing on the first output data to obtain the second output data.
  • the second processing includes mapping the first output data to the second output data.
  • the mapping of the first output data to the second output data is determined based on real-time environment parameters.
  • the second entity corresponds to the second processing module in Figure 4.
  • the structure of the second processing module is shown in Figure 12, including input nodes, output nodes and wireless channels.
  • the number of input nodes corresponds to the input dimension of the second entity, and the number of output nodes corresponds to the output dimension of the second entity.
  • a possible mapping between input nodes and output nodes can be expressed as:
  • x j is the j-th input value, where j ⁇ J, J represents the number of input nodes, y i is the i-th output value, where i ⁇ I, I represents the number of output nodes, and w ji is the weight between the j-th input value and the i-th output value.
  • w ji can be adjusted based on real-time environment parameters, such as using message passing algorithm (MPA), belief propagation (BP), minimum mean squared error (MMSE), and weighted minimum mean squared error. (weighted MMSE, WMMSE), maximum likelihood (maximum likelihood, ML), maximum a posteriori probability (maximum a posteriori, MAP) or other algorithms are updated.
  • MPA message passing algorithm
  • BP belief propagation
  • MMSE minimum mean squared error
  • WMMSE weighted minimum mean squared error
  • maximum likelihood maximum likelihood
  • ML maximum a posteriori probability
  • MAP maximum a posteriori probability
  • the second entity transmits the second output data to the third processing entity through the second interface, and accordingly, the third processing entity receives the second output data.
  • the third processing entity performs the third processing on the second output data to obtain the third data.
  • the first processing entity implements the coding and modulation function, and correspondingly, the third processing entity implements the demodulation and channel decoding functions.
  • the third processing can be understood as the reverse process of the first processing.
  • the input of the third processing entity, that is, the second output data is a symbol sequence; the output of the third processing entity, that is, the third data, is a bit sequence.
  • the structure of the third processing entity is shown in Figure 11, including an input adaptation module, a data space conversion module, an output adaptation module and an optional parameter configuration module.
  • the second output data corresponds to the input 1101 in the figure, and may be a modulated symbol sequence.
  • the input adaptation module is used to embed and extract features of the input 1101 so that the input 1101 adapts to the input dimension of the data space transformation module.
  • the output of the input adaptation module is symbol grouping, and the number of symbols in each group corresponds to the input dimension of the data space transformation module.
  • the data space conversion module is used to perform subspace conversion on the output of the input adaptation module to obtain a modulated symbol sequence to achieve demodulation and channel decoding functions.
  • the parameters of the data space conversion mode can be determined based on one or more of the channel coding rate, modulation order, and number of transmission resources.
  • the parameter configuration module is used to configure parameters of the data space conversion module according to one or more of channel coding rate, modulation order, and number of transmission resources.
  • the output adaptation module is used to adapt the output of the data space conversion module to subsequent modules to obtain output 1102.
  • the first processing entity implements quantization, source coding, channel coding and modulation functions
  • the third processing entity implements demodulation, channel decoding and source decoding functions.
  • the input of the third processing entity, that is, the second output data is the modulated symbol sequence; the output of the third processing entity, that is, the third data, is the symbol sequence.
  • the structure of the third processing entity is shown in Figure 11, including an input adaptation module, a data space conversion module, an output adaptation module and an optional parameter configuration module.
  • the first input data corresponds to input 1101 in the figure, and may be a modulated symbol sequence.
  • the input adaptation module is used to embed and extract features of the input 1101 so that the input 1101 adapts to the input dimension of the data space transformation module.
  • the output of the input adaptation module is symbol grouping, and the number of symbols in each group corresponds to the input dimension of the data space transformation module.
  • the data space conversion module is used to perform subspace conversion on the output of the input adaptation module to obtain modulated symbols to implement quantization, source coding, channel coding and modulation functions.
  • the parameters of the data space conversion module can be determined based on one or more of the quantization level, source coding scheme, channel coding rate, modulation order, and number of transmission resources.
  • the parameter configuration module is used to configure parameters of the data space conversion module according to one or more of channel coding rate, modulation order, and number of transmission resources.
  • the output adaptation module is used to adapt the output of the data space conversion module to subsequent modules to obtain output 1102.
  • the data space conversion module is a neural network model
  • the third processing entity is obtained through offline training.
  • the first input data is sensing data
  • the third data is sensing data recovered by the third processing entity.
  • the communication device corresponding to the third processing entity can further process the sensing time to obtain a sensing result.
  • data processing and data sending and receiving can be realized through a first processing entity, a second entity and a third processing entity that match each other.
  • the first processing entity and the third processing entity can be obtained through offline training, and the second entity can be obtained through offline training. It can be adjusted according to real-time environment parameters, adapting to the changing transmission environment with lower complexity, and achieving effective data sending and receiving.
  • the first processing entity and the third processing entity can be implemented through a neural network, and the second entity can be implemented through an adaptive algorithm or an adaptive iterative algorithm.
  • These algorithms include but are not limited to MPA, BP, MMSE, WMMSE, Maximum likelihood ML, MAP and other algorithms.
  • Figure 13 is a specific example of implementing the method provided by the embodiment of the present application through a neural network and an adaptive iterative algorithm.
  • the first processing module 410 corresponds to the first processing entity
  • the second processing module 420 corresponds to the second entity
  • the third processing module 430 corresponds to the third processing entity.
  • the first processing module 410 includes the neural network model NN Tx; the third processing module 430 includes the neural network model NN Rx.
  • NN Tx and NN Rx can be implemented through autoencoders or transformers, and are obtained through offline pre-training based on massive environmental parameter samples or training samples generated through mathematical model modeling.
  • the second processing module 420 includes an iterative algorithm module, wireless channel and optional beam forming, MIMO precoding and other modules.
  • the iterative operation module can be adjusted online based on real-time environment parameters.
  • the real-time environment parameter is determined based on a feedback indication from the third processing module, where the feedback indication includes channel state information or a channel feature vector, where the channel feature vector is a feature value vector extracted based on the channel state information.
  • the feedback indication also includes capability switch information for turning on or off the first processing module 410 and/or the second processing module 420.
  • the iterative algorithm module can be implemented through algorithms such as MPA, BP, MMSE, WMMSE, maximum likelihood ML or MAP.
  • the first processing entity, the second entity, and the third processing entity may each include a hardware structure and/or a software module.
  • Software modules are used to implement the above functions. Whether one of the above functions is performed as a hardware structure, a software module, or a hardware structure plus a software module depends on the specific application and design constraints of the technical solution.
  • an embodiment of the present application provides a communication device 1400.
  • the communication device 1400 may be a terminal or a network device, a device in a terminal device or a network device, or a device that can be used in conjunction with a terminal device or a network device.
  • the communication device 1400 may include modules or units that perform one-to-one correspondence with the methods/operations/steps/actions performed by the first processing entity, the second entity, and the third processing entity in the above method embodiment,
  • the unit may be a hardware circuit or software, or may be a combination of hardware circuit and software.
  • the communication device 1400 may include a processing unit 1410 and a transceiver unit 1420.
  • the processing unit 1410 may be configured to call the transceiver unit 1420 to perform functions of receiving and/or transmitting.
  • the communication device 1400 is configured to perform operations of the first processing entity, and the transceiver unit 1420 is configured to send a first registration request to the second entity.
  • the first registration request includes a first identifier, and the first identifier indicates One or more of the model information or data information of the communication device, wherein the model parameters of the second entity are adjusted according to the real-time environment parameters; the transceiver unit 1420 is also used to receive the first registration response sent by the second entity, the The first registration response indicates the input parameters of the second entity; the transceiver unit 1420 is also configured to send a first response confirmation, which indicates whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the communication device 1400 is configured to perform the operation of the second entity.
  • the transceiver unit 1420 is configured to receive a first registration request sent by the first processing entity.
  • the first registration request includes a first identification, a first
  • the identifier indicates one or more items of model information or data information of the first processing entity.
  • the model parameters of the communication device are based on the real-time environment. Adjust the environment parameters; the transceiver unit 1420 is used to send a first registration response to the first processing entity, the first registration response indicates the input parameters of the communication device; the input parameters of the communication device are used to determine whether the output parameters of the first processing entity Match the input parameters of the communication device.
  • the first processing entity and the second entity belong to the same communication device, that is, the communication device 1400 can implement both the operations performed by the first processing entity and the operations performed by the second entity.
  • the communication device 1400 is configured to perform the operation of the third processing entity, and the transceiver unit 1420 is configured to send a second registration request to the second entity, where the second registration request includes a second identifier indicating that the One or more of the model information or data information of the communication device, and the model parameters of the second entity are adjusted according to the real-time environment parameters; the transceiver unit 1420 is also used to receive a second registration response sent by the second entity. The response indicates the output parameters of the second entity; the transceiver unit 1420 is also configured to send a second response confirmation, which indicates whether the input parameters of the communication device match the output parameters of the second entity.
  • the communication device 1400 is used to perform operations of the second entity
  • Each functional module or unit in various embodiments of the present application may be integrated into one processor, or may exist independently, or two or more modules or units may be integrated into one module or unit.
  • the above integrated modules or units can be implemented in the form of hardware or software function modules.
  • the processing unit 1410 may be a processor
  • the transceiver unit 1420 may be a transceiver.
  • this embodiment of the present application also provides a communication device 1500 for implementing the functions of the first processing entity, the second entity, or the third processing entity in the above method.
  • the communication device may be a terminal, a network device (source access network device, target access network device or computing management function), or a device (such as a chip, circuit, etc.) in the terminal or network device, or a device that can be combined with Devices used to match terminals and network equipment.
  • the communication device 1500 includes at least one processor 1510.
  • the communication device 1500 may also include a communication interface 1520.
  • the communication interface may be a transceiver, a circuit, a bus, a module, or other types of communication interfaces, used for communicating with other devices through a transmission medium.
  • the communication interface 1520 is used for devices in the communication device 1500 to communicate with other devices.
  • the processor 1510 can perform the functions performed by the processing unit 1410 in the communication device 1400; the communication interface 1520 can be used to perform the functions performed by the transceiver unit 1420 in the communication device 1400.
  • the communication interface 1520 is used to send a first registration request to the second entity, and the first registration request includes the first identification, and the The first identifier indicates one or more of the model information or data information of the communication device, wherein the model parameters of the second entity are adjusted according to the real-time environment parameters; and is also used to receive the first registration response sent by the second entity, The first registration response indicates the input parameters of the second entity; and is also used to send a first response confirmation indicating whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the communication interface 1520 when the communication device 1500 is used to perform an operation performed by the second entity, the communication interface 1520 is used to receive a first registration request sent by the first processing entity, where the first registration request includes a first identification, and An identifier indicates one or more of the model information or data information of the first processing entity, and the model parameters of the communication device are adjusted according to real-time environment parameters; and is also used to send a first registration response to the first processing entity, The first registration response indicates the input parameters of the communication device; the input parameters of the communication device are used to determine whether the output parameters of the first processing entity match the input parameters of the communication device.
  • the communication interface 1520 when the communication device 1500 is used to perform an operation performed by the second entity, the communication interface 1520 is used to receive a second registration request sent by the third processing entity, where the second registration request includes a second identification, 2. Marking instructions One or more of the model information or data information of the third processing entity, the model parameters of the communication device are adjusted according to the real-time environment parameters; and is also used to send a second registration response to the third processing entity, the second registration response Indicates the output parameter of the communication device; the output parameter of the communication device is used to determine whether the input parameter of the third processing entity matches the output parameter of the second entity.
  • the communication interface 1520 is used to send a second registration request to the second entity, and the second registration request includes a second identification, and the second identification Indicate one or more of the model information or data information of the communication device, and the model parameters of the second entity are adjusted according to the real-time environment parameters; and are also used to receive a second registration response sent by the second entity, the second registration response Indicates the output parameters of the second entity; and is also used to send a second response confirmation indicating whether the input parameters of the communication device match the output parameters of the second entity.
  • Communication device 1500 may also include at least one memory 1530 for storing program instructions and/or data.
  • Memory 1530 and processor 1510 are coupled. Coupling in the embodiments of this application is an indirect coupling or communication connection between devices, units or modules, and may be electrical, mechanical or other forms for information interaction between devices, units or modules.
  • Processor 1510 may cooperate with memory 1530.
  • Processor 1510 may execute computer programs or instructions stored in memory 1530.
  • at least one of the at least one memory may be integrated with the processor.
  • the memory 1530 is located outside the communication device 1500 .
  • the embodiment of the present application does not limit the specific connection medium between the communication interface 1520, the processor 1510, and the memory 1530.
  • the memory 1530, the processor 1510 and the communication interface 1520 are connected through a bus 1540 in Figure 15.
  • the bus is represented by a thick line in Figure 15.
  • the connection methods between other components are only schematically explained. , is not limited.
  • the bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in Figure 15, but it does not mean that there is only one bus or one type of bus.
  • the communication device 1500 may be a chip system.
  • the chip system may be composed of chips, or may include chips and other discrete devices.
  • this embodiment of the present application also provides a communication device 1600 for implementing the functions of the first processing entity, the second entity, or the third processing entity in the above method.
  • the communication device may be a terminal, a network device, a device (such as a chip, a circuit, etc.) in the terminal or network device, or a device that can be used in conjunction with the terminal or network device.
  • the communication device includes a processor 1610, which is used to implement part or all of the functions of the first communication device, the second communication device or the terminal.
  • the processor 1610 when the communication device 1600 is used to perform an operation performed by the first processing entity, the processor 1610 is used to send a first registration request to the second entity, where the first registration request includes a first identification, and the The first identifier indicates one or more of the model information or data information of the communication device, wherein the model parameters of the second entity are adjusted according to the real-time environment parameters; and is also used to receive the first registration response sent by the second entity, The first registration response indicates the input parameters of the second entity; and is also used to send a first response confirmation indicating whether the output parameters of the first processing entity match the input parameters of the second entity.
  • the processor 1610 when the communication device 1600 is used to perform an operation performed by the second entity, the processor 1610 is used to receive a first registration request sent by the first processing entity, where the first registration request includes a first identifier, and the first registration request is sent by the first processing entity.
  • An identifier indicates one or more of the model information or data information of the first processing entity, and the model parameters of the communication device are adjusted according to real-time environment parameters; and is also used to send a first registration response to the first processing entity,
  • the first registration response indicates the input parameters of the communication device; the input parameters of the communication device are used to determine whether the output parameters of the first processing entity match the input parameters of the communication device.
  • the processor 1610 when the communication device 1600 is used to perform an operation performed by the second entity, the processor 1610 is used to receive a second registration request sent by the third processing entity, where the second registration request includes a second identification, and the second registration request is sent by the third processing entity.
  • the second identifier indicates one or more of the model information or data information of the third processing entity, and the model parameters of the communication device are adjusted according to real-time environment parameters; and is also used to send a second registration response to the third processing entity,
  • the second registration response indicates the output parameters of the communication device; the output parameters of the communication device are used to determine whether the input parameters of the third processing entity match the output parameters of the second entity.
  • the processor 1610 when the communication device 1600 is used for an operation performed by a third processing entity, the processor 1610 is configured to send a second registration request to the second entity, and the second registration request includes a second identification, and the second identification Indicate one or more of the model information or data information of the communication device, and the model parameters of the second entity are adjusted according to the real-time environment parameters; and are also used to receive a second registration response sent by the second entity, the second registration response Indicates the output parameters of the second entity; and is also used to send a second response confirmation indicating whether the input parameters of the communication device match the output parameters of the second entity.
  • the processor 1610 executes instructions stored in the memory 1620 to implement the functions implemented by the first processing entity, the second entity, or the third processing entity.
  • the communication device also includes a memory 1620.
  • the processor 1610 and the memory 1620 are integrated together.
  • the memory 1620 is external to the communication device 1600.
  • the processor 1610 may be a logic circuit, and the processor 1610 inputs/outputs messages or signaling through an input/output interface (not shown in the figure).
  • the logic circuit may be a signal processor, a chip, or other integrated circuits that can implement the method of the present application.
  • the terminal chip implements the functions of the terminal in the above method embodiment.
  • the terminal chip receives information from other modules in the terminal (such as radio frequency modules or antennas), and the information is sent to the terminal by other terminals or network equipment; or, the terminal chip sends information to other modules in the terminal (such as radio frequency modules or antennas) Output information that the terminal sends to other terminals or network devices.
  • the network device chip When the above communication device is a chip applied to a network device, the network device chip implements the functions of the network device in the above method embodiment.
  • the network device chip receives information from other modules in the network device (such as radio frequency modules or antennas), and the information is sent to the network device by the terminal or other network devices; or, the network device chip sends information to other modules in the network device (such as Such as radio frequency modules or antennas) output information, which is sent by network equipment to terminals or other network equipment.
  • the processor may be one or more central processing units (central processing units, CPU).
  • CPU central processing units
  • the processor may be a single CPU. Core CPU or multi-core CPU.
  • the processor may be a general processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or execute the disclosures in the embodiments of this application.
  • a general-purpose processor may be a microprocessor or any conventional processor, etc.
  • the steps of the methods disclosed in conjunction with the embodiments of the present application can be directly implemented by a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor.
  • the memory may include but is not limited to non-volatile memory such as hard disk drive (HDD) or solid-state drive (SSD), random access memory (Random Access Memory, RAM), erasable programmable read-only memory (Erasable Programmable ROM, EPROM), read-only memory (Read-Only Memory, ROM) or portable read-only memory (Compact Disc Read-Only Memory, CD- ROM) and so on.
  • Memory is, but is not limited to, any other medium that can be used to carry or store a computer program in the form of instructions or data structures and that can be accessed by a computer.
  • the memory in the embodiment of the present application can also be a circuit or any other device capable of realizing a storage function, used to store computer programs or instructions, and/or data.
  • Embodiments of the present application also provide a computer-readable storage medium that stores computer programs or instructions, and the computer programs or instructions are executed by a computer (for example, a processor) to implement the embodiments of the present application. Some or all of the steps of any method performed by any device.
  • Computer-readable storage media can be any available media that can be accessed by a computer or a data storage device such as a server, data center, or other integrated media that contains one or more available media.
  • the available media may be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as optical disks), or semiconductor media (such as solid state drives), etc.
  • Embodiments of the present application also provide a computer program product including a computer program or a set of instructions.
  • the computer program product is run on a computer, some or all of the steps of any method in the above aspects are executed.
  • the methods in the above embodiments may be implemented in whole or in part by software, hardware, firmware, or any combination thereof.
  • software When implemented using software, it may be implemented in whole or in part in the form of a computer program product.
  • the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable device.
  • Computer instructions may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium.
  • the chip may include a processor.
  • the chip may also include a memory (or storage module) and/or a transceiver (or communication module), or the chip may be coupled with a memory (or storage module) and/or a transceiver (or communication module), where the transceiver ( or communication module) can be used to support the chip to perform wired and/or wireless communications.
  • the memory (or storage module) can be used to store a program or a set of instructions.
  • the processor can call the program or the set of instructions to implement the above method embodiments. Operations performed by the terminal or network device in any possible implementation of the method embodiment.
  • the chip system may include the above chip, or may include the above chip and other discrete devices, such as a memory (or storage module) and/or a transceiver (or communication module).
  • this application also provides a communication system, which may include the above first processing entity, second entity and third processing entity.
  • the communication system can be used to implement the operations performed by the first processing entity, the second entity and the third processing entity in any of the above method embodiments and possible implementations of the method embodiment.
  • the communication system may have a structure as shown in Figure 2.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请提供通信方法以及相关装置。其中，一种通信方法可以用于第一通信设备的第一处理实体，该方法包括：第一处理实体向第二实体发送第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；第一处理实体接收第二实体发送的第一注册响应，第一注册响应指示所述第二实体的输入参数；以及第一处理实体发送第一响应确认，该第一响应确认指示所述第一处理实体的输出参数是否与所述第二实体的输入参数匹配。该方法能够适配环境的变化，支持有效的数据收发。

Description

一种通信方法及相关装置
本申请要求于2022年3月21日提交中国国家知识产权局，申请号为202210283863.9，发明名称为“一种通信方法及相关装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术领域，尤其涉及一种通信的方法及相关装置。
背景技术
人工智能(artifical intelligence，AI)技术将对移动通信网络技术的演进产生重要的推动作用。例如，将AI技术应用于网络层(如网络优化，移动性管理，资源分配等)或物理层(如信道编译码，信道预测、接收机等)等方面均有相关研究。
当前的AI与通信的结合，主要是针对特定场景中给定的系统参数，进行AI模型或训练方法的设计，当设备能力、场景(例如信道条件)或系统参数等环境发生改变后，需要重新设计相应的模型，并进行重新训练，难以在通用的通信系统中部署和使用。
发明内容
本申请实施例提供一种通信方法及相关装置，能够适配环境的变化，支持有效的数据收发。
第一方面，本申请实施例提供一种通信方法，该方法用于第一通信设备的第一处理实体，该方法包括：第一处理实体向第二实体发送第一注册请求，第一注册请求包括第一标识，该第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；第一处理实体接收第二实体发送的第一注册响应，该第一注册响应指示所述第二实体的输入参数；该第一处理实体发送第一响应确认，所述第一响应确认指示所述第一处理实体的输出参数是否与第二实体的输入参数匹配。
可选的，第二实体也可以称为自适应实体。
该方法中，第一处理实体向第二实体进行注册，以确定第一处理实体的输出参数与第二实体的输入参数匹配，其中，第二实体的模型参数可以根据实时环境参数进行调整，能够适配环境的变化，支持有效的数据收发。
一种可能的实现中，第一响应确认指示匹配成功，该方法还包括：第一处理实体对第一输入数据进行第一处理得到第一输出数据，第一输出数据为第一处理实体的输出；第一处理实体通过第一接口向所述第二实体传输所述第一输出数据。
该种实现中，当第一处理实体的输出参数与第二实体的输入参数匹配成功时，第一处理实体向第二实体传输第一输出数据，实现有效的数据传输。
一种可能的实现中，第一处理实体包括输入适配模块，数据空间转换模块和输出适配模块；
该种实现中，第一处理实体的输入适配模块和输出适配模块分别对第一处理实体的输入数据和输出数据进行维度变换等处理，以适配第一处理实体的前序处理模块的输出维度与数据空间转换模的输入维度，以及适配第一处理实体的输出维度与第一处理实体的后续处理模块(如第二实体)的输出维度。
一种可能的实现中，所述第一处理包括量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项，或者与量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项等效的处理。
通过该种实现，可以通过第一处理实体实现一项或多项传统的数据收发中进行的操作，实现多项操作时，相比于传统链路中，单独实现各个操作模块，更容易实现全局优化。
一种可能的实现中，第一处理包括：编码和调制，所述第一输入数据为比特流，第一输出数据为调制后的符号。
一种可能的实现中，第一处理包括量化、信源编码、信道编码以及调制，第一输入数据为符号序列，第一输出数据为调制后的符号序列。
由上述可能的实现可知，实现不同功能的第一处理实体均可以注册到与其匹配的第二实体中，以实现有效的数据收发。
一种可能的实现中，第一接口定义了第一处理实体与第二实体之间的交互流程和/或信令。示例性的，第一接口定义了传输所述注册请求、所述注册响应或所述响应确认中的一项或多项的过程。
一种可能的实现中，第一响应确认指示第一处理实体的输出参数与第二实体的输入参数匹配成功，该方法还包括：第一处理实体向第四实体发送第二注册请求，第二注册请求包括第一标识，第一标识指示第一处理实体的模型信息或数据信息中的一项或多项，第四实体的模型参数可以根据实时环境参数调整；第一处理实体接收第四实体发送的第二注册响应，第二注册响应指示第四实体的输入参数；第一处理实体发送第二响应确认，第二响应确认指示第一处理实体的输出参数是否与第四实体的输入参数匹配。
该实现中，当第一处理实体的输出参数与第二实体的输入参数匹配失败时，第一处理实体可以向第四实体发起注册，确保第一处理实体注册到与其匹配的自适应实体。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
该实现可以使得第二实体根据第一标识对应的模型信息，判断第一处理实体是否与第二实体的对应的模型信息相符，从而确定第一注册响应，进而使得第一处理实体注册到合适的第二实体上。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
该实现可以使得第二实体根据第一标识对应的数据信息，判断第一处理实体是否与第二实体的对应的数据信息相符，从而确定第一注册响应，进而使得第一处理实体注册到合适的第二实体上。
一种可能的实现中，第一处理实体的输出参数包括所述第一处理实体的输出数据的类型、维度或精度中的一项或多项，所述第二实体的输入参数包括所述第二实体的输入数据的类型、维度或精度中的一项或多项。
一种可能的实现中，所述第一处理实体和第二实体属于同一个通信设备。
第二方面，本申请实施例提供一种通信方法，该方法用于第二通信设备的第二实体，该方法包括：第二实体接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；第二实体向第一处理实体发送第一注册响应，第一注册响应指示第二实体的输入参数；第二实体的输入参数用于确定所述第一处理实体的输出参数是否与所述第二实体的输入参数匹配。
该方法中，第二实体接收第一处理实体的注册请求，并向第一处理实体反馈第二实体的输入参数，使得第一处理实体判断第一处理实体的输出参数与第二实体的输入参数匹配，其中，第二实体的模型参数可以根据实时环境参数进行调整，能够适配环境的变化，支持有效的数据收发。
一种可能的实现中，第二实体接收第一响应确认，该第一响应确认指示所述第一处理实体的输出参数与所述第二实体的输入参数匹配成功，该方法还包括：第二实体通过第一接口接收第一输出数据，第一输出数据为第一处理实体的输出数据；第二实体对该第一输出数据进行第二处理得到第二输出数据；第二实体输出该第二输出数据。
一种可能的实现中，第二实体输出第二输出数据，包括：第二实体向第三通信设备发送第二输出数据。
一种可能的实现中，第二实体对所述第一输出数据进行第二处理得到第二输出数据，包括：第二实体将第一输出数据映射至第二输出数据。其中，映射方式可以根据实时的环境参数调整。
该种实现中，通过根据实时的环境参数调整第一输出数据到第二输出数据的映射方式，可以适配变化的传输环境，支持有效的数据收发。
一种可能的实现中，第一接口定义了第一处理实体与第二实体之间的交互流程和/或信令。示例性的，第一接口定义了传输注册请求、注册响应或响应确认中的一项或多项的过程。
一种可能的实现中，第二实体发送所述第一注册响应的同时，启动第一响应确认计时，当所述响应确认计时到时，则向第一控制实体发送第一响应确认超时指示。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
一种可能的实现中，第一处理实体的输出参数包括所述第一处理实体的输出数据的类型、维度或精度中的一项或多项，所述第二实体的输入参数包括所述第二实体的输入数据的类型、维度或精度中的一项或多项。
一种可能的实现中，第一处理实体和第二实体属于同一个通信设备。
第二方面各种可能的实现的有益效果参见第一方面各种可能的实现的有益效果，此处不再赘述。
第三方面，本申请实施例提供一种通信方法，该方法用于第三通信设备的第三处理实体，该方法包括：第三处理实体向第二实体发送第二注册请求，第二注册请求包括第二标识，该第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；第三处理实体接收第二实体发送的第二注册响应，该第二注册响应指示所述第二实体的输出参数；该第三处理实体发送第二响应确认，所述第二响应确 认指示第三处理实体的输入参数是否与第二实体的输出参数匹配。
可选的，第二实体也可以称为自适应实体。
该方法中，第三处理实体向第二实体进行注册，以确定第三处理实体的输入参数与第二实体的输出参数匹配，其中，第二实体的模型参数可以根据实时环境参数进行调整，能够适配环境的变化，支持有效的数据收发。
一种可能的实现中，第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配，该方法还包括：第三处理实体获取第二实体传输的第二输出数据，并对该第二输出数据进行第三处理得到第三数据。
该种实现中，当第三处理实体的输入参数与第二实体的输出参数匹配成功后，第三处理实体接收第二实体传输的第二输出数据，并对第二输出数据进行第三处理得到第三数据，实现有效的数据传输。
一种可能的实现中，第三处理实体包括输入适配模块，数据空间转换模块和输出适配模块；输入配置模块用于适配第二输出数据的维度与数据空间转换模块的输入维度；数据空间转换模块用于对第二输出数据进行第三处理，得到第二中间输出数据；输出适配模块用于对第二中间输出数据进行维度变换得到第三数据。
该种实现中，第三处理实体的输入适配模块和输出适配模块分别对第三处理实体的输入数据和输出数据进行维度变换等处理，以适配第三处理实体的前序处理模块(如第二实体)的输出维度与数据空间转换模的输入维度，以及适配第三处理实体的输出维度与第三处理实体的后续处理模块的输出维度。
一种可能的实现中，第三处理为第一方面中的第一处理的逆操作，第三处理包括MIMO解码、解调、信道译码、信源解码中的一项或多项，或者与MIMO解码、解调、信道译码、信源解码中的一项或多项等效的处理。
通过该种实现，可以通过第三处理实体实现一项或多项传统的数据收发中进行的操作，实现多项操作时，相比于传统链路中，单独实现各个操作模块，更容易实现全局优化。
一种可能的实现中，第三处理包括：解调和信道译码，所述第二输出数据为调制后的符号，第三数据为比特流。
一种可能的实现中，第三处理包括解调、信道译码、信源解码，第二输出数据为调制后的符号序列，第三数据为符号序列。
由上述可能的实现可知，实现不同功能的第三处理实体均可以注册到与其匹配的第二实体中，以实现有效的数据收发。
一种可能的实现中，第二接口定义了第三处理实体与第二实体之间的交互流程和/或信令。示例性的，第二接口定义了传输所述注册请求、所述注册响应或所述响应确认中的一项或多项的过程。
一种可能的实现中，第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配失败，该方法还包括：第三处理实体向第四实体发送第二注册请求，第二注册请求包括第二标识，第二标识指示第三处理实体的模型信息或数据信息中的一项或多项，第四实体的模型参数可以根据实时环境参数调整；第三处理实体接收第四实体发送的第二注册响应，第二注册响应指示第四实体的输入参数；第三处理实体发送第二响应确认，第二响应确认指示第三处理实体的输入参数是否与第四实体的输入参数匹配。
该实现中，当第三处理实体的输入参数与第二实体的输出参数匹配失败时，第三处理实 体可以向第四实体发起注册，确保第三处理实体注册到与其匹配的自适应实体。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
该实现可以使得第二实体根据第二标识对应的模型信息，判断第三处理实体是否与第二实体的对应的模型信息相符，从而确定第二注册响应，进而使得第三处理实体注册到合适的第二实体上。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
该实现可以使得第二实体根据第二标识对应的数据信息，判断第三处理实体是否与第二实体的对应的数据信息相符，从而确定第二注册响应，进而使得第三处理实体注册到合适的第二实体上。
一种可能的实现中，第三处理实体的输入参数包括所述第三处理实体的输入数据的类型、维度或精度中的一项或多项，所述第二实体的输出参数包括所述第二实体的输出数据的类型、维度或精度中的一项或多项。
第四方面，本申请实施例提供一种通信方法，该方法用于第四通信设备的第二实体，该方法包括：第二实体接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；第二实体向第三处理实体发送第二注册响应，第二注册响应指示第二实体的输出参数；第二实体的输出参数用于确定第三处理实体的输入参数是否与第二实体的输出参数匹配。
该方法中，第二实体接收第三处理实体的注册请求，并向第三处理实体反馈第二实体的输出参数，使得第三处理实体判断第三处理实体的输入参数与第二实体的输出参数匹配，其中，第二实体的模型参数可以根据实时环境参数进行调整，能够适配环境的变化，支持有效的数据收发。
一种可能的实现中，第二实体接收第二响应确认，该第二响应确认指示所述第三处理实体的输入参数与所述第二实体的输出参数匹配成功，该方法还包括：第二实体对第一输出数据进行第二处理得到第二输出数据，第一输出数据为第一处理实体的输出数据；第二实体通过第二接口向第三处理实体传输该第二输出数据。
一种可能的实现中，第二实体对所述第一输出数据进行第二处理得到第二输出数据，包括：第二实体将第一输出数据映射至第二输出数据。其中，映射方式可以根据实时的环境参数调整。
该种实现中，通过根据实时的环境参数调整第一输出数据到第二输出数据的映射方式，可以适配变化的传输环境，支持有效的数据收发。
一种可能的实现中，第二接口定义了第三处理实体与第二实体之间的交互流程和/或信令。示例性的，第二接口定义了传输注册请求、注册响应或响应确认中的一项或多项的过程。
一种可能的实现中，第二实体发送所述第二注册响应的同时，启动第二响应确认计时，当所述响应确认计时到时，则向第三控制实体发送第二响应确认超时指示。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处 理数据精度。
一种可能的实现中，第三处理实体的输入参数包括所述第三处理实体的输入数据的类型、维度或精度中的一项或多项，所述第二实体的输出参数包括所述第二实体的输出数据的类型、维度或精度中的一项或多项。
第三方面和第四方面的各种可能的实现的有益效果参见第一方面各种可能的实现的有益效果，此处不再赘述。
第五方面，本申请实施例还提供一种通信装置，该通信装置可以用于第一方面的第一处理实体，该通信装置可以是终端或网络设备，也可以是终端或网络设备中的装置(例如，芯片，或者芯片系统，或者电路)，或者是能够和终端或网络设备匹配使用的装置。
一种可能的实现中，该通信装置可以包括执行第一方面中所描述的方法/操作/步骤/动作所一一对应的模块或单元，该模块或单元可以是硬件电路，也可是软件，也可以是硬件电路结合软件实现。
一种可能的实现中，该通信装置可以包括处理单元和收发单元。处理单元用于调用收发单元执行接收和/或发送的功能。具体地，收发单元用于向第二实体发送第一注册请求，第一注册请求包括第一标识，该第一标识指示该通信装置的模型信息或数据信息中的一项或多项，其中，第二实体的模型参数根据实时环境参数进行调整；收发单元还用于接收第二实体发送的第一注册响应，该第一注册响应指示第二实体的输入参数；收发单元还用于发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
一种可能的实现中，处理单元还用于对第一输入数据进行第一处理得到第一输出数据；收发单元还用于通过第一接口向第二实体传输该第一输出数据。
一种可能的实现中，该收发单元包括输入适配模块和输出适配模块，该处理单元包括数据空间转换模块，输入配置模块用于适配第一输入数据的维度与数据空间转换模块的输入维度；数据空间转换模块用于对第一输入数据进行第一处理，得到第一中间输出数据；输出适配模块用于对第一中间输出数据进行维度变换得到第一输出数据。
一种可能的实现中，第一处理包括量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项，或者与量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项等效的处理。
一种可能的实现中，第一响应确认指示该通信装置的输出参数与第二实体的输入参数匹配成功，收发单元还用于向第四实体发送第二注册请求，第二注册请求包括第一标识，第一标识指示第一处理实体的模型信息或数据信息中的一项或多项，第四实体的模型参数可以根据实时环境参数调整；收发单元还用于接收第四实体发送的第二注册响应，第二注册响应指示第四实体的输入参数；收发单元还用于发送第二响应确认，第二响应确认指示第一处理实体的输出参数是否与第四实体的输入参数匹配。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
一种可能的实现中，第一处理实体的输出参数包括所述第一处理实体的输出数据的类型、维度或精度中的一项或多项，所述第二实体的输入参数包括所述第二实体的输入数据的类型、维度或精度中的一项或多项。
一种可能的实现中，该通信装置和第二实体属于同一个通信设备。
一种可能的实现中，上述处理单元为处理器，上述收发单元为收发器。
第六方面，本申请实施例还提供一种通信装置，该通信装置可以用于第二方面的第二实体，该通信装置可以是终端或网络设备，也可以是网络设备中的装置(例如，芯片，或者芯片系统，或者电路)，或者是能够和网络设备匹配使用的装置。
一种可能的实现中，该通信装置可以包括执行第二方面中所描述的方法/操作/步骤/动作所一一对应的模块或单元，该模块或单元可以是硬件电路，也可是软件，也可以是硬件电路结合软件实现。
一种可能的实现中，该通信装置可以包括处理单元和收发单元。处理单元用于调用收发单元执行接收和/或发送的功能。具体地，收发单元用于接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；收发单元用于向第一处理实体发送第一注册响应，第一注册响应指示该通信装置的输入参数；该通信装置的输入参数用于确定第一处理实体的输出参数是否与该通信装置的输入参数匹配。
一种可能的实现中，收发单元用于接收第一响应确认，该第一响应确认指示所述第一处理实体的输出参数与该通信装置的输入参数匹配成功，收发单元还用于通过第一接口接收第一输出数据，第一输出数据为第一处理实体的输出数据；处理单元用于对该第一输出数据进行第二处理得到第二输出数据；收发单元还用于输出该第二输出数据。
一种可能的实现中，收发单元具体用于向第三通信设备发送第二输出数据。
一种可能的实现中，处理单元具体用于将第一输出数据映射至第二输出数据。其中，映射方式可以根据实时的环境参数调整。
一种可能的实现中，处理单元还用于在收发单元发送第一注册响应的同时，启动第一响应确认计时；收发单元还用于当所述响应确认计时到时，向第一控制实体发送第一响应确认超时指示。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
一种可能的实现中，第一处理实体的输出参数包括所述第一处理实体的输出数据的类型、维度或精度中的一项或多项，所述第二实体的输入参数包括所述第二实体的输入数据的类型、维度或精度中的一项或多项。
一种可能的实现中，该通信装置和第一处理实体属于同一个通信设备。
第七方面，本申请实施例还提供一种通信装置，该通信装置可以用于第三方面的第三处理实体，该通信装置可以是终端或网络设备，也可以是终端或网络设备中的装置(例如，芯片，或者芯片系统，或者电路)，或者是能够和终端或网络设备匹配使用的装置。
一种可能的实现中，该通信装置可以包括执行第三方面中所描述的方法/操作/步骤/动作所一一对应的模块或单元，该模块或单元可以是硬件电路，也可是软件，也可以是硬件电路结合软件实现。
一种可能的实现中，该通信装置可以包括处理单元和收发单元。处理单元用于调用收发单元执行接收和/或发送的功能。具体地，收发单元用于向第二实体发送第二注册请求，第二 注册请求包括第二标识，该第二标识指示该通信装置的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；收发单元还用于接收第二实体发送的第二注册响应，该第二注册响应指示第二实体的输出参数；收发单元还用于发送第二响应确认，所述第二响应确认指示该通信装置的输入参数是否与第二实体的输出参数匹配。
一种可能的实现中，第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配，该收发单元还用于获取第二实体传输的第二输出数据，该处理单元还用于对该第二输出数据进行第三处理得到第三数据。
一种可能的实现中，收发单元包括输入适配模块和输出适配模块，处理单元包括数据空间转换模块；输入配置模块用于适配第二输出数据的维度与数据空间转换模块的输入维度；数据空间转换模块用于对第二输出数据进行第三处理，得到第二中间输出数据；输出适配模块用于对第二中间输出数据进行维度变换得到第三数据。
一种可能的实现中，第三处理为第一方面中的第一处理的逆操作，第三处理包括MIMO解码、解调、信道译码、信源解码中的一项或多项，或者与MIMO解码、解调、信道译码、信源解码中的一项或多项等效的处理。
一种可能的实现中，第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配失败，收发单元还用于向第四实体发送第二注册请求，第二注册请求包括第二标识，第二标识指示第三处理实体的模型信息或数据信息中的一项或多项，第四实体的模型参数可以根据实时环境参数调整；收发单元还用于接收第四实体发送的第二注册响应，第二注册响应指示第四实体的输入参数；收发单元还用于发送第二响应确认，第二响应确认指示第三处理实体的输入参数是否与第四实体的输入参数匹配。
第八方面，本申请实施例还提供一种通信装置，该通信装置可以用于第四方面的第二实体，该通信装置可以是终端或网络设备，也可以是终端或网络设备中的装置(例如，芯片，或者芯片系统，或者电路)，或者是能够和终端或网络设备匹配使用的装置。
一种可能的实现中，该通信装置可以包括执行第四方面中所描述的方法/操作/步骤/动作所一一对应的模块或单元，该模块或单元可以是硬件电路，也可是软件，也可以是硬件电路结合软件实现。
一种可能的实现中，该通信装置可以包括处理单元和收发单元。处理单元用于调用收发单元执行接收和/或发送的功能。具体地，收发单元用于接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；收发单元用于向第三处理实体发送第二注册响应，第二注册响应指示该通信装置的输出参数；该通信装置的输出参数用于确定第三处理实体的输入参数是否与所述第二实体的输出参数匹配。
一种可能的实现中，收发单元用于接收第二响应确认，该第二响应确认指示第三处理实体的输入参数与该通信装置的输出参数匹配成功。处理单元用于对第一输出数据进行第二处理得到第二输出数据，该第一输出数据为第一处理实体的输出数据；收发单元还用于输出该第二输出数据。
对该第一输出数据进行第二处理得到第二输出数据；收发单元还用于输出该第二输出数据。
一种可能的实现中，处理单元具体用于将第一输出数据映射至第二输出数据。其中，映射方式可以根据实时的环境参数调整。
一种可能的实现中，处理单元还用于在收发单元发送第二注册响应的同时，启动第二响应确认计时。收发单元还用于当所述响应确认计时到时，向第三控制实体发送第二响应确认超时指示。
一种可能的实现中，模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
一种可能的实现中，数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
一种可能的实现中，第三处理实体的输入参数包括所述第三处理实体的输入数据的类型、维度或精度中的一项或多项，所述第二实体的输出参数包括所述第二实体的输出数据的类型、维度或精度中的一项或多项。
第九方面，本申请实施例还提供一种通信装置，包括处理器，用于实现第一方面的方法以及其各种可能的实现。一种可能的实现中，处理器通过逻辑电路实现上述方法；又一种可能的实现中，处理器通过执行指令以实现上述方法。
具体地，处理器用于输出第一注册请求，该第一注册请求包括第一标识，该第一标识指示该通信装置的模型信息或数据信息中的一项或多项，其中，第二实体的模型参数根据实时环境参数进行调整；处理器还用于接收第二实体发送的第一注册响应，该第一注册响应指示第二实体的输入参数；处理器还用于发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
一种可能的实现中，处理单元还用于对第一输入数据进行第一处理得到第一输出数据，以及通过第一接口向第二实体传输该第一输出数据。
第十方面，本申请实施例还提供一种通信装置，包括处理器，用于实现第二方面的方法以及其各种可能的实现。一种可能的实现中，处理器通过逻辑电路实现上述方法；又一种可能的实现中，处理器通过执行指令以实现上述方法。
具体地，处理器用于接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；处理器输出第一注册响应，第一注册响应指示该通信装置的输入参数；该通信装置的输入参数用于确定第一处理实体的输出参数是否与该通信装置的输入参数匹配。
第十一方面，本申请实施例还提供一种通信装置，包括处理器，用于实现第三方面的方法以及其各种可能的实现。一种可能的实现中，处理器通过逻辑电路实现上述方法；又一种可能的实现中，处理器通过执行指令以实现上述方法。可选的，该通信装置还包括用于存储该指令的存储器。
具体地，处理器用于输出第二注册请求，第二注册请求包括第二标识，该第二标识指示该通信装置的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；处理器还用于接收第二实体发送的第二注册响应，该第二注册响应指示第二实体的输出参数；处理器还用于输出第二响应确认，所述第二响应确认指示该通信装置的输入参数是否与第二实体的输出参数匹配。
第十二方面，本申请实施例还提供一种通信装置，包括处理器，用于实现第四方面的方法以及其各种可能的实现。一种可能的实现中，处理器通过逻辑电路实现上述方法；又一种可能的实现中，处理器通过执行指令以实现上述方法。可选的，该通信装置还包括用于存储 该指令的存储器。
具体地，处理器用于接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；处理器还用于输出第二注册响应，第二注册响应指示该通信装置的输出参数；该通信装置的输出参数用于确定所述第三处理实体的输入参数是否与该通信装置的输出参数匹配。
第十三方面，本申请实施例还提供一种通信装置，包括处理器，用于执行存储器中存储的计算机程序(或计算机可执行指令)，当计算机程序(或计算机可执行指令)被执行时，使得该装置执行如第一方面及第一方面各个可能的实现中的方法，或者使得该装置执行如第二方面及第二方面各个可能的实现中的方法，或者使得该装置执行如第三方面及第三方面各个可能的实现中的方法，或者使得该装置执行如第四方面及第四方面各个可能的实现中的方法。
在一种可能的实现中，处理器和存储器集成在一起；
在另一种可能的实现中，上述存储器位于该通信装置之外。
该通信装置还包括通信接口，该通信接口用于该通信装置与其他设备进行通信，例如数据和/或信号的发送或接收。示例性的，通信接口可以是收发器、电路、总线、模块或其它类型的通信接口。
第十四方面，本申请实施例还提供一种通信装置，用于执行上述第一方面及其各种可能的实现中的方法。
第十五方面，本申请实施例还提供一种通信装置，用于执行上述第二方面及其各种可能的实现中的方法，和/或，第四方面及其各种可能的实现中的方法。
第十六方面，本申请实施例还提供一种通信装置，用于执行上述第三方面及其各种可能的实现中的方法。
上述第五至第十六方面的通信装置以及其各种可能的实现带来的有益效果参见第一方面及其各种可能的实现的方法的有益效果，此处不再赘述。
第十七方面，本申请实施例还提供一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序(或计算机可执行指令)，其中，该计算机程序(或计算机可执行指令)被处理器执行时，使得上述第一方面及其任一种可能的实现、第二方面及其任一种可能的实现、第三方面及其任一种可能的实现、或第四方面及其任一种可能的实现中所述的方法的部分或全部步骤被执行。
第十八方面，本申请实施例还提供了一种包括计算机可执行指令的计算机程序产品，当该计算机程序产品被运行时，使得上述第一方面及其任一种可能的实现、第二方面及其任一种可能的实现、第三方面及其任一种可能的实现、或第四方面及其任一种可能的实现中所述的方法的部分或全部步骤被执行。
第十九方面，本申请实施例还提供了一种包括计算机可执行指令的计算机程序，当该计算机程序被运行时，使得上述第一方面及其任一种可能的实现、第二方面及其任一种可能的实现、第三方面及其任一种可能的实现、或第四方面及其任一种可能的实现中所述的方法的部分或全部步骤被执行。
第二十方面，本申请实施例还提供一种芯片系统，该芯片系统包括处理器，还可以包括存储器，用于实现上述第一方面及其任一种可能的实现、第二方面及其任一种可能的实现、 第三方面及其任一种可能的实现或者第四方面及其任一种可能的实现中所述的方法。该芯片系统可以由芯片构成，也可以包含芯片和其他分立器件。
第二十一方面，本申请实施例还提供一种通信系统，包括第五方面及其各种可能的实现提供的通信装置、第六或八方面及其各种可能的实现提供的通信装置、第七方面及其各种可能的实现提供的通信装置。
附图说明
下面将对本申请实施例涉及的一些附图进行说明。
图1是一种自编码器的结构示意图。
图2是本申请实施例适用的通信系统的架构示意图。
图3是一种人工智能通信收发机的示意图。
图4是本申请实施例提供的通信架构示意图。
图5是本申请实施例提供的通信方法500的交互示意图。
图6是本申请实施例提供的通信方法600的交互示意图。
图7是本申请实施例提供的通信方法700的交互示意图。
图8是本申请实施例提供的通信方法800的交互示意图。
图9是本申请实施例提供的通信方法900的交互示意图。
图10是本申请实施例提供的通信方法900的交互示意图。
图11是本申请实施例提供的处理实体的结构示意图。
图12是本申请实施例提供的自适应实体的结构示意图。
图13通过神经网络和自适应迭代算法实现本申请实施例提供的方法的示例。
图14为本申请实施例提供的通信装置的结构示意图。
图15为本申请实施例提供的通信装置的结构示意图。
图16为本申请实施例提供的通信装置的结构示意图。
具体实施方式
本申请实施例提供一种通信方法及相关装置，能够适配环境的变化，支持有效的数据收发。
下面结合本申请实施例中的附图对本申请实施例进行描述。
首先对本申请实施例涉及的术语进行解释：
本申请实施例涉及的自编码器(AutoEncoder)为一种AI技术。图1为自编码器的一种结构示意图。如图1所示，通过f函数(编码器(Encoder))将输入数据x映射/压缩到隐空间(latent space)的变量z，再通过g函数(解码器(Decoder))从隐空间变量z恢复出数据 其中f函数和g函数可以用神经网络实现，则优化目标可以是表示寻找f函数和g函数的参数，使得恢复数据的误差最小。一般来说，通信系统完成的任务与自编码器很相似，可以将通信系统类比于一种分布式的自编码器(distributed AutoEncoder，DAE)。对于无线通信系统，可以认为发射机将隐空间变量(波形)通过信道发送，并在接收机进行信息恢复。计算机领域的自编码器假设编码器和解码器之间的前向与反向传播为理想传播，而在无需通信系统中，信道是可变的，因此在训练和推 理过程中需要考虑信道传输带来的影响。
本申请中“/”表示“或”。术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，B和/或C，可以表示：单独存在B，同时存在B和C，单独存在C这三种情况。本申请实施例的说明书和权利要求书中的术语“第一”和“第二”等是用于区别不同的对象，没有明确说明的情况下不用于描述对象的特定顺序。例如，第一通信装置和第二通信装置等是用于区别不同的通信装置，而不是用于描述目标对象的特定顺序。在本申请实施例中，“示例性的”或者“例如”等词用于表示作例子、例证或说明，本申请实施例中被描述为“示例性的”或者“例如”的任何实施例或方案不应被解释为比其它实施例或设计方案更优选或更具优势。在本申请实施例的描述中，除非另有说明，“多个”的含义是指两个或两个以上。例如，多个处理单元是指两个或两个以上的处理单元。
本申请的技术方案可以应用于第三代合作伙伴计划(3rd generation partnership project，3GPP)相关的蜂窝系统，例如，长期演进(long term evolution，LTE)系统等第四代(4th generation，4G)通信系统，新无线(new radio，NR)系统等第五代(5th generation，5G)通信系统，支持多种无线技术融合的通信系统，或者是5G之后的演进的通信系统。
图1为适用于本申请实施例的通信系统的示例。参见图1，通信系统100包括至少一个终端设备110以及至少一个接入网设备120。
本申请实施例中提及的终端设备110，可以是一种具有无线收发功能的设备，可以和接入网设备120或其他终端设备进行通信。终端设备110具体可以指用户设备(user equipment，UE)、接入终端、用户单元(subscriber unit)、用户站、移动台(mobile station)、远方站、远程终端、移动设备、用户终端、无线通信设备、用户代理或用户装置。终端设备还可以是卫星电话、蜂窝电话、智能手机、无线数据卡、无线调制解调器、机器类型通信设备、可以是无绳电话、会话启动协议(session initiation protocol，SIP)电话、无线本地环路(wireless local loop，WLL)站、个人数字处理(personal digital assistant，PDA)、具有无线通信功能的手持设备、计算设备或连接到无线调制解调器的其它处理设备、车载设备、高空飞机上搭载的通信设备、可穿戴设备、无人机、机器人、设备到设备通信(device-to-device，D2D)中的终端、车辆外联(vehicle to everything，V2X)中的终端、虚拟现实(virtual reality，VR)终端设备、增强现实(augmented reality，AR)终端设备、工业控制(industrial control)中的无线终端、无人驾驶(self driving)中的无线终端、远程医疗(remote medical)中的无线终端、智能电网(smart grid)中的无线终端、运输安全(transportation safety)中的无线终端、智慧城市(smart city)中的无线终端、智慧家庭(smart home)中的无线终端或者5G之后演进的通信网络中的终端设备等，本申请不作限制。
接入网设备120是具有无线收发功能的设备，用于与终端设备110进行通信。接入网设备可以为无线接入网(radio access network，RAN)中的节点，可以称为RAN节点或网络设备，又可以称为基站。可以是LTE中的演进型基站(evolved Node B，eNB或eNodeB)；或者可以是gNodeB(gNB)等5G网络中的基站或者5G之后演进的公共陆地移动网络(public land mobile network，PLMN)中的基站，或非第三代合作伙伴项目(3rd generation partnership project，3GPP)接入设备等。可选的，本申请实施例中的网络设备可以包括各种形式的基站，例如：宏基站、微基站(也称为小站)、中继站、传输点(transmitting and receiving point，TRP)、发射点(transmitting point，TP)、移动交换中心以及设备到设备(Device-to-Device，D2D)、车辆外联(vehicle-to-everything，V2X)、机器到机器(machine-to-machine，M2M)通信中 承担基站功能的设备等，还可以包括云接入网(cloud radio access network，C-RAN)系统中的集中式单元(centralized unit，CU)和分布式单元(distributed unit，DU)、非陆地通信网络(non-terrestrial network，NTN)通信系统中的网络设备，本申请实施例对此不作具体限定。
图2所示的通信系统还可以包括核心网设备(图中未示出)，接入网设备120可以和核心网设备进行交互，向终端提供通信服务。例如核心网设备可以为终端提供通信连接、认证、管理、策略控制以及对数据业务完成承载等。
本申请实施例中，用于实现终端设备的功能的装置可以是终端设备；也可以是能够支持终端设备实现该功能的装置，例如芯片系统。该装置可以被安装在终端设备中或者和终端设备匹配使用。本申请实施例中，芯片系统可以由芯片构成，也可以包括芯片和其他分立器件。
本申请实施例中，用于实现接入网设备的功能的装置可以是接入网设备；也可以是能够支持接入网设备实现该功能的装置，例如芯片系统。该装置可以被安装在接入网设备中或者和接入网设备匹配使用。
本申请实施例中，本申请实施例中，终端设备还可以具备AI处理能力，接入网设备也可以具备AI处理能力，适用于AI与通信结合的场景。一种可能的AI与通信结合的场景为AI通信收发机，如图3所示，发射机和接收机的一个或多个模块由一个AI模型(如神经网络)实现，从而获得优异的全局性能。
当前的AI与通信的结合，主要是针对特定场景中给定的系统参数，进行AI模型或训练方法的设计，当设备能力、场景(例如信道条件)或系统参数等环境发生改变后，需要重新设计相应的模型，并进行重新训练，难以在通用的通信系统中部署和使用。
为解决上述问题本申请实施例提供一种通信方法、相关装置以及系统，能够适配信道条件、用户需求等环境的变化，实现有效的数据收发。
图4为本申请实施例提供的一种通信系统架构图。如图4所示，该通信系统架构包括第一处理模块410，第二处理模块420，第三处理模块430，第一接口440以及第二接口450。第一处理模块410获取并处理数据1得到数据2，并通过第一接口440将数据2传输至第二处理模块420；第二处理模块420获取并处理数据2得到数据3，并将数据3通过第二接口450传输至第三处理模块430，第三处理模块430获取并处理数据2得到数据4。
第一处理模块410可以设置在发送端，数据1可以是比特流或符号，示例性的，第一处理模块410可以对数据1进行信源编码、信道编码、调制、多输入多输出(multiple-input multiple-output，MIMO)预编码等操作的一项或多项，或者进行与上述一项或多项操作的等效操作；第三处理模块430可以设置在接收端，示例性的，第三处理模块430可以对获取的数据3进行解调、信道译码、信源译码等操作的一些或多项，或者进行与上述一项或多项操作的等效操作，得到数据4，其中数据4可以是对数据1进行恢复得到的比特流或符号。
第一处理模块410和/或第三处理模块430可以通过神经网络模型实现。可选的，第一处理模块410和/或第三处理模块430可以采用海量环境数据(如信道数据)通过离线预训练进行配置，实际使用或推理时可以不再进行训练。
第一接口440定义了第一处理模块410和第二处理模块420之间的交互流程和/或信令。第二接口450定义了第二处理模块420和第三处理模块430之间的交互流程和/或信令。第一接口440和/或第二接口450可以是逻辑接口，也可以是物理接口。
第二处理模块420也可以称为自适应(selfadaptive)模块，在通信过程中，可以基于环境数据进行调整，上述“调整”可以通过神经网络训练实现，也可以通过解析表达式求解实 现。
上述通信架构中，第一处理模块410和第三处理模块430可以采用离线训练，第二处理模块420可以根据实际环境的变化进行调整，可以通过较低的复杂度，适配环境的变化，实现有效的数据收发。
第一处理模块和第三处理模块由海量环境数据离线训练获得。通信设备中可以预先存储一个或多个第一处理模块，和/或，一个或多个第三处理模块。可选的，该一个或多个第一处理模，和/或，一个或多个第三处理模块可能是从模型服务器中下载的，或者是由第三方提供的。实际使用时，通信设备可以根据自身的计算能力、存储能力、功耗、以及任务类型选择第一处理模块或第三处理模块。而第二处理模块是通信设备本地设置的可以基于实时环境数据进行调整的模块，随着通信环境的变化，第二处理模块的输入/输出参数或模型参数等可能会发生变化，可能导致第一处理模块和第二处理模块不匹配，或者第三处理模块和第二处理模块不匹配的问题。因此通信设备在选择第一处理模块/第三处理模块和第二处理模块的选择时，需要考虑它们之间的适配问题。此外，收发两端的通信设备可以分别选择第一处理模块和第三处理模块共同实现数据收发，第一处理模块的选择和第三处理模块的选择也需要适配才能实现有效通信。
针对上述问题，本申请实施例提供一种注册的流程，发送侧的通信设备的第一处理模块可以通过注册流程与该通信设备内部的第二处理模块适配，还可以通过注册到第二处理模块与接收端的第三处理模块适配，其中第二处理模块可以根据实际的传输环境调整，使得各个处理模块完成数据处理，并适应变化的通信环境，实现有效的数据收发。
一种可能的实现中，可以通过处理实体和自适应实体之间的接口，例如，上述第一接口和第二接口，定义了处理实体和自适应实体之间的交互流程和/或信令。交互流程可以包括注册过程，数据传输过程等。信令涉及的参数可以包括处理实体和自适应实体之间传输的标识信息、数据类型、数据精度、输入维度、输出维度等等。
第一接口和第二接口定义的信令涉及的参数的值可以是标准预定义的，也可以是通过半静态配置或动态配置的。一种可能的实现中，标识信息可以和具体实现的任务类型相关。可选的，任务类型可以通过无线资源控制(radio resource control，RRC)信令指示，例如，在RRC中增加的任务类型或任务标识的字段。一种可能的实现中，数据类型和数据精度可以和任务的性能需求相关。可选的，任务的性能需求可以通过RRC信令指示，例如，在RRC中增加性能需求的字段。一种可能的实现中，输入维度和输出维度可以和实际的传输资源、传输策略等相关，传输资源或传输策略可以通过下行控制信息(downlink control information，DCI)信令指示。传输策略可以为调制和编码方案(modulation and coding scheme，MCS)。示例性的，基于传输资源和MCS等级确定信息比特数量、编码后比特数量、调制后调制符号数等参数，并用这些参数确定输入维度和/或输出维度。
如图5所示，本申请实施例提供一种通信方法500。该方法适用于发送端，涉及第一控制实体、第一处理实体和第二实体。其中，第一控制实体用于控制或管理第一处理实体、第二实体以及它们之间的交互；第一处理实体和第二实体分别对应图4所示的通信架构中的第一处理模块410和第二处理模块420。第一处理实体对应的模型是通过离线训练得到，第二实体对应的模型可以根据环境信息进行调整。第一控制实体，第一处理实体和第二实体在物理上可以设置在相同或不同的通信设备上，该通信设备可以是上述终端设备或接入网设备。
下面介绍通信方法500的具体内容。
S500(可选步骤)，第一控制实体向第一处理实体发送第一注册指示，相应的，第一处理实体接收该注册指示，该注册指示包括第二实体的标识信息。
一种可能的实现中，第一控制实体从一个或多个处理实体中选择目标处理实体，即第一处理实体，以及从一个或多个自适应实体中选择目标自适应实体，即第二实体，并向第一处理实体发送包括第二实体的标识信息的注册指示。
可选的，第一控制实体根据任务类型选择目标处理实体和目标自适应实体。该任务类型可以对应：量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项。
一种可能的实现中，第一控制实体为物理层之上的协议层实体，也可以称为高层实体。例如，该第一控制实体为媒体接入控制(media access control，MAC)层实体，该注册指示为MAC层信令。
S501.第一处理实体向第二实体发送第一注册请求，相应的，第二实体接收该第一注册请求，该第一注册请求中包含第一标识。
一种可能的实现中，第一处理实体根据S500中接收的第一注册指示向第二实体发送第一注册请求。
另一种可能的实现中，第一处理实体根据自身对应的任务类型选择第二实体，并向其发送第一注册请求。
具体地，第一标识指示第一处理实体的模型信息或数据信息中的一项或多项。
模型信息可以称为模型描述，可以包括以下一项或多项：模型用途，模型类型，模型规模，模型精度或模型性能。
其中，模型用途指示第一处理实体对应的任务类型。例如，模型用途包括：量化、信源编码、信道编码、调制、MIMO预编码中的一项或多项。其中，模型类型指示第一处理实体采用的AI模型的种类。模型类型可以分级指示，例如包括大类和子类。示例性的，大类可以包括全连接神经网络(fully connected network,FCN),卷积神经网络(convolutional neural network，CNN),递归神经网络(recurrent neural network，RNN)或Transformer等；其中，CNN的子类可以包括：基于空间利用的CNN、基于深度的CNN、多路径CNN、基于注意力机制的CNN等；RNN的子类可以包括：门递归单元(GRU，gated recurrent unit)、长短期记忆(LSTM，long short-term memory)等。其中，模型规模指示模型的大小。模型为神经网络时，模型规模可以包括神经网络层、每层神经元、权重和偏置中的一项或多项参数的数量。其中，模型精度指示模型参数使用的精度。模块精度可以包括：二进制、整形或浮点型等，其中浮点型还可以包括float16、float32等。其中，模型性能指示模型的推理性能。模型的推理性能可以通过该模型对参考数据集的进行推理的准确度表示。
数据信息可以称为数据描述。数据信息可以包括数据类型，数据维度，数据精度中的一项或多项。
其中，数据类型指示模型可以处理的数据的类型。数据类型可以包括：比特、整数、实数、复数等。其中，数据维度指示模型的输入维度和/或输出维度。数据维度可以包括：输入数据向量中元素的个数，或输出数据向量中元素的个数。其中，数据精度指示模型可以处理的数据的精度。数据精度可以包括：二进制、整形或浮点型等，其中浮点型还可以包括float16、float32等。
S502.第二实体向第一处理实体发送第一注册响应，相应的，第一处理实体接收该第一注册响应，该第一注册响应中包含第二实体的输入参数。
具体地，第二实体根据接收到的第一注册请求中的第一标识确定第一处理实体的模型信息和/或数据信息是否符合该第二实体的需求。第二实体确定第一处理实体的模型信息和/或数据信息符合该第二实体的需求，则向第一处理实体发送第一注册响应。
可选的，第二实体确定第一处理实体的模型信息和/或数据信息不符合该第二实体的需求时，可以向第一处理实体反馈注册失败指示；以及，可选的，第一处理实体向第一控制实体上报该注册失败指示。
可选的，第二实体确定第一处理实体的模型信息和/或数据信息不符合该第二实体的需求时，可以不反馈注册失败指示，隐式指示注册失败。第一处理实体发送第一注册请求的同时，启动第一注册响应计时，若在计时超时之前接收第一注册响应，则清零计时；若在计时超时，则向第一控制实体发送第一注册响应超时指示。
一种可能的实现中，第二实体的输入参数包括第二实体的输入数据的类型、维度或精度中的一项或多项。
S503(图中未单独示出).第一处理实体根据接收到的第一注册响应发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
第一处理实体的输出参数包括第一处理实体的输出数据的类型、维度或精度中的一项或多项。
具体地，第一处理实体发送第一响应确认之前判断第一处理实体的输出参数是否与第二实体的输入参数匹配。例如，该第一处理实体的输出数据的类型是否与第二实体的输入数据的类型匹配，该第一处理实体的输出数据的维度是否与第二实体的输入数据的维度匹配，和/或该第一处理实体的输出数据的精度是否与第二实体的输入数据的精度匹配。
第一处理实体确定第一处理实体的输出参数与第二实体的输入参数匹配成功则向第二实体发送第一响应确认(S503a)；若匹配失败则向第一控制实体发送第一响应确认(S503b)。
S503a.第一处理实体向第二实体发送第一响应确认，第一响应确认指示第一处理实体的输出参数与第二实体的输入参数匹配成功，相应的，第二实体接收该第一响应确认。
一种可能的实现中，第一处理实体确定第一处理实体的输出参数与第二实体的输入参数匹配成功后，第一处理实体对第一输入数据进行第一处理得到第一输出数据，并通过第一接口向第二实体传输第一输出数据。
第一处理实体、第二实体和第二处理实体的数据处理和数据传输的过程，具体参见下文中图10相关的描述。
S503b.第一处理实体向第一控制实体发送第一响应确认，该第一响应确认指示第一处理实体的输出参数与第二实体的输入参数匹配失败，相应的，第一控制实体接收该第一响应确认。
可选的，第一处理实体确定第一处理实体的输出参数与第二实体的输入参数匹配失败，可以向第一处理实体反馈参数匹配失败指示。
可选的，第一处理实体确定第一处理实体的输出参数与第二实体的输入参数匹配失败后，可以不反馈参数匹配失败指示。第二实体发送第一注册响应的同时，启动第一响应确认计时，若在计时超时之前接收第一响应确认，则清零计时；若在计时超时，则向第一控制实体发送第一响应确认超时指示。第二实体通过隐式的方式获知参数匹配失败，可以节省信令开销。
第一处理实体对应的模型的参数是通过离线训练得到，第二实体对应的模型的参数可以根据环境信息进行调整。实际通信过程中，第一处理实体无需重新训练第一处理实体的模型， 而是通过图5所示的注册流程，注册到合适的(模型信息和/或数据信息相符，输入输出参数匹配)第二实体，进行数据处理和传输。该方法可以实现通过较低的复杂度，适配实际环境的变化，实现有效的数据收发。
如图6所示，本申请实施例提供一种通信方法600。该方法适用于接收端，涉及第三控制实体、第三处理实体和第二实体。其中，第三控制实体用于控制或管理第三处理实体、第二实体以及它们之间的交互；第三处理实体和第二实体分别对应图4所示的通信架构中的第三处理模块430和第二处理模块420。第三处理实体对应的模型是通过离线训练得到，第二实体对应的模型可以根据环境信息进行调整。第三控制实体，第三处理实体和第二实体在物理上可以设置在相同或不同的通信设备上，该通信设备可以是上述终端设备或接入网设备。
下面介绍通信方法600的具体内容。
S600(可选步骤)，第三控制实体向第三处理实体发送第二注册指示，相应的，第三处理实体接收该第二注册指示，该第二注册指示包括第二实体的标识信息。
一种可能的实现中，第三控制实体从一个或多个处理实体中选择第三处理实体，以及从一个或多个自适应实体中选择第二实体，并向第三处理实体发送包括第二实体的标识信息的注册指示。
可选的，第三控制实体根据任务类型选择第三处理实体和第二实体。该任务类型可以对应：解调、信道译码、信道译码中的一项或多项。
一种可能的实现中，第三控制实体为物理层之上的协议层实体，也可以称为高层实体。例如，该第三控制实体为MAC层实体，该注册指示为MAC层信令。
当第一处理实体和第三处理实体共同实现数据收发时，第三控制实体通过和第三控制实体对应的协议层的交互机制(例如，MAC层协议)与第一控制实体交互，确保第三处理实体和第一处理实体注册到同一个第二实体上。
S601.第三处理实体向第二实体发送第二注册请求，相应的，第二实体接收该第二注册请求，该第二注册请求中包含第一标识。
一种可能的实现中，第三处理实体根据S600中接收的第二注册指示向第二实体发送第二注册请求。
另一种可能的实现中，第三处理实体根据自身对应的任务类型选择第二实体，并向其发送第二注册请求。
具体地，第一标识指示第三处理实体的模型信息或数据信息中的一项或多项。
模型信息可以称为模型描述，可以包括以下一项或多项：模型用途，模型类型，模型规模，模型精度或模型性能。
其中，模型用途指示第三处理实体对应的任务类型。例如，模型用途包括：解调、信道译码、信道译码中的一项或多项。其他模型信息可以参考S501处相关的描述，此处不再赘述。
数据信息可以称为数据描述。数据信息可以包括数据类型，数据维度，数据精度中的一项或多项。
其中，数据类型指示模型可以处理的数据的类型。其中，数据维度指示模型的输入维度和/或输出维度。数据维度可以包括：输入数据向量中元素的个数，或输出数据向量中元素的个数。其他数据信息相关的描述参数可以参考S501处相关的描述，此处不再赘述。
S602.第二实体向第三处理实体发送第二注册响应，相应的，第三处理实体接收该第二注册响应，该第二注册响应中包含第二实体的输出参数。
一种可能的实现中，第二实体的输出参数包括第二实体的输出数据的类型、维度或精度中的一项或多项。
第二实体向第三处理实体发送第二注册响应的其他方面，可以参考S502中第二实体向第一处理实体发送第一注册响应的相关描述，此处不再赘述。
可选的，第二实体确定第三处理实体的模型信息和/或数据信息不符合该第二实体的需求时，可以向第三处理实体反馈注册失败指示；以及，可选的，第三处理实体向第三控制实体上报该注册失败指示。
可选的，第二实体确定第三处理实体的模型信息和/或数据信息不符合该第二实体的需求时，可以不反馈注册失败指示，隐式指示注册失败。第三处理实体发送第二注册请求的同时，启动第二注册响应计时，若在计时超时之前接收第二注册响应，则清零计时；若在计时超时，则向第三控制实体发送第二注册响应超时指示。
S603(图中未单独示出).第三处理实体根据接收到的第二注册响应发送第二响应确认，该第二响应确认指示第三处理实体的输出参数是否与第二实体的输入参数匹配。
第三处理实体的输入参数包括第三处理实体的输入数据的类型、维度或精度中的一项或多项。
具体地，第三处理实体发送第二响应确认之前判断第三处理实体的输入参数是否与第二实体的输出参数匹配。例如，该第三处理实体的输入数据的类型是否与第二实体的输出数据的类型匹配，该第三处理实体的输入数据的维度是否与第二实体的输出数据的维度匹配，和/或该第三处理实体的输入数据的精度是否与第二实体的输出数据的精度匹配。
第三处理实体确定第三处理实体的输入参数与第二实体的输出参数匹配成功则向第二实体发送第二响应确认(S603a)；若匹配失败则向第三控制实体发送第二响应确认(S603b)。
S603a.第三处理实体向第二实体发送第二响应确认，第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配成功，相应的，第二实体接收该第二响应确认。
一种可能的实现中，第三处理实体确定第三处理实体的输入参数与第二实体的输出参数匹配成功后，第三处理实体通过第二接口接收第二实体传输的第二输出数据，并对该第二输出数据进行第二处理得到第三数据。具体内容参见下文中图10相关的描述。
S603b.第三处理实体向第三控制实体发送第二响应确认，该第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配失败，相应的，第三控制实体接收该第二响应确认。
可选的，第三处理实体确定第三处理实体的输入参数与第二实体的输出参数匹配失败，可以向第二实体反馈参数匹配失败指示。
可选的，第三处理实体确定第三处理实体的输入参数与第二实体的输出参数匹配失败，可以不反馈参数匹配失败指示。第二实体发送第二注册响应的同时，启动第二响应确认计时，若在计时超时之前接收第二响应确认，则清零计时；若在计时超时，则向第三控制实体发送第二响应确认超时指示。第二实体通过隐式的方式获知参数匹配失败，可以节省信令开销。
第三处理实体对应的模型是通过离线训练得到，第二实体对应的模型可以根据环境信息进行调整。实际通信过程中，第三处理实体无需重新训练第三处理实体的模型，而是通过图6所示的注册流程，注册到合适的(模型信息和/或数据信息相符，输入输出参数匹配)第二 实体，进行数据处理和传输。该方法可以实现通过较低的复杂度，适配实际环境的变化，实现有效的数据收发。
上述通信方法500和通信方法600中，注册过程涉及的三个步骤：注册请求、注册响应、响应确认都可能造成注册失败，例如注册响应超时、响应确认超时或两个实体之间的输入输出参数不匹配会导致处理实体向第二实体注册失败，对于任意一注册阶段的失败，都可以通过选择新的第一处理实体/第三处理实体向第二实体注册流程，或第一处理实体/第三处理实体向新的第二实体发起新的注册流程实现。图7-9介绍了本申请实施例提供的几种可能的注册失败后的流程。
如图7所示，本申请实施例提供一种通信方法700。该方法描述了第一处理实体向第二实体发送注册请求后，产生注册失败的情况的处理方法。通信方法700以第一处理实体向第二实体注册为例进行了描述，同样也适用于第三处理实体向第二实体注册。该方法涉及第一控制实体，第一处理实体，第二实体以及第四实体，其中第四实体与第二实体类似，其模型可以根据环境信息进行调制。
下面介绍通信方法700的具体内容。
S700.第一控制实体向第一处理实体发送第一注册指示，相应的，第一处理实体接收该注册指示，该注册指示包括第二实体的标识信息。
S701.第一处理实体向第二实体发送第一注册请求，相应的，第二实体接收该第一注册请求，该第一注册请求中包含第一标识。
S700和S710具体内容参见S500和S501，此处不再赘述。
S702.第一处理实体发送第一注册请求时，启动第一注册响应计时。
当第一注册响应计时超时前接收第二实体发送的第一注册响应，则停止计时或计时重置。计时时长可以是预先定义的。
当第一注册响应计时超时则执步骤S703。
S703.第一处理实体向第一控制实体发送第一注册响应超时指示，相应的，第一控制实体接收该第一注册响应超时指示。
S704.第一控制实体向第一处理实体发送第三注册指示，相应的，第一处理实体接收该第三注册指示。第三注册指示包括第四实体的标识。
具体地，第一控制实体接收第一注册响应超时指示后，从一个或多个自适应实体中选择另一个实体，即第四实体，并向第一处理实体发送包括第四实体的标识信息的第三注册指示。
可选的，第四实体也可以是根据第一处理实体的输出参数对第二实体进行更新后的实体。
S705.第一处理实体向第四实体发送第三注册请求，相应的，第四实体接收该第一注册请求，该第一注册请求中包含第一标识。S705的具体内容参见S501，此处不再赘述。
方法700中，第一处理实体发第一注册请求后启动第一注册响应计时，若计时超时则向第一控制实体上报超时指示，由第一控制实体向第一处理实体指示另一个自适应实体，避免长时间等待第一处理实体，提高效率。
如图8所示，本申请实施例提供一种通信方法800。该方法描述了第二实体向第一处理实体发送注册响应后产生注册失败的情况的处理方法，例如，因第二实体的输入参数与第一处理实体的输出参数不匹配而导致注册失败的后续操作。通信方法800以第一处理实体向第二实体注册为例进行了描述，同样也适用于第三处理实体向第二实体注册。该方法涉及第一控制实体，第一处理实体，第二实体以及第四实体，其中第四实体与第二实体类似，其模型 可以根据环境信息进行调制。
S800(可选步骤)，第一控制实体向第一处理实体发送第一注册指示，相应的，第一处理实体接收该注册指示，该注册指示包括第二实体的标识信息。
S801.第一处理实体向第二实体发送第一注册请求，相应的，第二实体接收该第一注册请求，该第一注册请求中包含第一标识。
S802.第二实体向第一处理实体发送第一注册响应，相应的，第一处理实体接收该第一注册响应，该第一注册响应中包含第二实体的输入参数。
S800-S802具体描述参见S500-S502，此处不再赘述。
S803.第一处理实体向第一控制实体发送第一响应确认，该第一响应确认指示第一处理实体的输出参数与第二实体的输入参数匹配失败，相应的，第一控制实体接收该第一响应确认。
第一处理实体的输出参数包括第一处理实体的输出数据的类型、维度或精度中的一项或多项。具体地，第一处理实体发送第一响应确认之前判断第一处理实体的输出参数是否与第二实体的输入参数匹配。例如，该第一处理实体的输出数据的类型是否与第二实体的输入数据的类型匹配，该第一处理实体的输出数据的维度是否与第二实体的输入数据的维度匹配，和/或该第一处理实体的输出数据的精度是否与第二实体的输入数据的精度匹配。
第一处理实体确定第一处理实体的输出参数与第二实体的输入参数匹配失败则向第一控制实体发送第一响应确认。
S804.第一控制实体向第一处理实体发送第三注册指示，相应的，第一处理实体接收该第三注册指示。第三注册指示包括第四实体的标识。
具体地，第一控制实体接收指示匹配失败的第一注册响应后，从一个或多个自适应实体中选择另一个实体，即第四实体，并向第一处理实体发送包括第四实体的标识信息的第三注册指示。
S805.第一处理实体向第四实体发送第三注册请求，相应的，第四实体接收该第一注册请求，该第一注册请求中包含第一标识。S705的具体内容参见S501，此处不再赘述。
方法800中，第一处理实体确定该第一处理实体的输出参数于第一注册响应中指示的第一实体的输入参数匹配失败时，向第一控制实体发送指示匹配失败的第一注册响应，由第一控制实体向第一处理实体指示另一个可以注册的自适应实体，避免多次向第二实体注册，从而提高效率。
如图9所示，本申请实施例提供一种通信方法900。该方法描述了因第一注册响应计时超时而导致注册失败的后续操作。通信方法900以第一处理实体向第二实体注册为例进行了描述，同样也适用于第三处理实体向第二实体注册。该方法涉及第一控制实体，第一处理实体，第二实体以及第四处理实体，其中第四处理实体与第一处理实体类似，第四处理实体对应的模型是通过离线训练得到。
S900(可选步骤)，第一控制实体向第一处理实体发送第一注册指示，相应的，第一处理实体接收该注册指示，该注册指示包括第二实体的标识信息。
S901.第一处理实体向第二实体发送第一注册请求，相应的，第二实体接收该第一注册请求，该第一注册请求中包含第一标识。
S902.第二实体向第一处理实体发送第一注册响应，相应的，第一处理实体接收该第一注册响应，该第一注册响应中包含第二实体的输入参数。
S900-S902具体描述可以参见S500-S502，此处不再赘述。
S903.第二实体发送第一注册请求时，启动第一响应确认计时。
当第一响应确认计时超时前接收第二实体发送的第一响应确认，则停止计时或计时重置。其中，计时时长可以是预先定义的。
当第一响应确认计时超时则执步骤S703。
S904.第二实体向第一控制实体发送第一响应确认超时指示，相应的，第一控制实体接收该第一响应确认超时指示。
S905.第一控制实体向第四处理实体发送第四注册指示，相应的，第四处理实体接收该第四注册指示，其中第四注册指示包含第二实体的标识。
具体地，第一控制实体接收第一响应确认超时指示后，从一个或多个处理实体中选择不同于第一处理实体的目标处理实体，即第四处理实体，并向第四处理实体发送包含第二实体的标识的第四注册指示。
第一控制实体从一个或多个处理实体中选择目标处理实体的具体内容参见S500处相关的描述，此处不再赘述。
S906.第四处理实体向第二实体发送第三注册请求，相应的，第二实体接收该第三注册请求，其中第三注册请求包含第三标识。
具体地第四处理实体根据接收到的第三注册指示向第二实体发送第三注册请求。该第三注册请求包含第三标识。第三标识指示第四处理实体的模型信息或数据信息中的一项或多项。具体内容可以参考S501中第一标识相关的描述此处不再赘述。
方法900的后续步骤同方法500中第二实体接收第一处理实体的第一注册请求后的步骤，此处不再赘述。方法900中，当第一处理实体在第一响应确认计时超时前未向第二实体反馈第一响应确认时，第二实体向第一控制实体上报第一响应超时确认指示，由第一控制实体选择另一个目标处理实体注册到第二实体，避免第一处理实体多次向第二实体注册失败，从而提高效率。
当收发两端的目标处理实体通过上述图5-图9中描述的任一种可能的方法注册到目标自适应实体后，可以进行数据处理和数据收发。则如图10所示，本申请实施例提供一种数据传输方法1000。该方法以第一处理实体、第二实体和第三处理实体为例进行描述，其中第一处理实体和第二实体可以设置在发送端，第三处理实体可以设置在接收端。第一处理实体的输出参数与第二实体的输入参数匹配，第二实体的输出参数与第三处理实体的输入参数匹配。第一处理实体对数据进行第一处理，第三处理实体对数据进行第二处理，第二处理可以是第一处理的逆操作。
下面介绍通信方法1000的具体内容。
S1001.第一处理实体对第一输入数据进行第一处理得到第一输出数据。
一种可能的实现中，第一处理实体实现编码调制功能，第一处理包括信道编码和调制，或信道编码和调制的等效处理，第一输入数据为信源编码后的比特流，第一输出数据为调制后的符号序列。
示例性的，第一处理实体的结构如图11所示，包括输入适配模块，数据空间转换模块，输出适配模块以及可选的参数配置模块。其中第一输入数据对应图中的输入1101，可以为信源编码后的比特流。其中，输入适配模块用于对输入1101进特征嵌入和提取，使得输入1101适配数据空间转换模块的输入维度。输入适配模块的输出为比特分组，每个分组中的比特的 数据量与数据空间转换模块的输入维度对应。数据空间转换模块用于对输入适配模块的输出进行子空间转换得到调制后的符号序列，以实现编码和调制功能。数据空间转换模的参数(包括其输入维度)可以根据信道编码码率、调制阶数、传输资源数中的一项或多项确定。可选的，参数配置模块用于根据信道编码码率、调制阶数、传输资源数中的一项或多项配置数据空间转换模块的参数。输出适配模块，用于将数据空间转换模块的输出适配至特定的数据类型、数据维度和数据精度得到第一输出数据，即输出1102。
示例性的，输入配置模块用于适配第一输入数据的维度与数据空间转换模块的输入维度；数据空间转换模块用于对第一输入数据进行第一处理，得到第一中间输出数据；输出适配模块用于对第一中间输出数据进行维度变换得到第一输出数据。其中对第一中间输出数据的维度变换可以包括从低维到高维的变化，例如通过对第一中间输出数据复制实现升维，以应对信道条件差的情况。
一种可能的实现中，第一处理实体实现量化、信源编码、信道编码和调制功能，第一处理包括量化、信源编码、信道编码和调制，或者包括量化、信源编码、信道编码和调制的等效处理，第一输入数据为符号序列，第一输出数据为调制后符号序列。可选的，第一输入数据为感知数据。
示例性的，第一处理实体的结构如图11所示，包括输入适配模块，数据空间转换模块，输出适配模块以及可选的参数配置模块。其中第一输入数据对应图中的输入1101，可以为符号序列。其中，输入适配模块用于对输入1101进特征嵌入和提取，使得输入1101适配数据空间转换模块的输入维度。输入适配模块的输出为符号分组，每个分组中符号的数量与数据空间转换模块的输入维度对应。数据空间转换模块用于对输入适配模块的输出进行子空间转换得到调制后的符号，以实现量化、信源编码、信道编码和调制功能。数据空间转换模的参数(包括其输入维度)可以根据量化等级、信源编码方案、信道编码码率、调制阶数和传输资源数中的一项或多项确定。可选的，参数配置模块用于根据信道编码码率、调制阶数和传输资源数中的一项或多项配置数据空间转换模块的参数。输出适配模块，用于将数据空间转换模块的输出适配至特定的数据类型、数据维度和数据精度得到第一输出数据，即输出1102。
一种可能的实现中，数据空间转换模块为神经网络模型，第一处理实体通过离线训练得到。
S1002.第一处理实体通过第一接口传输第一输出数据，相应的，第二实体通过该第一接口接收该第一输出数据。
S1003.第二实体对第一输出数据进行第二处理得到第二输出数据。
一种可能的实现中，第二处理包括将第一输出数据映射至第二输出数据。第一输出数据至第二输出数据的映射根据实时环境参数确定。
示例性的，第二实体对应图4中的第二处理模块。第二处理模块的结构如图12所示，包括输入节点，输出节点以及无线信道。其中输入节点数与第二实体的输入维度对应，输出节点数与第二实体的输出维度对应。
输入节点和输出节点之间的一种可能的映射可以表达为：
其中，xj为第j个输入值，其中j∈J，J表示输入节点数，yi为第i个输出值，其中i∈I， I表示输出节点数，wji为第j个输入值和第i个输出值之间的权重。其中wji可以基于实时环境参数调整，例如使用信息传递算法(message passing algorithm，MPA)、置信传播(belief propagation，BP)、最小均方误差(minimum mean squared error，MMSE)、加权最小均方误差(weighted MMSE，WMMSE)、最大似然(maximum likelihood，ML)、最大后验概率(maximum a posteriori，MAP)或其他算法进行更新。上述映射表达式为示例性的基础表达式，实际应用时可能对上述映射表达式进行变换等，本申请不做限定。
S1004.第二实体通过第二接口向第三处理实体传输第二输出数据，相应的，第三处理实体接收该第二输出数据。
S1005.第三处理实体对第二输出数据进行第三处理得到第三数据。
一种可能的实现中，第一处理实体实现编码调制功能，相应的，第三处理实现解调和信道译码功能。第三处理可以理解为是第一处理的逆过程。第三处理实体的输入，即第二输出数据，为符号序列；第三处理实体的输出，即第三数据，为比特序列。
示例性的，第三处理实体的结构如图11所示，包括输入适配模块，数据空间转换模块，输出适配模块以及可选的参数配置模块。其中第二输出数据对应图中的输入1101，可以为调制后的符号序列。其中，输入适配模块用于对输入1101进特征嵌入和提取，使得输入1101适配数据空间转换模块的输入维度。输入适配模块的输出为符号分组，每个分组中符号的数量与数据空间转换模块的输入维度对应。数据空间转换模块用于对输入适配模块的输出进行子空间转换得到调制后的符号序列，以实现解调和信道译码功能。数据空间转换模的参数(包括其输入维度)可以根据信道编码码率、调制阶数、传输资源数中的一项或多项确定。可选的，参数配置模块用于根据信道编码码率、调制阶数、传输资源数中的一项或多项配置数据空间转换模块的参数。输出适配模块，用于将数据空间转换模块的输出适配至后续模块，得到输出1102。
一种可能的实现中，第一处理实体实现量化、信源编码、信道编码和调制功能，相应的，第三处理实现解调、信道译码和信源译码功能。第三处理实体的输入，即第二输出数据，为调制后的符号序列；第三处理实体的输出，即第三数据，为符号序列。
示例性的，第三处理实体的结构如图11所示，包括输入适配模块，数据空间转换模块，输出适配模块以及可选的参数配置模块。其中第一输入数据对应图中的输入1101，可以为调制后的符号序列。其中，输入适配模块用于对输入1101进特征嵌入和提取，使得输入1101适配数据空间转换模块的输入维度。输入适配模块的输出为符号分组，每个分组中符号的数量与数据空间转换模块的输入维度对应。数据空间转换模块用于对输入适配模块的输出进行子空间转换得到调制后的符号，以实现量化、信源编码、信道编码和调制功能。数据空间转换模的参数(包括其输入维度)可以根据量化等级、信源编码方案、信道编码码率、调制阶数、传输资源数的一项或多项确定。可选的，参数配置模块用于根据信道编码码率、调制阶数、传输资源数的一项或多项配置数据空间转换模块的参数。输出适配模块，用于将数据空间转换模块的输出适配至后续模块，得到输出1102。
一种可能的实现中，数据空间转换模块为神经网络模型，第三处理实体通过离线训练得到。
一种可能的实现中，第一输入数据为感知数据，第三数据为第三处理实体恢复处的感知数据。可选的，第三处理实体对应的通信设备还可以对该感知时间进行进一步处理得到感知结果。
通信方法1000中，可以通过互相匹配的第一处理实体、第二实体以及第三处理实体，实现数据处理和数据收发，其中第一处理实体和第三处理实体可以通过离线训练得到，第二实体可以根据实时环境参数进行调整，以较低的复杂度适配变化的传输环境，实现有效的数据收发。
本申请实施例中第一处理实体和第三处理实体可以通过神经网络实现，第二实体可以通过通过自适应算法或自适应迭代算法实现，这些算法包括但不限于MPA、BP、MMSE、WMMSE、最大似然ML、MAP等算法。如图13所示，图13为通过神经网络和自适应迭代算法实现本申请实施例提供的方法的具体示例。第一处理模块410对应第一处理实体，第二处理模块420对应第二实体，第三处理模块430对应第三处理实体。
第一处理模块410包括神经网络模型NN Tx；第三处理模块430包括神经网络模型NN Rx。NN Tx和NN Rx可以通过自编码器或transformer实现，并基于海量环境参数样本或通过数学模型建模产生的训练样本进行离线预训练得到。
第二处理模块420包括迭代算法模块，无线信道以及可选的波束赋形，MIMO预编码等模块。迭代运算模块可以基于实时环境参数进行在线调整。
一种可能的实现中，该实时环境参数是根据第三处理模块的反馈指示确定的，该反馈指示包括信道状态信息或信道特征向量，其中信道特征向量是根据信道状态信息提取的特征值向量。可选的，该反馈指示还包括能力开关信息，用于开启或关闭第一处理模块410和/或第二处理模块420。
一种可能的实现中，迭代算法模块可以通过MPA、BP、MMSE、WMMSE、最大似然ML或MAP等算法实现。
为了实现上述本申请实施例提供的方法中的各功能，第一处理实体、第二实体、第三处理实体均可以包括硬件结构和/或软件模块，以硬件结构、软件模块、或硬件结构加软件模块的形式来实现上述各功能。上述各功能中的某个功能以硬件结构、软件模块、还是硬件结构加软件模块的方式来执行，取决于技术方案的特定应用和设计约束条件。
如图14所示，本申请实施例提供了一种通信装置1400。该通信装置1400可以是终端或网络设备，也可以是终端设备或网络设备中的装置，或者是能够和终端设备、网络设备匹配使用的装置。一种可能的实现中，该通信装置1400可以包括执行上述方法实施例中第一处理实体、第二实体、第三处理实体执行的方法/操作/步骤/动作所一一对应的模块或单元，该单元可以是硬件电路，也可是软件，也可以是硬件电路结合软件实现。一种可能的实现中，该通信装置1400可以包括处理单元1410和收发单元1420。处理单元1410可以用于调用收发单元1420执行接收和/或发送的功能。
一种可能的实现中，通信装置1400用于执行第一处理实体的操作，收发单元1420，用于向第二实体发送第一注册请求，第一注册请求包括第一标识，该第一标识指示该通信装置的模型信息或数据信息中的一项或多项，其中，第二实体的模型参数根据实时环境参数进行调整；收发单元1420还用于接收第二实体发送的第一注册响应，该第一注册响应指示第二实体的输入参数；收发单元1420还用于发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
一种可能的实现中，通信装置1400用于执行第二实体的操作，具体地，收发单元1420用于接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环 境参数进行调整；收发单元1420用于向第一处理实体发送第一注册响应，第一注册响应指示该通信装置的输入参数；该通信装置的输入参数用于确定第一处理实体的输出参数是否与该通信装置的输入参数匹配。
一种可能的实现中，第一处理实体和第二实体属于同一个通信装置，即通信装置1400既可以实现第一处理实体所执行的操作，也可以实现第二实体所执行的操作。
一种可能的实现中，通信装置1400用于执行第三处理实体的操作，收发单元1420用于向第二实体发送第二注册请求，第二注册请求包括第二标识，该第二标识指示该通信装置的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；收发单元1420还用于接收第二实体发送的第二注册响应，该第二注册响应指示第二实体的输出参数；收发单元1420还用于发送第二响应确认，所述第二响应确认指示该通信装置的输入参数是否与第二实体的输出参数匹配。
一种可能的实现中，通信装置1400用于执行第二实体的操作，
在本申请各个实施例中的各功能模块或单元可以集成在一个处理器中，也可以是单独物理存在，也可以两个或两个以上模块或单元集成在一个模块或单元中。上述集成的模块或单元既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。一种可能的实现中，处理单元1410可以是处理器，收发单元1420可以是收发器。
参见图15，本申请实施例还提供了一种通信装置1500，用于实现上述方法中第一处理实体、第二实体或第三处理实体功能。该通信装置可以是终端、网络设备(源接入网设备、目标接入网设备或计算管理功能)，也可以是终端、网络设备中的装置(如，芯片、电路等)，或者是能够和终端、网络设备匹配使用的装置。通信装置1500包括至少一个处理器1510，通信装置1500还可以包括通信接口1520。在本申请实施例中，通信接口可以是收发器、电路、总线、模块或其它类型的通信接口，用于通过传输介质和其它设备进行通信。例如，通信接口1520用于通信装置1500中的装置可以和其它设备进行通信。
处理器1510可以执行通信装置1400中处理单元1410所执行的功能；通信接口1520可以用于执行通信装置1400中收发单元1420所执行的功能。
一种可能的实现中，当通信装置1500用于执行第一处理实体所执行的操作时，通信接口1520，用于向第二实体发送第一注册请求，第一注册请求包括第一标识，该第一标识指示该通信装置的模型信息或数据信息中的一项或多项，其中，第二实体的模型参数根据实时环境参数进行调整；还用于接收第二实体发送的第一注册响应，该第一注册响应指示第二实体的输入参数；还用于发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
一种可能的实现中，当通信装置1500用于执行第二实体执行的操作时，通信接口1520，用于接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；还用于向第一处理实体发送第一注册响应，第一注册响应指示该通信装置的输入参数；该通信装置的输入参数用于确定第一处理实体的输出参数是否与该通信装置的输入参数匹配。
一种可能的实现中，当通信装置1500用于执行第二实体执行的操作时，通信接口1520，用于接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所 述第三处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；还用于向第三处理实体发送第二注册响应，第二注册响应指示该通信装置的输出参数；该通信装置的输出参数用于确定所述第三处理实体的输入参数是否与所述第二实体的输出参数匹配。
一种可能的实现中，当通信装置1500用于第三处理实体执行的操作时，通信接口1520用于向第二实体发送第二注册请求，第二注册请求包括第二标识，该第二标识指示该通信装置的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；还用于接收第二实体发送的第二注册响应，该第二注册响应指示第二实体的输出参数；还用于发送第二响应确认，所述第二响应确认指示该通信装置的输入参数是否与第二实体的输出参数匹配。
通信装置1500还可以包括至少一个存储器1530，用于存储程序指令和/或数据。存储器1530和处理器1510耦合。本申请实施例中的耦合是装置、单元或模块之间的间接耦合或通信连接，可以是电性，机械或其它的用于装置、单元或模块之间的信息交互的形式。处理器1510可能和存储器1530协同操作。处理器1510可能执行存储器1530中存储的计算机程序或指令。在一种可能的实现中，至少一个存储器中的至少一个可以与处理器集成在一起。在另一种可能的实现中，存储器1530位于该通信装置1500之外。
本申请实施例中不限定上述通信接口1520、处理器1510以及存储器1530之间的具体连接介质。本申请实施例在图15中以存储器1530、处理器1510以及通信接口1520之间通过总线1540连接，总线在图15中以粗线表示，其它部件之间的连接方式，仅是进行示意性说明，并不引以为限。该总线可以分为地址总线、数据总线、控制总线等。为便于表示，图15中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
一种可能的实现中，该通信装置1500可以为芯片系统。本申请实施例中，芯片系统可以由芯片构成，也可以包含芯片和其他分立器件。
参见图16，本申请实施例还提供了一种通信装置1600，用于实现上述方法中第一处理实体、第二实体或第三处理实体的功能。该通信装置可以是终端、网络设备，也可以是终端、网络设备中的装置(如，芯片、电路等)，或者是能够和终端、网络设备匹配使用的装置。该通信装置包括处理器1610，该处理器用实现上述第一通信装置、第二通信装置或终端的部分或全部功能。
一种可能的实现中，当通信装置1600用于执行第一处理实体所执行的操作时，处理器1610，用于向第二实体发送第一注册请求，第一注册请求包括第一标识，该第一标识指示该通信装置的模型信息或数据信息中的一项或多项，其中，第二实体的模型参数根据实时环境参数进行调整；还用于接收第二实体发送的第一注册响应，该第一注册响应指示第二实体的输入参数；还用于发送第一响应确认，该第一响应确认指示第一处理实体的输出参数是否与第二实体的输入参数匹配。
一种可能的实现中，当通信装置1600用于执行第二实体执行的操作时，处理器1610，用于接收第一处理实体发送的第一注册请求，第一注册请求包括第一标识，第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；还用于向第一处理实体发送第一注册响应，第一注册响应指示该通信装置的输入参数；该通信装置的输入参数用于确定第一处理实体的输出参数是否与该通信装置的输入参数匹配。
一种可能的实现中，当通信装置1600用于执行第二实体执行的操作时，处理器1610，用于接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，该通信装置的模型参数根据实时环境参数进行调整；还用于向第三处理实体发送第二注册响应，第二注册响应指示该通信装置的输出参数；该通信装置的输出参数用于确定所述第三处理实体的输入参数是否与所述第二实体的输出参数匹配。
一种可能的实现中，当通信装置1600用于第三处理实体执行的操作时，处理器1610用于向第二实体发送第二注册请求，第二注册请求包括第二标识，该第二标识指示该通信装置的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；还用于接收第二实体发送的第二注册响应，该第二注册响应指示第二实体的输出参数；还用于发送第二响应确认，所述第二响应确认指示该通信装置的输入参数是否与第二实体的输出参数匹配。
一种可能的实现中，处理器1610通过执行存储器1620中存储的指令，以实现第一处理实体、第二实体或第三处理实体实现的功能。可选的，该通信装置还包括存储器1620。可选的，处理器1610和存储器1620集成在一起。可选的，存储器1620在通信装置1600之外。
一种可能的实现中，处理器1610可以为逻辑电路，处理器1610通过输入输出接口(图中未示出)输入/输出消息或信令。其中，逻辑电路可以是信号处理器、芯片，或其他可以实现本申请方法的集成电路。
当上述通信装置为应用于终端的芯片时，该终端芯片实现上述方法实施例中终端的功能。该终端芯片从终端中的其它模块(如射频模块或天线)接收信息，该信息是其他终端或网络设备发送给终端的；或者，该终端芯片向终端中的其它模块(如射频模块或天线)输出信息，该信息是终端发送给其他终端或网络设备的。
当上述通信装置为应用于网络设备的芯片时，该网络设备芯片实现上述方法实施例中网络设备的功能。该网络设备芯片从网络设备中的其它模块(如射频模块或天线)接收信息，该信息是终端或其他网络设备发送给该网络设备的；或者，该网络设备芯片向网络设备中的其它模块(如射频模块或天线)输出信息，该信息是网络设备发送给终端或其他网络设备的。
本申请实施例中，处理器(例如处理器1510，处理器1610)可以是一个或多个中央处理单元(central processing unit，CPU)，在处理器是一个CPU的情况下，该CPU可以是单核CPU，也可以是多核CPU。处理器可以是通用处理器、数字信号处理器、专用集成电路、现场可编程门阵列或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件，可以实现或者执行本申请实施例中公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件处理器执行完成，或者用处理器中的硬件及软件模块组合执行完成。
本申请实施例中，存储器(例如存储器1530，存储器1620)可包括但不限于硬盘(hard disk drive，HDD)或固态硬盘(solid-state drive，SSD)等非易失性存储器，随机存储记忆体(Random Access Memory，RAM)、可擦除可编程只读存储器(Erasable Programmable ROM，EPROM)、只读存储器(Read-Only Memory，ROM)或便携式只读存储器(Compact Disc Read-Only Memory，CD-ROM)等等。存储器是能够用于携带或存储具有指令或数据结构形式的计算机程序并能够由计算机存取的任何其他介质，但不限于此。本申请实施例中的存储器还可以是电路或者其它任意能够实现存储功能的装置，用于存储计算机程序或指令，和/或数据。
本申请实施例还提供一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序或指令，所述计算机程序或指令被计算机(例如，处理器)执行，以实现本申请实施例中由任意装置执行的任意一种方法的部分或全部步骤。
本申请所提供的装置如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质(例如软盘、硬盘、磁带)、光介质(例如光盘)、或者半导体介质(例如固态硬盘)等。
本申请实施例还提供了一种包括计算机程序或一组指令的计算机程序产品，当所述计算机程序产品在计算机上运行时，使得以上各方面的任意一种方法的部分或者全部步骤被执行。
在上述实施例中的方法可全部或部分地通过软件、硬件、固件、或其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输。
本申请还提供一种芯片或芯片系统，该芯片可包括处理器。该芯片还可包括存储器(或存储模块)和/或收发器(或通信模块)，或者，该芯片与存储器(或存储模块)和/或收发器(或通信模块)耦合，其中，收发器(或通信模块)可用于支持该芯片进行有线和/或无线通信，存储器(或存储模块)可用于存储程序或一组指令，该处理器调用该程序或该组指令可用于实现上述方法实施例、方法实施例的任意一种可能的实现方式中由终端或者网络设备执行的操作。该芯片系统可包括以上芯片，也可以包含上述芯片和其他分立器件，如存储器(或存储模块)和/或收发器(或通信模块)。
基于与上述方法实施例相同构思，本申请还提供一种通信系统，该通信系统可包括以上第一处理实体、第二实体和第三处理实体。该通信系统可用于实现上述方法实施例、方法实施例的任意一种可能的实现方式中由第一处理实体、第二实体和第三处理实体执行的操作。示例性的，该通信系统可具有如图2所示结构。
在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。
以上所述，仅为本申请的一些具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可对这些实施例做出另外的变更和修改。因此，所附权利要求意欲解释为包括上述实施例以及落入本申请范围的说是有变更和修改。因此，本申请保护范围应以所述权利要求的保护范围为准。

Claims (37)

  1. 一种通信方法，所述方法用于第一通信设备的第一处理实体，其特征在于，包括：
    第一处理实体向第二实体发送第一注册请求，所述第一注册请求包括第一标识，所述第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，所述第二实体的模型参数根据实时环境参数进行调整；
    所述第一处理实体接收所述第二实体发送的第一注册响应，所述第一注册响应指示所述第二实体的输入参数；
    所述第一处理实体发送第一响应确认，所述第一响应确认指示所述第一处理实体的输出参数是否与所述第二实体的输入参数匹配。
  2. 根据权利要求1所述的方法，其特征在于，所述第一响应确认指示所述第一处理实体的输出参数与所述第二实体的输入参数匹配成功，所述方法还包括：
    所述第一处理实体对第一输入数据进行第一处理得到第一输出数据，所述第一输出数据为所述第一处理实体的输出；
    所述第一处理实体通过第一接口向所述第二实体传输所述第一输出数据。
  3. 根据权利要求2所述的方法，其特征在于，所述第一处理实体包括输入适配模块，数据空间转换模块和输出适配模块；
    所述输入配置模块用于适配所述第一输入数据的维度与所述数据空间转换模块的输入维度；
    所述数据空间转换模块用于对所述第一输入数据进行第一处理，得到第一中间输出数据；
    所述输出适配模块用于对所述第一中间输出数据进行维度变换得到第一输出数据。
  4. 根据权利要求2或3所述的方法，其特征在于，
    所述第一处理包括编码和调制，所述第一输入数据为比特流，所述第一输出数据为调制后的符号；或者
    所述第一处理包括量化、信源编码、信道编码以及调制，所述第一输入数据为符号序列，所述第一输出数据为调制后的符号序列。
  5. 根据权利要求2至4任一项所述的方法，其特征在于，所述第一接口还用于传输所述注册请求、所述注册响应或所述响应确认中的一项或多项。
  6. 根据权利要求2至5任一项所述的方法，其特征在于，所述第一接口定义了所述第一处理实体和所述第二实体交互的流程和/或信令。
  7. 根据权利要求1所述的方法，其特征在于，所述第一响应确认指示匹配失败，所述方法还包括：
    所述第一处理实体向第四实体发送第二注册请求，所述第二注册请求包括第一标识，所述第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，所述第四实体的模型参数根据实时环境参数调整；
    所述第一处理实体接收所述第四实体发送的第二注册响应，所述第二注册响应指示所述第四实体的输入参数；
    所述第一处理实体发送第二响应确认，所述第二响应确认指示所述第一处理实体的输出参数是否与所述第四实体的输入参数匹配。
  8. 一种通信方法，所述方法用于第二通信设备的第二实体的，其特征在于，
    第二实体接收第一处理实体发送的第一注册请求，所述第一注册请求包括第一标识，所述第一标识指示所述第一处理实体的模型信息或数据信息中的一项或多项，所述第二实体的模型参数根据实时环境参数进行调整；
    所述第二实体向所述第二实体发送第一注册响应，所述第一注册响应指示所述第二实体的输入参数；所述第二实体的输入参数用于确定所述第一处理实体的输出参数是否与所述第二实体的输入参数匹配。
  9. 根据权利要求8所述的方法，其特征在于，还包括：所述第二实体接收第一响应确认，所述第一响应确认指示所述第一处理实体的输出参数与所述第二实体的输入参数匹配成功，所述方法还包括：
    所述第二实体通过第一接口接收第一输出数据，所述第一输出数据为所述第一处理实体的输出数据；
    所述第二实体对所述第一输出数据进行第二处理得到第二输出数据；
    所述第二实体输出所述第二输出数据。
  10. 根据权利要求9所述的方法，其特征在于，所述第二实体输出所述第二输出数据，包括：
    所述第二实体向第三通信设备发送所述第二输出数据。
  11. 根据权利要求9或10所述的方法，其特征在于，所述第二实体对所述第一输出数据进行第二处理得到第二输出数据，包括：
    所述第二实体将所述第一输出数据映射至第二输出数据。
  12. 根据权利要求9至11任一项所述的方法，其特征在于，所述第一接口还用于传输所述第一注册请求、所述第一注册响应或所述第一响应确认中的一项或多项。
  13. 根据权利要求9至12任一项所述的方法，其特征在于，所述第一接口定义了所述第一处理实体和所述第二实体交互的流程和/或信令。
  14. 根据权利要去8至13任一项所述的方法，其特征在于，所述第二实体发送所述第一注册响应的同时，启动第一响应确认计时，当所述响应确认计时到时，则向第一控制实体发送第一响应确认超时指示。
  15. 根据权利要求1至14任一项所述的方法，其特征在于，所述模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
  16. 根据权利要求1至15任一项所述的方法，其特征在于，所述数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
  17. 根据权利要求1至16任一项所述的方法，其特征在于，所述第一处理实体的输出参数包括所述第一处理实体的输出数据的类型、维度或精度中的一项或多项，所述第二实体的输入参数包括所述第二实体的输入数据的类型、维度或精度中的一项或多项。
  18. 根据权利要求1至17任一项所述的方法，其特征在于，所述第二实体的模型参数根据实时环境参数进行调整。
  19. 根据权利要求1至18任一项所述的方法，其特征在于，所述第一处理实体和第二实体属于同一个通信设备。
  20. 一种通信方法，其特征在于，
    第三处理实体向第二实体发送第二注册请求，所述第二注册请求包括第二标识，所述第 二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，所述第二实体的模型参数根据实时环境参数进行调整；
    所述第三处理实体接收所述第二实体发送的第二注册响应，所述第二注册响应指示所述第二实体的输出参数；
    该第三处理实体发送第二响应确认，所述第二响应确认指示第三处理实体的输入参数是否与第二实体的输出参数匹配。
  21. 根据权利要求20所述的方法，其特征在于，
    所述第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配，所述方法还包括：
    所述第三处理实体获取所述第二实体传输的第二输出数据，并对所述第二输出数据进行第三处理得到第三数据。
  22. 根据权利要求21所述的方法，其特征在于，
    所述第三处理实体包括输入适配模块，数据空间转换模块和输出适配模块；
    所述输入配置模块用于适配所述第二输出数据的维度与所述数据空间转换模块的输入维度；
    所述数据空间转换模块用于对所述第二输出数据进行第三处理，得到第二中间输出数据；
    所述输出适配模块用于对所述第二中间输出数据进行维度变换得到第三数据。
  23. 根据权利要求20-22任一项所述的方法，其特征在于，
    所述第三处理包括：解调和信道译码，所述第二输出数据为调制后的符号，所述第三数据为比特流；或者，
    所述第三处理包括解调、信道译码、信源解码，所述第二输出数据为调制后的符号序列，所述第三数据为符号序列。
  24. 根据权利要求20-23任一项所述的方法，其特征在于，所述第二响应确认指示第三处理实体的输入参数与第二实体的输出参数匹配失败，所述该方法还包括：
    所述第三处理实体向所述第四实体发送第二注册请求，所述第二注册请求包括第二标识，第二标识指示第三处理实体的模型信息或数据信息中的一项或多项，所述第四实体的模型参数根据实时环境参数调整；
    所述第三处理实体接收所述第四实体发送的第二注册响应，所述第二注册响应指示第四实体的输入参数；
    所述第三处理实体发送第二响应确认，所述第二响应确认指示第三处理实体的输入参数是否与第四实体的输入参数匹配。
  25. 一种通信方法，所述方法包括：
    第二实体接收第三处理实体发送的第二注册请求，第二注册请求包括第二标识，第二标识指示所述第三处理实体的模型信息或数据信息中的一项或多项，第二实体的模型参数根据实时环境参数进行调整；
    第二实体向第三处理实体发送第二注册响应，第二注册响应指示第二实体的输出参数；
    第二实体的输出参数用于确定第三处理实体的输入参数是否与第二实体的输出参数匹配。
  26. 根据权利要求25所述的方法，其特征在于，所诉第二实体接收第二响应确认，所述第二响应确认指示所述第三处理实体的输入参数与所述第二实体的输出参数匹配成功，所述方法还包括：
    所述第二实体对第一输出数据进行第二处理得到第二输出数据，第一输出数据为第一处理实体的输出数据；
    所述第二实体通过第二接口向第三处理实体传输该第二输出数据。
  27. 根据权利要求26所述的方法，其特征在于，所述第二实体对所述第一输出数据进行第二处理得到第二输出数据，包括：
    所述第二实体将第一输出数据映射至第二输出数据，其中，映射方式根据实时的环境参数调整。
  28. 根据权利要求25-27任一项所述的方法，其特征在于，
    所述第二实体在发送所述第二注册响应的同时，启动第二响应确认计时；当所述响应确认计时到时，则向第三控制实体发送第二响应确认超时指示。
  29. 根据权利要求20-28任一项所述的方法，其特征在于，所述模型信息包括以下一项或多项：模型用途、模型类型、模型规模、模型精度或模型性能。
  30. 根据权利要求20-29任一项所述的方法，其特征在于，所述数据信息包括以下一项或多项：处理数据类型、处理数据维度或处理数据精度。
  31. 根据权利要求20-30任一项所述的方法，其特征在于，
    所述第三处理实体的输入参数包括所述第三处理实体的输入数据的类型、维度或精度中的一项或多项；
    所述第二实体的输出参数包括所述第二实体的输出数据的类型、维度或精度中的一项或多项。
  32. 一种通信装置，包括用于执行权利要求1-7，5-19任一项所述的方法的模块或单元；或者包括用于执行权利要求8-19任一项所述的方法的模块或单元；或者包括用于执行权利要求20-24，29-31任一项所述的方法的模块或单元；或者包括用于执行权利要求25-31任一项所述的方法的模块或单元。
  33. 一种通信装置，其特征在于，包括处理器，所述处理器与存储器耦合，所述存储器存储指令，所述处理器用于执行所述指令，使得所述通信装置执行如权利要求1-7，5-19任一项所述的方法；或者，使得所述通信装置执行如权利要求8-19任一项所述的方法；或者，使得所述通信装置执行如权利要求20-24，29-31任一项所述的方法；或者，使得所述通信装置执行如权利要求25-31任一项所述的方法。
  34. 一种计算机可读存储介质，包括计算机程序或指令，当所述计算机程序或指令在计算机上执行时，使得权利要求1至31任一项所述的方法被执行。
  35. 一种包含指令的计算机程序产品，当其在计算机上执行时，使得权利要求1至31任一项所述的方法被执行。
  36. 一种通信系统，其特征在于，包括用于执行1-7，5-19任一项所述的方法的装置以及用于执行权利要求8-19任一项所述的方法的装置。
  37. 一种通信系统，其特征在于，包括用于执行20-24，29-31任一项所述的方法的装置以及用于执行权利要求25-31任一项所述的方法的装置。
PCT/CN2023/082648 2022-03-21 2023-03-21 一种通信方法及相关装置 WO2023179577A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210283863.9A CN116828486A (zh) 2022-03-21 2022-03-21 一种通信方法及相关装置
CN202210283863.9 2022-03-21

Publications (1)

Publication Number Publication Date
WO2023179577A1 true WO2023179577A1 (zh) 2023-09-28

Family

ID=88099954

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/082648 WO2023179577A1 (zh) 2022-03-21 2023-03-21 一种通信方法及相关装置

Country Status (2)

Country Link
CN (1) CN116828486A (zh)
WO (1) WO2023179577A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110073301A (zh) * 2017-08-02 2019-07-30 强力物联网投资组合2016有限公司 工业物联网中具有大数据集的数据收集环境下的检测方法和系统
US20210397796A1 (en) * 2020-06-18 2021-12-23 Citrix Systems, Inc. Autonomous learning of entity values in artificial intelligence conversational systems
US20220020064A1 (en) * 2019-08-29 2022-01-20 Tencent Technology (Shenzhen) Company Limited Feature processing method and apparatus for artificial intelligence recommendation model, electronic device, and storage medium
US20220019186A1 (en) * 2018-12-07 2022-01-20 Moka Mind Software Ltda Method and system for smart environment management
US20220060390A1 (en) * 2020-08-21 2022-02-24 Huawei Technologies Co., Ltd. System and methods for supporting artificial intelligence service in a network

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110073301A (zh) * 2017-08-02 2019-07-30 强力物联网投资组合2016有限公司 工业物联网中具有大数据集的数据收集环境下的检测方法和系统
US20220019186A1 (en) * 2018-12-07 2022-01-20 Moka Mind Software Ltda Method and system for smart environment management
US20220020064A1 (en) * 2019-08-29 2022-01-20 Tencent Technology (Shenzhen) Company Limited Feature processing method and apparatus for artificial intelligence recommendation model, electronic device, and storage medium
US20210397796A1 (en) * 2020-06-18 2021-12-23 Citrix Systems, Inc. Autonomous learning of entity values in artificial intelligence conversational systems
US20220060390A1 (en) * 2020-08-21 2022-02-24 Huawei Technologies Co., Ltd. System and methods for supporting artificial intelligence service in a network

Also Published As

Publication number Publication date
CN116828486A (zh) 2023-09-29

Similar Documents

Publication Publication Date Title
CN111819872A (zh) 信息传输方法、装置、通信设备及存储介质
WO2021217519A1 (zh) 用于调整神经网络的方法和装置
WO2022141397A1 (zh) 一种神经网络的训练方法以及相关装置
WO2023020502A1 (zh) 数据处理方法及装置
WO2023179577A1 (zh) 一种通信方法及相关装置
WO2019238014A1 (zh) 反馈信息的传输方法和装置
CN115835185A (zh) 一种人工智能模型下载方法、装置及系统
WO2023066111A1 (zh) 数据重传方法及相关产品
WO2023283785A1 (zh) 信号处理的方法及接收机
WO2024008004A1 (zh) 一种通信方法及装置
WO2023231881A1 (zh) 一种模型应用方法及装置
WO2023207783A1 (zh) 一种通信方法、装置及系统
CN112888076A (zh) 一种调度方法及装置
WO2024026846A1 (zh) 一种人工智能模型处理方法及相关设备
WO2023231933A1 (zh) 一种通信方法及装置
WO2023115254A1 (zh) 处理数据的方法及装置
WO2023185890A1 (zh) 一种数据处理方法及相关装置
WO2023011381A1 (zh) 数据处理方法及装置
US20230259742A1 (en) Communication method, apparatus, and system
WO2022199133A1 (zh) 压缩码本的发送方法、信道状态信息矩阵的获取方法
WO2023197950A1 (zh) 一种通信方法及相关装置
WO2024046215A1 (zh) 一种通信方法及装置
WO2022042523A1 (zh) 一种通信方法及装置
WO2023231843A1 (zh) 一种信号处理方法及装置
WO2023060503A1 (zh) 信息处理方法、装置、设备、介质、芯片、产品及程序

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: 23773824

Country of ref document: EP

Kind code of ref document: A1