Communication method and related apparatus

Abstract

Disclosed in the embodiments of the present application are a communication method and a related apparatus. The method comprises: a first communication apparatus receiving first data, wherein the first data is carried at a physical layer; and the first communication apparatus receiving second data, wherein the second data is carried at the physical layer, there is an association relationship between the first data and the second data, the first data is used for determining a truth value of model training or model monitoring of a first model, the second data is used for determining model input data of the first model, and the first model is used for processing physical layer data of the first communication apparatus. Since first data and second data are carried at a physical layer, the first data and the second data have the characteristic of a small data size, such that the processing overheads for a first communication apparatus to perform model training or model monitoring on the basis of the first data and the second data can be reduced, and the communication overheads of the first communication apparatus can also be reduced.

Description

Communication method and related apparatus

[0001] The present application claims priority from the Chinese patent application No. 202411095373.1 filed on August 9, 2024, and entitled "A communication method and related apparatus", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of communication, and in particular, to a communication method and related apparatus. BACKGROUND

[0003] With the development of communication technology, in a communication system, in addition to the traditional communication service, the service performed by the communication device can also include other new services, such as artificial intelligence (AI) services. Generally, a wireless communication system capable of processing AI services can also be referred to as an air interface AI system.

[0004] Currently, a communication device can be used as a participating node of the air interface AI system, and the computing power of the communication device is applied to a certain link of the air interface AI system. Generally, the AI function introduced in the wireless communication network needs to rely on a model to be implemented. For example, the communication device can process the input data through the model to obtain the output data. In order to implement the training or monitoring of the model, the transmitter (also referred to as the sending end) and the receiver (also referred to as the receiving end) can interact with the ground truth, which can be used as a training set, a validation set and / or a test set.

[0005] However, the current transmitter and receiver interact with the ground truth through the application layer, and the application layer data has the characteristic of large data volume, so there is a problem of large communication overhead and processing overhead. SUMMARY

[0006] Embodiments of the present application propose a communication method, a first communication device receives first data and second data carried in a physical layer, and performs model training or model monitoring on a first model based on the first data and the second data, so as to save the communication overhead and the processing overhead.

[0007] In a first aspect, embodiments of the present application propose a communication method, which is applied to a first communication device.

[0008] In a possible implementation, the first communication apparatus is a terminal device, which can be the terminal device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, chip system, functional module, control unit, circuit, processor, or integrated circuit in the foregoing devices or apparatus, and the like, without limitation.

[0009] In another possible implementation, the first communication apparatus is an access network device, which can be the access network device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, chip system, module, control unit, circuit, or processor in the foregoing devices or apparatus, or a centralized unit (CU) and / or distributed unit (DU), without limitation.

[0010] The method comprises: receiving, by the first communication apparatus, first data, the first data being carried in a physical layer; receiving, by the first communication apparatus, second data, the second data being carried in the physical layer, the first data and the second data having an association relationship, the first data being used to determine a true value of model training or model monitoring of a first model, and the second data being used to determine model input data of the first model, the first model being used to process physical layer data of the first communication apparatus.

[0011] Exemplarily, the first data is carried in the physical layer, in other words, the first data is a group of data carried in the physical layer and transmitted in a bit stream form.

[0012] Exemplarily, the second data is carried in the physical layer, in other words, the second data is a group of data carried in the physical layer and transmitted in a bit stream form.

[0013] It should be noted that the first data can be directly used as the true value of the model training or model monitoring of the first model, or the first data can be processed and then used as the true value of the model training or model monitoring of the first model. The second data can be directly input to the model input data of the first model, or the second data can be processed and then input to the first model.

[0014] In the technical solution, the first communication device receives the first data and the second data carried in the physical layer, and the first data and the second data are used for model training or model monitoring of the first model. Since the first data and the second data are carried in the physical layer, the physical layer data has the characteristics of small granularity, small data volume and simple processing, so that the processing overhead of the first communication device for training or monitoring the first model can be reduced, and the communication overhead of the first communication device can also be reduced. The first data and the second data of the present application are obtained from the physical layer and do not need to be processed by other transmission layers, and can be directly used for training or monitoring of the first model. The processing overhead of the first communication device can be saved.

[0015] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: performing, by the first communication device, model training or model monitoring on the first model according to the first data and the second data.

[0016] With reference to the first aspect, in a possible implementation manner of the first aspect, the first data is received by receiving transmission data from the second communication device, the transmission data is determined based on user plane data and / or control plane data, and the first data is determined based on the transmission data.

[0017] In other words, the first data multiplexes part of the transmission data. The user plane data is, for example, service-related data, and the control plane data is, for example, signaling-related data.

[0018] In the technical solution, the first data can multiplex the transmission data, so that the second communication device does not need to additionally send the true value, and the current data transmission process can be used to realize the true value transmission, thereby effectively saving the communication overhead.

[0019] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: receiving first information, the first information being used to indicate that the first data and the second data have an association relationship.

[0020] In a possible implementation manner, the first information is further used to indicate the model training or the model monitoring of the first model.

[0021] In another possible implementation manner, the first information is further used to indicate that the first data and the second data are used for the model training or the model monitoring of the first model.

[0022] In another possible implementation manner, the first information is further used to indicate that the first data is used to determine the true value of the model training or the model monitoring of the first model.

[0023] The first information is further used to indicate that the second data is used to determine the model input data of the first model.

[0024] In a possible implementation manner of the first aspect, the first information comprises one or more of the following: first sub-information, the first sub-information being used for indicating one or more sets of association relationships, the association relationship comprising: an association relationship between data used for determining the model input data of the first model and data used for determining the true value; or second sub-information, the second sub-information being used for indicating a target association relationship, the target association relationship being that the first data and the second data have an association relationship, and the target association relationship belonging to the one or more sets of association relationships.

[0025] In a possible implementation manner of the first aspect, the first information further comprises one or more of the following: identification information of a data set, the data set comprising the first data and / or the second data; identification information of the first model; identification information of a second model, the second model being a model used by the second communication apparatus, and the second model being used for generating the second data; or identification information of a model set, the model set comprising one or more models, and the one or more models comprising the first model.

[0026] For example, the first data and the second data of multiple interactions between the first communication apparatus and the second communication apparatus are a data set. Then, the first communication apparatus uses data of the data set to uniformly perform model training or model monitoring of the first model. At this time, when the second communication apparatus sends the first data and the second data to the first communication apparatus, the second communication apparatus also needs to send identification information of the data set.

[0027] In the technical solution, the first communication apparatus can be explicitly indicated by the first information that the first data and the second data have an association relationship, so as to facilitate the first communication apparatus to subsequently perform model training or model monitoring of the first model based on the first data and the second data, save processing overhead of the first communication apparatus, and improve model training or model monitoring efficiency.

[0028] In a possible implementation manner of the first aspect, the first information comprises one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data.

[0029] In the technical solution, the second communication apparatus can also indicate, by the first information, a transmission resource carrying the first data, so as to facilitate the first communication apparatus to determine the first data according to the first information. In addition, the first information can also indicate a transmission resource carrying the second data, so as to facilitate the first communication apparatus to determine, according to the transmission resource carrying the first data and the transmission resource carrying the second data, that the data carried by the two parts of the transmission resources have an association relationship, and improve implementation flexibility of the solution.

[0030] With reference to the first aspect, in a possible implementation of the first aspect, the method further includes: receiving second information, the second information being used to indicate one or more of the following resources: frequency domain resources carrying the second data, time domain resources carrying the second data, or spatial domain resources carrying the second data.

[0031] For example, the frequency domain resources carrying the second data include: a frequency band, a subcarrier, a resource element (RE), or an RE pattern, where the RE pattern includes one or more REs; the time domain resources carrying the second data include: a slot, a symbol, or a subframe; and the spatial domain resources carrying the second data include: an antenna port.

[0032] In the above technical solution, the time domain, frequency domain, and spatial domain resources of the second data can be explicitly indicated by the second information, so that the first communication device can quickly determine the second data, the processing overhead of the first communication device is saved, and the model training or model monitoring efficiency is improved.

[0033] With reference to the first aspect, in a possible implementation of the first aspect, the method further includes: receiving third information, the third information being used to indicate a relative position of the first data in the transmission data.

[0034] In the above technical solution, when the first data is multiplexed in the transmission data, the second communication device can indicate the relative position of the first data in the transmission data by using the third information, so that the first communication device can determine the first data according to the third information. The first communication device can quickly determine the first data, the processing overhead of the first communication device is saved, and the model training or model monitoring efficiency is improved.

[0035] With reference to the first aspect, in a possible implementation of the first aspect, the third information includes one or more of the following information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data.

[0036] With reference to the first aspect, in a possible implementation of the first aspect, the first data further includes a first reference signal, and the first reference signal is used for channel estimation of a wireless resource carrying the first data; and the second data further includes a second reference signal, and the second reference signal is used for channel estimation of a wireless resource carrying the second data by the first model.

[0037] With reference to the first aspect, in a possible implementation of the first aspect, the first information is further used to indicate that the first reference signal and the second reference signal have an association relationship.

[0038] In the technical solution, the first information can explicitly indicate that the first reference signal and the second reference signal have the association relationship, so as to facilitate the first communication device to subsequently perform model training or model monitoring of the first model based on the first reference signal and the second reference signal, save processing overhead of the first communication device, and improve model training or model monitoring efficiency.

[0039] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: receiving fourth information, the fourth information being used to indicate that the first communication device feeds back gradient information related to the first model; and transmitting the gradient information according to the fourth information.

[0040] In a possible implementation manner, the fourth information is further used to indicate first data and second data on which the gradient information is based.

[0041] In another possible implementation manner, the fourth information includes one or more of the following information:

[0042] Time domain resources carrying the gradient information, or an association relationship between the time domain resources carrying the gradient information and time domain resources of the second data on which the gradient information is based.

[0043] In another possible implementation manner, the gradient information is generated according to one or more groups of the first data and the second data.

[0044] In the technical solution, the first communication device can implement reverse update of the gradient information, and improve training effect of the second communication device on the second model under the condition of low communication overhead.

[0045] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: receiving fifth information, the fifth information being used to configure an effective condition of the first data and the second data, and the effective condition including one or more of the following: successful execution of channel decoding on the transmission data; successful demodulation of the transmission data; or a condition of performing channel estimation on the first data, wherein the condition of performing channel estimation includes one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), average delay, Doppler spread, Doppler shift, or delay spread.

[0046] In the technical solution, the first communication device can determine the effective condition of the first data and the second data according to the fifth information, and use the first data and the second data to perform model training or model monitoring of the first model in the case where the effective condition is met, thereby improving model training or model monitoring effect.

[0047] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: performing first processing on the first data to generate third data; performing second processing on the second data using the first model to generate fourth data, the third data being a true value of the fourth data; and determining a loss value of the first model based on the third data and the fourth data.

[0048] In a possible implementation manner, the first processing is non-artificial intelligence (AI) processing, and the first processing includes any one or more of the following processing: channel decoding, demodulation, or channel estimation based on the first reference signal; and the second processing is AI processing, and the second processing includes any one or more of the following processing: AI channel decoding, AI demodulation, or AI channel estimation based on the second reference signal.

[0049] In a possible implementation manner, the first reference signal is a reference signal determined in a non-artificial intelligence (AI) processing manner; and the second reference signal is a reference signal determined in an AI processing manner.

[0050] With reference to the first aspect, in a possible implementation manner of the first aspect, the method further includes: sending capability information of the first communication device, the capability information of the first communication device indicating one or more of the following information: whether the first communication device supports model training or model monitoring based on the first data and the second data; frequency domain resources reserved by the first communication device for the second data; time domain resource requirements of the first communication device for the first data and the second data; accuracy of channel estimation performed by the first communication device based on the first reference signal; conditions of channel estimation performed by the first communication device based on the first reference signal, the conditions including one or more of the following: signal-to-noise ratio, time delay spread requirement, average time delay requirement, Doppler shift requirement, or Doppler spread requirement; data format of input data and / or output data supported by the first communication device; whether the first communication device supports feedback of gradient information related to the first model; time length of updating, by the first communication device, the gradient information related to the first model; or time length of feedback, by the first communication device, of the gradient information related to the first model.

[0051] In the above technical solution, the first communication device can further report its capability information, so as to save unnecessary communication overhead.

[0052] In a possible implementation manner of the first aspect, the method further includes: sending capability information of the first communication device, the capability information of the first communication device indicating one or more of the following information: whether the first communication device supports model training or model monitoring based on the first data and the second data; frequency domain resources reserved by the first communication device for the second data; time domain resource requirements of the first communication device for the first data and the second data; accuracy of channel estimation performed by the first communication device based on the first reference signal; conditions of channel estimation performed by the first communication device based on the first reference signal, the conditions including one or more of the following: signal-to-noise ratio, time delay spread requirement, average time delay requirement, Doppler shift requirement, or Doppler spread requirement; data format of input data and / or output data supported by the first communication device; whether the first communication device supports feedback of gradient information related to the first model; time length of updating, by the first communication device, the gradient information related to the first model; or time length of feedback, by the first communication device, of the gradient information related to the first model.

[0051] In the above technical solution, the first communication device can further report its capability information, so as to save unnecessary communication overhead.

[0052] The second aspect, the embodiments of the present application propose a communication method, the method is applied to the second communication device.

[0053] In a possible implementation, the second communication apparatus is a terminal device, which can be the terminal device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a functional module, a control unit, a circuit, a processor, or an integrated circuit in the foregoing devices or apparatuses, and the like, without any limitation in particular.

[0054] In another possible implementation, the second communication apparatus is an access network device, which can be the access network device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a module, a control unit, a circuit, or a processor in the foregoing devices or apparatuses, or a centralized unit (CU) and / or a distributed unit (DU), without any limitation in particular.

[0055] The method comprises: sending first data to the first communication apparatus, the first data being carried in a physical layer; and sending second data to the first communication apparatus, the second data being carried in the physical layer, the first data and the second data having an association relationship, the first data being used to determine a true value of model training or model monitoring of a first model, and the second data being used to determine model input data of the first model, the first model being a model of the first communication apparatus.

[0056] For example, the first data is carried in the physical layer, that is, the first data is a group of data carried in the physical layer and transmitted in a bit stream form.

[0057] For example, the second data is carried in the physical layer, that is, the second data is a group of data carried in the physical layer and transmitted in a bit stream form.

[0058] It should be noted that the first data can be directly used as the true value of the model training or the model monitoring of the first model, or the first data can be processed and then used as the true value of the model training or the model monitoring of the first model. The second data can be directly input to the model input data of the first model, or the second data can be processed and then input to the first model.

[0059] In the foregoing technical solution, the first communication apparatus receives the first data and the second data carried in the physical layer, and the first data and the second data are used for model training or model monitoring of the first model. Since the first data and the second data are carried in the physical layer, the physical layer data has the characteristics of small granularity, small data volume, and simple processing, so that the processing overhead of the first communication apparatus for training or monitoring the first model can be reduced, and the communication overhead of the first communication apparatus can also be reduced. The first data and the second data of the present application are obtained from the physical layer and do not need to be processed through other transmission layers, and can be directly used for training or monitoring of the first model. The processing overhead of the first communication apparatus can be saved.

[0060] With reference to the second aspect, in a possible implementation of the second aspect, the sending the first data to the first communication device comprises: sending transmission data to the first communication device, the transmission data being determined based on the user plane data and / or the control plane data, and the first data being determined based on the transmission data.

[0061] In other words, the first data multiplexes the transmission data. The user plane data is, for example, service-related data, and the control plane data is, for example, signaling-related data.

[0062] Optionally, the second communication device can also perform bit flipping on the first data, and then send the flipped first data as part of the transmission data to the first communication device.

[0063] In the above technical solution, the first data can multiplex the transmission data, so that the second communication device does not need to send additional true values, and the current data transmission process can be used to implement true value transmission, which can effectively save communication overhead.

[0064] With reference to the second aspect, in a possible implementation of the second aspect, the method further comprises: sending first information to the first communication device, the first information being used to indicate that the first data and the second data have an association relationship.

[0065] In a possible implementation, the first information is also used to indicate model training or model monitoring of the first model.

[0066] In another possible implementation, the first information is also used to indicate that the first data and the second data are used for model training or model monitoring of the first model.

[0067] In another possible implementation, the first information is also used to indicate that the first data is used to determine a true value of model training or model monitoring of the first model.

[0068] The first information is also used to indicate that the second data is used to determine model input data of the first model.

[0069] In another possible implementation, the first information comprises one or more of the following information: first sub-information, the first sub-information being used to indicate one or more association relationships, the association relationship comprising an association relationship between data used to determine model input data of the first model and data used to determine a true value; or second sub-information, the second sub-information being used to indicate a target association relationship, the target association relationship being that the first data and the second data have an association relationship, and the target association relationship belonging to one or more association relationships.

[0070] In a possible implementation of the second aspect, the first information further includes one or more of the following: identification information of a data set including the first data and / or the second data; identification information of the first model; identification information of a second model used by the second communication apparatus, the second model being used to generate the second data; or identification information of a model set including one or more models, the one or more models including the first model.

[0071] For example, the first data and the second data of multiple interactions between the first communication apparatus and the second communication apparatus are a data set. Then, the first communication apparatus uses data of the data set to uniformly perform model training or model monitoring of the first model. At this time, when the second communication apparatus sends the first data and the second data to the first communication apparatus, the second communication apparatus also needs to send identification information of the data set.

[0072] In the above technical solution, the first communication apparatus can be explicitly indicated by the first information that the first data and the second data have a correlation relationship, so as to facilitate the first communication apparatus to subsequently perform model training or model monitoring of the first model based on the first data and the second data, save processing overhead of the first communication apparatus, and improve model training or model monitoring efficiency.

[0073] In combination with the second aspect, in a possible implementation of the second aspect, the first information includes one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data.

[0074] In the above technical solution, the second communication apparatus can also indicate, by the first information, a transmission resource carrying the first data, so as to facilitate the first communication apparatus to determine the first data according to the first information. In addition, the first information can also indicate a transmission resource carrying the second data, so as to facilitate the first communication apparatus to determine, according to the transmission resource carrying the first data and the transmission resource carrying the second data, that the data carried by the two parts of the transmission resources have a correlation relationship, and improve implementation flexibility of the solution.

[0075] In combination with the second aspect, in a possible implementation of the second aspect, the method further includes: sending, to the first communication apparatus, second information, the second information being used to indicate one or more of the following resources: a frequency domain resource carrying the second data, a time domain resource carrying the second data, or a space domain resource carrying the second resource.

[0076] For example, the frequency domain resource carrying the second data includes a frequency band, a subcarrier, a resource element (RE), or an RE pattern, where the RE pattern includes one or more REs; the time domain resource carrying the second data includes a slot, a symbol, or a subframe; and the space domain resource carrying the second data includes an antenna port.

[0077] In the technical solution described above, the time domain, frequency domain, and space domain resources of the second data can also be explicitly indicated by the second information, so that the first communication device can quickly determine the second data, the processing overhead of the first communication device is saved, and the model training or model monitoring efficiency is improved.

[0078] With reference to the second aspect, in a possible implementation manner of the second aspect, the method further includes: sending, to the first communication device, third information, where the third information is used to indicate a relative position of the first data in the transmission data.

[0079] In the technical solution described above, when the first data is multiplexed in the transmission data, the second communication device can also indicate the relative position of the first data in the transmission data by using the third information, so that the first communication device can determine the first data according to the third information. This facilitates the first communication device to quickly determine the first data, saves the processing overhead of the first communication device, and improves the model training or model monitoring efficiency.

[0080] With reference to the second aspect, in a possible implementation manner of the second aspect, the third information includes one or more of the following pieces of information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data.

[0081] With reference to the second aspect, in a possible implementation manner of the second aspect, the first data further includes a first reference signal, and the first reference signal is used for channel estimation of a wireless resource carrying the first data; and the second data further includes a second reference signal, and the second reference signal is used for channel estimation of a wireless resource carrying the second data by the first model.

[0082] With reference to the second aspect, in a possible implementation manner of the second aspect, the first information is further used to indicate that the first reference signal and the second reference signal have an association relationship.

[0083] In the technical solution described above, the first communication device can subsequently perform model training or model monitoring of the first model based on the first reference signal and the second reference signal, the processing overhead of the first communication device is saved, and the model training or model monitoring efficiency is improved.

[0084] In a possible implementation manner of the second aspect, the method further includes: sending fourth information to the first communication device, the fourth information being used to indicate that the first communication device feeds back gradient information related to the first model; and sending the gradient information according to the fourth information.

[0085] In a possible implementation manner, the fourth information is further used to indicate the first data and the second data on which the gradient information is based.

[0086] In another possible implementation manner, the fourth information includes one or more of the following information:

[0087] The time domain resource carrying the gradient information, or an association relationship between the time domain resource carrying the gradient information and the time domain resource of the second data on which the gradient information is based.

[0088] In another possible implementation manner, the gradient information is generated according to one or more groups of the first data and the second data.

[0089] In the above technical solution, the first communication device can realize reverse updating of the gradient information, and the training effect of the second model by the second communication device is improved under the condition of lower communication overhead.

[0090] In a possible implementation manner of the second aspect, the method further includes: sending fifth information to the first communication device, the fifth information being used to configure an effective condition of the first data and the second data, and the effective condition including one or more of the following: successful execution of channel decoding on the transmission data; successful demodulation of the transmission data; or a condition of performing channel estimation on the first data, wherein the condition of performing channel estimation includes one or more of the following: signal-to-noise ratio, signal-to-interference-plus-noise ratio, average delay, Doppler spread, Doppler shift, or delay spread.

[0091] In the above technical solution, the first communication device can determine the effective condition of the first data and the second data according to the fifth information, and use the first data and the second data to perform model training or model monitoring of the first model in the case where the effective condition is met, thereby improving the effect of model training or model monitoring.

[0092] With reference to the second aspect, in a possible implementation manner of the second aspect, the method further includes: receiving capability information of the first communication apparatus, the capability information of the first communication apparatus indicating one or more of the following: whether the first communication apparatus supports model training or model monitoring based on the first data and the second data; frequency domain resources reserved by the first communication apparatus for the second data; time domain resource requirements of the first communication apparatus for the first data and the second data; accuracy of channel estimation performed by the first communication apparatus based on the first reference signal; conditions of channel estimation performed by the first communication apparatus based on the first reference signal, the conditions including one or more of the following: signal-to-noise ratio, time delay spread requirement, average time delay requirement, Doppler shift requirement, or Doppler spread requirement; data format of input data and / or output data supported by the first communication apparatus; whether the first communication apparatus supports feedback of gradient information related to the first model; time length of updating, by the first communication apparatus, the gradient information related to the first model; or time length of feedback, by the first communication apparatus, of the gradient information related to the first model.

[0093] In the above technical solution, the first communication apparatus can also report its own capability information, so as to save unnecessary communication overhead.

[0094] With reference to the second aspect, in a possible implementation manner of the second aspect, the method further includes: performing third processing on the sixth data to generate the first data; determining the fifth data from the sixth data, the fifth data belonging to the sixth data; and performing fourth processing on the fifth data to generate the second data.

[0095] In a possible implementation manner, the third processing is non-artificial intelligence (AI) processing, and the third processing includes any one or more of the following processing: channel coding, modulation, generation of the first reference signal, or resource mapping of the first reference signal; and the fourth processing is AI processing, and the fourth processing includes any one or more of the following processing: AI channel coding, AI modulation, generation of the second reference signal, or AI resource mapping of the second reference signal.

[0096] In a third aspect, an embodiment of the present application provides a communication method, the method being applied to a first communication apparatus.

[0097] In a possible implementation manner, the first communication apparatus is a terminal device, which can be a terminal device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a functional module, a control unit, a circuit, a processor, or an integrated circuit in the foregoing devices or apparatus, without limitation.

[0098] In an alternative possible implementation form of the first communication device, the first communication device is an access network device, which can be the access network device itself, a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a module, a control unit, a circuit, or a processor in the aforementioned devices or apparatuses, or a centralized unit (CU) and / or a distributed unit (DU), without limitation.

[0099] The method comprises: receiving fourth information, the fourth information being used to indicate that the first communication device feeds back gradient information related to the first model; and transmitting the gradient information related to the first model according to the fourth information. The first model is used to perform a second process, the second process being an AI process, and the second process comprising any one or more of the following processes: AI channel decoding, AI demodulation, or AI channel estimation based on a second reference signal, the second reference signal being an AI reference signal.

[0100] In the above technical solution, the first communication device can implement backward updating of gradient information, and the training effect of the second communication device on the second model is improved under the condition of low communication overhead.

[0101] Optionally, the fourth information is also used to indicate first data and second data on which the gradient information is based, the first data and the second data having a correlation relationship. The first data is used to determine a true value of model training or model monitoring of the first model, and the second data is used to determine model input data of the first model.

[0102] In a possible implementation form of the third aspect, the first data is carried in a physical layer, and the second data is carried in the physical layer, and the first model is used to process physical layer data of the first communication device. For example, the first data is carried in the physical layer, in other words, the first data is a set of data carried in the physical layer and transmitted in the form of a bit stream. For example, the second data is carried in the physical layer, in other words, the second data is a set of data carried in the physical layer and transmitted in the form of a bit stream.

[0103] In an alternative possible implementation form of the third aspect, the first data is carried in a higher layer, for example, an application layer, and the second data is carried in the higher layer, for example, the application layer.

[0104] In a possible implementation form of the third aspect, the fourth information comprises one or more of the following information: time domain resources carrying the gradient information, or a correlation relationship between the time domain resources carrying the gradient information and time domain resources of the second data on which the gradient information is based.

[0105] In a possible implementation form of the third aspect, the gradient information is generated according to one or more sets of the first data and the second data.

[0106] With reference to the third aspect, in a possible implementation manner of the third aspect, the method further includes: receiving the first data; receiving second data, the first data and the second data having a correlation relationship, the first data being used to determine a true value of model training or model monitoring of the first model, and the second data being used to determine model input data of the first model; and processing the first data and the second data according to the first model to generate gradient information related to the first model.

[0107] With reference to the third aspect, in a possible implementation manner of the third aspect, the first data is borne in a physical layer; the second data is borne in the physical layer, and the first model is used to process physical layer data of the first communication device.

[0108] In the fourth aspect, an embodiment of the present application provides a communication method, which is applied to a second communication device.

[0109] In a possible implementation manner, the second communication device is a terminal device, and the second communication device can be the terminal device itself, or a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a functional module, a control unit, a circuit, a processor, or an integrated circuit in the foregoing device or apparatus, without limitation.

[0110] In another possible implementation manner, the second communication device is an access network device, and the second communication device can be the access network device itself, or a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a module, a control unit, a circuit, or a processor in the foregoing device or apparatus, or a centralized unit (CU) and / or a distributed unit (DU), without limitation.

[0111] The method includes: sending fourth information, the fourth information being used to indicate that the first communication device feeds back gradient information related to the first model; and receiving the gradient information related to the first model from the first communication device. The first model is used to perform second processing, the second processing being AI processing, and the second processing includes any one or more of the following processing: AI channel decoding, AI demodulation, or AI channel estimation based on a second reference signal, the second reference signal being an AI reference signal.

[0112] In the foregoing technical solution, the first communication device can implement reverse update of gradient information, and the training effect of the second communication device on the second model is improved under the condition of low communication overhead.

[0113] Optionally, the fourth information further indicates first data and second data based on which the gradient information is generated, the first data and the second data having a correlation relationship. The first data is used to determine a true value of model training or model monitoring of the first model, and the second data is used to determine model input data of the first model.

[0114] In a possible implementation, the first data is carried in a physical layer, the second data is carried in a physical layer, and the first model is used to process physical layer data of the first communication device. For example, the first data is carried in a physical layer, in other words, the first data is a set of data carried in a physical layer and transmitted in a bit stream form. For example, the second data is carried in a physical layer, in other words, the second data is a set of data carried in a physical layer and transmitted in a bit stream form.

[0115] In another possible implementation, the first data is carried in a high layer, for example, an application layer; and the second data is carried in a high layer, for example, an application layer.

[0116] With reference to the fourth aspect, in a possible implementation of the fourth aspect, the fourth information includes one or more of the following information: a time domain resource carrying the gradient information, or a correlation relationship between the time domain resource carrying the gradient information and a time domain resource of the second data based on which the gradient information is generated.

[0117] With reference to the fourth aspect, in a possible implementation of the fourth aspect, the gradient information is generated according to one or more sets of the first data and the second data.

[0118] With reference to the fourth aspect, in a possible implementation of the fourth aspect, the method further includes: transmitting the first data; and transmitting the second data, the first data and the second data having a correlation relationship, the first data being used to determine a true value of model training or model monitoring of the first model, and the second data being used to determine model input data of the first model.

[0119] With reference to the fourth aspect, in a possible implementation of the fourth aspect, the first data is carried in a physical layer; the second data is carried in a physical layer, and the first model is used to process physical layer data of the first communication device.

[0120] In the fifth aspect, the fifth aspect of the present application provides a communication device, which is a first communication device or a second communication device, and includes a transceiver module and a processing module. The constituent modules of the communication device can also be used to perform the steps performed in each possible implementation of the first aspect, the second aspect, the third aspect, or the fourth aspect, and achieve the corresponding technical effects. For details, refer to the first aspect, the second aspect, the third aspect, or the fourth aspect, which will not be described here.

[0121] In a sixth aspect, the sixth aspect of the present application provides a communication apparatus, comprising at least one processor, wherein the at least one processor is coupled with a memory; the memory is configured to store programs or instructions; and the at least one processor is configured to execute the programs or instructions, so that the apparatus implements the method in any possible implementation manner of any one of the first aspect, the second aspect, the third aspect or the fourth aspect. Optionally, the communication apparatus can comprise the memory.

[0122] In a seventh aspect, the seventh aspect of the present application provides a communication apparatus, comprising at least one logic circuit and an input / output interface; and the logic circuit is configured to execute the method in any possible implementation manner of any one of the first aspect, the second aspect, the third aspect or the fourth aspect.

[0123] In an eighth aspect, the eighth aspect of the present application provides a communication system, comprising the first communication apparatus and / or the second communication apparatus.

[0124] In a ninth aspect, the ninth aspect of the present application provides a computer readable storage medium, configured to store one or more computer-executable instructions, when the computer-executable instructions are executed by a processor, the processor executes the method in any possible implementation manner of any one of the first aspect, the second aspect, the third aspect or the fourth aspect.

[0125] In a tenth aspect, the tenth aspect of the present application provides a computer program product (or computer program), when a computer program in the computer program product is executed by a processor, the processor executes the method in any possible implementation manner of any one of the first aspect, the second aspect, the third aspect or the fourth aspect.

[0126] In an eleventh aspect, the eleventh aspect of the present application provides a chip or chip system, comprising at least one processor, configured to support the communication apparatus to implement the method in any possible implementation manner of any one of the first aspect, the second aspect, the third aspect or the fourth aspect.

[0127] In a possible design, the chip or chip system can further comprise a memory, configured to store necessary programs and data of the communication apparatus. The chip system can be composed of a chip, or can comprise a chip and other discrete devices. Optionally, the chip system further comprises an interface circuit, configured to provide programs and / or data for the at least one processor.

[0128] The technical effects brought by the fifth aspect to the eleventh aspect can be referred to the technical effects brought by the first aspect, the second aspect, the third aspect or the fourth aspect, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0129] FIG. 1a to FIG. 1e are schematic diagrams of AI processing procedures involved in the present application;

[0130] FIG. 2a is a schematic diagram of an OSI model;

[0131] FIG. 2b is a schematic diagram of a conventional receiver;

[0132] FIG. 2c is a schematic diagram of an AI receiver;

[0133] FIG. 2d is a schematic diagram of model training of an AI receiver;

[0134] FIG. 3a is a schematic diagram of a scenario of a conventional transmitter and receiver;

[0135] FIG. 3b is a schematic diagram of a scenario of an AI transmitter and AI receiver;

[0136] FIG. 4a is a schematic diagram of a model training method in an embodiment of the present application;

[0137] FIG. 4b is a schematic diagram of another model training method in an embodiment of the present application;

[0138] FIG. 4c is a schematic diagram of another model training method in an embodiment of the present application;

[0139] FIG. 4d is a schematic diagram of an architecture of a communication system 1000 applied in an embodiment of the present application;

[0140] FIG. 5 is a schematic diagram of an embodiment of a communication method in an embodiment of the present application;

[0141] FIG. 6 is a schematic diagram of data processing involved in a communication method in an embodiment of the present application;

[0142] FIG. 7 is a schematic diagram of first data in an embodiment of the present application;

[0143] FIG. 8 is a schematic diagram of an application scenario in an embodiment of the present application;

[0144] FIG. 9 is a schematic diagram of another embodiment of a communication method in an embodiment of the present application;

[0145] FIG. 10 is a schematic diagram of an application scenario in an embodiment of the present application;

[0146] FIG. 11 is a schematic diagram of a data processing method in an embodiment of the present application;

[0147] FIG. 12 is a schematic diagram of another model training in an embodiment of the present application;

[0148] FIG. 13 is a schematic diagram of another application scenario in an embodiment of the present application;

[0149] FIG. 14 is a schematic diagram of one scenario of data processing in an embodiment of the present application;

[0150] FIG. 15 is a schematic diagram of one model training in an embodiment of the present application;

[0151] FIG. 16 is a schematic diagram of another application scenario in an embodiment of the present application;

[0152] FIG. 17 is a schematic diagram of one model training in an embodiment of the present application;

[0153] FIG. 18 is a schematic diagram of an embodiment of another communication method in an embodiment of the present application;

[0154] FIG. 19 is a schematic diagram of another application scenario in an embodiment of the present application;

[0155] FIG. 20 is a schematic diagram of a scenario of another communication method in an embodiment of the present application;

[0156] FIG. 21 is a schematic diagram of one structure of a communication apparatus in an embodiment of the present application;

[0157] FIG. 22 is a schematic diagram of another structure of a communication apparatus in an embodiment of the present application;

[0158] FIG. 23 is a schematic diagram of another structure of a communication apparatus in an embodiment of the present application. DETAILED DESCRIPTION

[0159] The terms "first", "second", "third", "fourth" and the like in the description and in the claims of the present application, if any, are used for distinguishing between similar objects talking about the embodiments and do not necessarily have a particular chronological, spatial or logical order. It is to be understood that the terms so used are interchangeable under appropriate circumstances and embodiments of the application might operate in other than the particular sequences, or according to other sequences, except where specifically stated to the contrary. Moreover, the terms "comprise", "comprising", "include", "including", and the like, as used herein, are specifically intended to be construed as open-ended terms i.e., the terms do not exclude additional, unrecited elements but rather, the terms are merely to be construed as specifying the presence of the stated features, integers, steps or components as being included, but do not preclude the presence or addition of one or more other features, integers, steps, components or groups thereof.

[0160] It should be understood that the term "and / or" in this document is merely used to describe associated objects, and can represent three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist simultaneously, and B exists alone, where A and B can be single or multiple. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects. In addition, "at least one of the following" or similar expressions in this document are used to represent any combination of the listed items; for example, at least one of A, B and (or) C can represent the following six cases: A exists alone, B exists alone, C exists alone, A and B exist simultaneously, B and C exist simultaneously, and A and C exist simultaneously, where A, B and C can be single or multiple.

[0161] First, some terms in the embodiments of the present application are explained and described to facilitate understanding by those skilled in the art.

[0162] (1) Configuration and pre-configuration: In this application, configuration and pre-configuration will be used together. Configuration means that the access network device sends some parameter configuration information or parameter values to the terminal device through messages or signaling, so that the terminal device determines the communication parameters or transmission resources according to these values or information. Pre-configuration is similar to configuration, which can be parameter information or parameter values agreed by the access network device and the terminal device in advance, or parameter information or parameter values adopted by the access network device or the terminal device according to standard protocols, or parameter information or parameter values pre-stored in the access network device or the terminal device. The present application does not limit this.

[0163] Further, these values and parameters can be changed or updated.

[0164] (2) The terms "system" and "network" in the embodiments of the present application can be used interchangeably. "At least one" means one or more, and "multiple" means two or more. "And / or" describes the association between the associated objects, and means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist simultaneously, and B exists alone, where A and B can be single or multiple. The character " / " generally represents an "or" relationship between the associated objects. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, "at least one of A, B and C" includes A, B, C, AB, AC, BC or ABC. In addition, unless otherwise specified, the ordinal numbers "first", "second", etc. mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the order, time sequence, priority or importance of the multiple objects.

[0165] (3) In the embodiments of the present application, “sending” and “receiving” represent the direction of signal transmission. For example, “sending information to XX” can be understood as that the destination of the information is XX, which can include direct transmission through the air interface, or indirect transmission through the air interface by other units or modules. “Receiving information from YY” can be understood as that the source of the information is YY, which can include direct reception from YY through the air interface, or indirect reception from YY through the air interface by other units or modules. “Sending” can also be understood as “output” of a chip interface, and “receiving” can also be understood as “input” of a chip interface.

[0166] In other words, sending and receiving can be between devices, such as between an access network device and a terminal device, or within a device, such as between components, modules, chips, software modules or hardware modules within a device through a bus, wire or interface.

[0167] It can be understood that the information may be processed as necessary between the source and the destination, such as encoding and modulation, but the destination can understand the valid information from the source. Similar expressions in the present application can be similarly understood and will not be repeated here.

[0168] (4) Artificial Intelligence (AI).

[0169] AI can enable machines to have human intelligence, such as enabling machines to apply computer hardware and software to simulate certain intelligent behaviors of humans. To achieve artificial intelligence, a machine learning method can be used. In the machine learning method, the machine learns (or trains) a neural network model using training data. The neural network model can also be referred to as an AI model, an AI large model, a large language model (LLM), or a model. The model represents the mapping between the input and the output. The learned model can be used for inference (or prediction), i.e., the model can be used to predict the output corresponding to a given input. The output can also be referred to as an inference result (or prediction result).

[0170] Machine learning can include supervised learning, unsupervised learning, and reinforcement learning. Unsupervised learning can also be referred to as non-supervised learning.

[0171] Supervised learning learns the mapping relationship from sample values to sample labels according to the collected sample values and sample labels, and uses an AI model to express the learned mapping relationship. The process of training a machine learning model is the process of learning this mapping relationship. In the training process, the sample values are input into the model to obtain the predicted values of the model, and the model parameters are optimized by calculating the error between the predicted values of the model and the sample labels (ideal values). After the mapping relationship is learned, the learned mapping can be used to predict new sample labels. The learned mapping relationship of supervised learning can include linear mapping or nonlinear mapping. According to the type of label, the learned task can be divided into classification task and regression task.

[0172] Unsupervised learning uses algorithms to discover the internal patterns of samples according to the collected sample values. In unsupervised learning, a class of algorithms uses the sample itself as a supervision signal, that is, the model learns the mapping relationship from the sample to the sample, which is called self-supervised learning. In training, the model parameters are optimized by calculating the error between the predicted values of the model and the sample itself. Self-supervised learning can be used for signal compression and decompression recovery applications. Common algorithms include autoencoders and generative adversarial networks.

[0173] Reinforcement learning is different from supervised learning, and is a class of algorithms that learn the strategy for solving problems by interacting with the environment. Unlike supervised and unsupervised learning, the reinforcement learning problem does not have clear "correct" action label data. The algorithm needs to interact with the environment to obtain the reward signal of the environment feedback, and then adjust the decision action to obtain a larger reward signal value. In the following power control, the reinforcement learning model adjusts the downlink transmission power of each user according to the system total throughput rate feedback by the wireless network, and then expects to obtain a higher system throughput rate. The goal of reinforcement learning is to seek a decision action that can obtain the maximum cumulative reward in a long period of time. Reinforcement learning training is achieved through iterative interaction with the environment.

[0174] A neural network (NN) is a specific model in machine learning technology. According to the universal approximation theorem, a neural network can theoretically approximate any continuous function, so that the neural network has the ability to learn any mapping. Traditional communication systems need to use rich expert knowledge to design communication modules, while a deep learning communication system based on a neural network can automatically discover the implicit pattern structure from a large amount of data set, establish the mapping relationship between data, and obtain better performance than traditional modeling methods.

[0175] The idea of a neural network comes from the neuron structure of the brain tissue. For example, each neuron performs a weighted sum operation on its input values, and outputs the operation result through an activation function.

[0176] Figure 1a shows a schematic diagram of a neuron structure. Assume the input to the neuron is x = [x0, x1, ..., x...]. n The weights corresponding to each input are w = [w, w1, ..., w2]. n ], where n is a positive integer, w i and x i It can be any possible type, such as a decimal, an integer (e.g., 0, a positive integer, or a negative integer), or a complex number. i As x i The weights are used to assign weights to x. i Weighting is applied. The bias for the weighted summation of the input values ​​is, for example, b. Activation functions can take many forms. Assuming a neuron's activation function is y = f(z) = max(0, z), then the neuron's output is: For example, if the activation function of a neuron is y = f(z) = z, then the output of that neuron is: Here, b can be any possible type, such as a decimal, an integer (e.g., 0, a positive integer, or a negative integer), or a complex number. The activation functions of different neurons in a neural network can be the same or different.

[0177] Furthermore, neural networks generally consist of multiple layers, each of which may include one or more neurons. Increasing the depth and / or width of a neural network can improve its expressive power, providing more powerful information extraction and abstract modeling capabilities for complex systems. The depth of a neural network can refer to the number of layers it includes, and the number of neurons in each layer can be called the width of that layer. In one implementation, a neural network includes an input layer and an output layer. The input layer processes the received input information through neurons and passes the processing result to the output layer, which then obtains the output of the neural network. In another implementation, a neural network includes an input layer, hidden layers, and an output layer. The input layer processes the received input information through neurons and passes the processing result to the hidden layer. The hidden layer calculates the received processing result and passes the calculation result to the output layer or the next adjacent hidden layer, ultimately obtaining the output of the neural network. A neural network may include one hidden layer or multiple sequentially connected hidden layers, without limitation.

[0178] The neural network is, for example, a deep neural network (DNN). Depending on the construction of the network, the DNN can include a feed forward neural network (FNN), a convolutional neural network (CNN), and a recurrent neural network (RNN).

[0179] Fig. 1b is a schematic diagram of a FNN network. The FNN network is characterized by that each neuron in a layer is fully connected to each neuron in the next layer. This characteristic makes the FNN usually require a large amount of storage space and lead to a high computational complexity.

[0180] The CNN is a neural network specially designed to process data with a similar grid structure. For example, time series data (e.g., discrete sampling on a time axis) and image data (e.g., two-dimensional discrete sampling) can be considered as data with a similar grid structure. Instead of using all input information at once, the CNN uses a fixed-size window to extract partial information for convolution operation, which greatly reduces the computational load of model parameters. In addition, different convolution kernels can be used for each window according to the different types of information extracted by the window (e.g., people and objects in the same image are different types of information), which enables the CNN to better extract features of the input data.

[0181] The RNN is a DNN network that uses feedback time series information. The input of the RNN includes a new input value at the current time and the output value of itself at the previous time. The RNN is suitable for obtaining sequence features with temporal correlation, and is particularly suitable for speech recognition, channel coding and decoding, etc.

[0182] In the above model training process of machine learning, a loss function can be defined. The loss function describes the gap or difference between the output value of the model and the ideal target value. The loss function can be embodied in various forms, and the specific form of the loss function is not limited. The model training process can be regarded as the following process: by adjusting part or all of the parameters of the model, the value of the loss function is less than a threshold value or meets the target requirement.

[0183] The model can also be referred to as an AI model, a rule, or other names. The AI model can be considered as a specific method to realize AI functions. The AI model represents the mapping relationship or function between the input and output of the model. The AI functions can include one or more of the following: data collection, model training (or model learning), model information publishing, model inference (or model reasoning, reasoning, or prediction, etc.), model monitoring or model verification, or inference result publishing, etc. The AI function can also be referred to as an AI (related) operation, or an AI-related function.

[0184] The implementation process of the fully connected neural network will be described below with reference to the accompanying drawings. The fully connected neural network is also called a multilayer perceptron (MLP).

[0185] As shown in FIG. 1c, an MLP includes an input layer (left side), an output layer (right side), and multiple hidden layers (middle). Each layer of the MLP includes a plurality of nodes, which are called neurons. The neurons of adjacent two layers are connected to each other.

[0186] Optionally, considering the neurons of adjacent two layers, the output h of the neuron of the next layer is the weighted sum of all the neurons x of the previous layer connected to the neuron and is subjected to an activation function, which can be expressed as: h = f(wx + b).

[0187] where w is a weight matrix, b is a bias vector, and f is an activation function.

[0188] Further optionally, the output of the neural network can be recursively expressed as: y = f n (w n f n-1 (…)+b n ).

[0189] where n is the index of the layer of the neural network, n is greater than or equal to 1, and n is less than or equal to N, where N is the total number of layers of the neural network.

[0190] In other words, the neural network can be understood as a mapping relationship from a set of input data to a set of output data. Usually, the neural network is randomly initialized, and the process of obtaining this mapping relationship from the random w and b using the existing data is called training of the neural network.

[0191] Optionally, the specific way of training is to evaluate the output result of the neural network by using a loss function.

[0192] As shown in FIG. 1d, the error can be propagated backward, and the neural network parameters (including w and b) can be iteratively optimized by the gradient descent method until the loss function reaches the minimum value, i.e., the "better point (e.g., optimal point)" in FIG. 1d. It can be understood that the neural network parameters corresponding to the "better point (e.g., optimal point)" in FIG. 1d can be used as the neural network parameters in the trained AI model information.

[0193] Further optionally, the process of gradient descent can be expressed as:

[0194] wherein, θ is a parameter to be optimized (including w and b), L is a loss function, η is a learning rate, and controls the step size of gradient descent, denotes a derivative operation, denotes the derivative of L with respect to θ. Further, the process of back propagation utilizes the chain rule of partial derivatives.

[0195] As shown in FIG. 1e, the gradient of the previous layer parameters can be recursively calculated from the gradient of the next layer parameters, which can be expressed as:

[0196] wherein, w ij is the weight of node j connected to node i, and s i is the input weighted sum on node i.

[0197] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments.

[0198] (5) Wireless protocol stack.

[0199] The wireless protocol stack will be introduced below taking the wireless protocol stack of a new radio (NR) communication system as an example. The wireless protocol stack of NR can be divided into two planes, i.e., user plane (UP) protocol and control plane (CP) protocol, wherein the user plane protocol stack is the protocol cluster adopted for user data transmission, and the control plane protocol stack is the protocol cluster adopted for system control signaling transmission.

[0200] Specifically, the control plane protocol stack includes: a non-access stratum (NAS), a radio resource control (RRC) layer, a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a medium access control (MAC) layer, and a physical layer (PHY). The user plane protocol stack includes: a service data adaptation protocol (SDAP) layer, a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a medium access control (MAC) layer, and a physical layer (PHY). The application layer generally refers to the PDCP layer and / or the SDAP layer, and the application layer data is generally processed or transmitted at the PDCP layer and / or the SDAP layer. The application layer can also include other protocol layers above the radio protocol stack, including but not limited to an Internet protocol (IP) layer, a transmission control protocol (TCP) layer, or a (User Datagram Protocol, UDP) layer, etc.

[0201] (6) AI receiver.

[0202] For ease of understanding, please refer to FIG. 2b and FIG. 2c, FIG. 2b is a schematic diagram of a conventional receiver, and FIG. 2c is a schematic diagram of an AI receiver. The AI receiver refers to using an AI module to replace part or all of the modules of the conventional receiver, such as: a channel decoding module, an equalization demodulation module (or a demodulation module), or a channel estimation module, etc. The AI module runs a model so that the AI module performs the function corresponding to the module.

[0203] In the embodiments of the present application, for ease of description, the receiver is referred to as a first communication device, and the transmitter is referred to as a second communication device. The first communication device can be a terminal device or an access network device; the second communication device can be a terminal device or an access network device, and the embodiments of the present application do not limit this.

[0204] Next, the model-specific process of the receiver is described. Please refer to FIG. 2d, which is a schematic diagram of model training of an AI receiver. In the embodiments of the present application, the model run by the receiver is referred to as a first model, and the model run by the first communication device is referred to as a first model. The first model can include one or more models. After the receiver obtains the input data from the channel, the input data is transmitted to the first model. After the input data is processed by the first model, output data is obtained. After the receiver obtains the true value, the loss value of the first model is calculated according to the true value and the output data.

[0205] The first model can include any one or more models with module functions, including but not limited to: a channel estimation module, a demodulation module (or an equalization demodulation module), or a channel decoding module, etc. Alternatively, the first model is used to implement the functions of any one or more AI modules, including but not limited to: an AI channel estimation module, an AI demodulation module (or an AI equalization demodulation module), or an AI channel decoding module, etc.

[0206] (7) The AI transmitter and the AI receiver form a double-end model scenario.

[0207] FIGS. 2a and 2b are schematic diagrams of a single-end model scenario. Further, the transmitter side can also use AI modules to replace part or all of the modules of the traditional transmitter, such as: a channel coding module, a modulation module, or an AI reference signal module, etc. Such a transmitter using AI modules can be referred to as an AI transmitter. For ease of understanding, please refer to FIGS. 3a and 3b, which are schematic diagrams of a traditional transmitter and receiver scenario and a schematic diagram of an AI transmitter and AI receiver scenario. In FIG. 3b, the traditional reference signal module in the AI transmitter is replaced by an AI reference signal module, and the traditional modulation module is replaced by an AI modulation module; the traditional channel estimation module in the AI receiver is replaced by an AI channel estimation module, and the traditional demodulation module is replaced by an AI demodulation module.

[0208] In combination with the foregoing description, the model of the first communication device (receiver) is referred to as a first model, and the model of the second communication device (sender) is referred to as a second model for the sake of distinction. The first communication device runs the first model, and the second communication device runs the second model, and thus the first communication device and the second communication device constitute a two-end model scenario. Next, taking an example in which the first communication device is a terminal device, the second communication device is an access network device, the first model is an AI channel state information (CSI) compression model or AI CSI compression model, and the second model is an AI CSI reconstruction model or AI CSI reconstruction model, the model training manner of some current two-end model scenarios is introduced, in which the first model and the second model are used in cooperation to implement AI compression and AI reconstruction of CSI information.

[0209] Model training manner one:

[0210] Referring to FIG. 4a, FIG. 4a is a schematic diagram of a model training manner in an embodiment of the present application. Taking an example in which the access network device side leads the model training, the access network device trains the first model and the second model as a whole. Then, the trained first model is sent to the terminal device for use, and the trained second model is used by the access network device itself, to implement deployment of the two-end model.

[0211] Since the first model and the second model are both trained by the access network device, the input and output of the first model and the second model are controlled by one node, and thus the specific training manner of the model training manner one is similar to the training manner of the single-end model. For example, when the access network device side trains, the current sounding reference signal (SRS) measurement process can be followed to obtain a large amount of channel information.

[0212] However, in the model training manner one, after one end completes the model training, the trained model needs to be sent to the other end, and thus there is a problem of large communication overhead, and in addition, implementation of model transmission is relatively complex for air interface transmission.

[0213] Model training manner two:

[0214] Referring to FIG. 4b, FIG. 4b is a schematic diagram of another model training manner in an embodiment of the present application. The access network device side and the terminal device side can both design their own models by themselves, without the need to send their own models to the other end. However, in order to implement model training, the training data set needs to be sent from one end to the other end. Next, taking an example in which the access network device side sends the training data set to the terminal device side for description.

[0215] The access network device side first trains the first model and the second model, and then inputs the existing training data into the first model, the training data being an input vector (Vin), and the output of the first model for the Vin being a quantization vector (Vq). The Vin serves as the training data of the first model, and the Vq serves as the ground truth of the Vin or the Vq serves as the ground truth of the model training of the first model, and the Vin and the corresponding Vq constitute a training data set. Then, the access network device transmits the training data set to the terminal device, and the terminal device trains the local first model based on the training data set. After the training is completed, the first model on the terminal device side is equivalent to the first model on the access network device side.

[0216] However, in the second model training mode, the format of the training data set transmitted between the two ends is relatively complex, and the data amount is large, and therefore, the training data set is generally borne in a high layer, for example, borne in an application layer. The training data set borne in the application layer has the characteristic of a large data amount, and therefore, has the problem of a large communication overhead. In addition, processing the training data set with a large data amount requires that the first communication device and the second communication device have strong processing capability, for example, need to be processed through multiple protocol layers, and therefore, has the problem of a large processing overhead.

[0217] The third model training mode:

[0218] Please refer to FIG. 4c, which is a schematic diagram of another model training mode in the embodiment of the present application. The access network device side and the terminal device side can both design their own models by themselves, without the need to transmit their own models to the other end. In the second model training mode, the model on the access network device side cannot be changed after the access network device side transmits the training data set to the terminal device side. If the model on the access network device side needs to be changed, the access network device needs to send new training data set to the terminal device. Based on the problems in the second model training mode, in the third model training mode, the end receiving the training data set is allowed to backward propagate the model training gradient information of the model to the other end. The other end updates its own model based on the gradient information of the model.

[0219] With reference to FIG. 4c, the terminal device side is taken as an example to describe the model training starting first. The terminal device side does not need to assume that the access network device side uses the second model, and the terminal device side performs local training of the first model to obtain a forward inference result (FP), which includes training data. Then, the terminal device side sends the forward inference result and the corresponding true value to the access network device. In the AI CSI compression feedback, the true value refers to the input data of the first model corresponding to the forward inference result. The access network device trains the local second model according to the forward inference result and the corresponding true value. The access network device side inputs the forward inference result into the second model, compares the output of the second model with the true value corresponding to the forward inference result, and calculates a loss value. Then, the access network device calculates gradient information of the second model based on the loss value. Then, the access network device sends the gradient information to the first device through back propagation. The first device updates the local first model based on the gradient information. Through the above method, the model training of the first model and the second model is realized through iteration.

[0220] However, in the third model training method, the forward inference result and the gradient information transmitted between the two ends are implemented at a high layer, for example, at an application layer. The training data set carried at the application layer has the characteristic of a large amount of data, and therefore has the problem of a large communication overhead. In addition, processing the forward inference result and the gradient information with a large amount of data requires the first communication device and the second communication device to have strong processing capability, for example, to be processed through multiple protocol layers, and therefore has the problem of a large processing overhead.

[0221] In view of the above problems in the model training, an embodiment of the present application provides a communication method. A first communication device receives first data, and the first data is carried at a physical layer. The first communication device receives second data, and the second data is carried at the physical layer. The first data and the second data have an association relationship. The first data is used to determine a true value of model training or model monitoring of a first model. The second data is used to determine model input data of the first model. The first model is used to process physical layer data of the first communication device. Since the first data and the second data are carried at the physical layer, the first data and the second data can be directly processed at the physical layer without being processed through a high layer protocol. Therefore, the first data and the second data have the characteristic of a small amount of data, and therefore the processing overhead of the first communication device for obtaining the first data and the second data, and the processing overhead of the first communication device for performing model training or model monitoring based on the first data and the second data can be reduced, and the communication overhead of the first communication device can also be reduced.

[0222] The embodiments of the present application will be described in detail below with reference to the drawings. It should be noted that the communication method proposed by the embodiments of the present application can be applied to a single-end model scenario in addition to a double-end model scenario. Exemplarily, the single-end model scenario includes that a first communication device deploys a first model, and a second communication device does not have an AI function corresponding to the first model, for example, the first communication device adopts AI demodulation, and the second communication device adopts non-AI modulation; or the second communication device deploys a second model, and the first communication device does not have an AI function corresponding to the second model, and the embodiments of the present application do not limit this. First, the communication system to which the embodiments of the present application are applied is introduced. Please refer to FIG. 4d, which is an architecture schematic diagram of a communication system 1000 to which the embodiments of the present application are applied.

[0223] As shown in FIG. 4d, the communication system includes a radio access network 100 and a core network 200. Optionally, the communication system 1000 can also include the Internet 300. The radio access network 100 can include at least one access network device (which can also be understood as a kind of network device, such as 110a and 110b in FIG. 4d), and can also include at least one terminal (which can also be understood as the terminal device introduced in the foregoing, such as 120a-120j in FIG. 4d). In addition, the access network device (or referred to as the radio access network device) can be a macro base station (such as 110a in FIG. 4d), or a micro base station or indoor station (such as 110b in FIG. 4d), or a relay node or donor node, etc. It can be understood that all or part of the functions of the access network device in the present application can also be implemented by software functions running on hardware, or by virtualized functions instantiated on a platform (such as a cloud platform). The embodiments of the present application do not limit the specific technology and specific device form adopted by the radio access network device.

[0224] For ease of description, the communication system shown in FIG. 4d is described by taking the access network device as a base station and the terminal device as a terminal as an example. It can be understood that when the communication system includes an integrated access and backhaul (IAB) network, the base station can be an IAB node. It should be noted that the base station and the access network device in the embodiments of the present application can be replaced with each other.

[0225] In the present application, the base station and the terminal can be fixed in position or movable. The base station and the terminal can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted, can also be deployed on the water surface, and can also be deployed on aircraft, balloons and artificial satellites in the air. The embodiments of the present application do not limit the application scenarios of the base station and the terminal.

[0226] The roles of the base stations and the terminals can be relative, for example, the helicopter or the drone 120i in FIG. 4d can be configured as a mobile base station, and for those terminals 120j accessing to the wireless access network 100 through the 120i, the terminal 120i is a base station. But for the base station 110a, the 120i is a terminal, that is, the 110a and the 120i communicate through the wireless air interface protocol. Of course, the 110a and the 120i can also communicate through the interface protocol between base stations and base stations, at this time, the 120i is also a base station relative to the 110a. Therefore, the base stations and the terminals can be collectively referred to as communication devices, the 110a and the 110b in FIG. 4d can be referred to as communication devices with base station functions, and the 120a-120j in FIG. 4d can be referred to as communication devices with terminal functions.

[0227] The base stations and the terminals, the base stations and the base stations, and the terminals and the terminals can communicate through the licensed spectrum, or through the unlicensed spectrum, or through the licensed spectrum and the unlicensed spectrum at the same time. The communication can be through the spectrum below 6 gigahertz (GHz), or through the spectrum above 6 GHz, or through the spectrum below 6 GHz and the spectrum above 6 GHz at the same time. The embodiments of the present application do not limit the spectrum resources used for wireless communication.

[0228] The technical solutions of the present application can be applied to the cellular communication system related to the 3rd generation partnership project (3GPP). For example, the fourth generation (4G) communication system, the 5G communication system, the communication system after the 5G communication system. For example, the future communication system. For example, the fourth generation communication system can include the long term evolution (LTE) communication system. The fifth generation communication system can include the new radio (NR) communication system. The technical solutions of the present application can also be applied to the wireless fidelity (WiFi) system, the communication system supporting multiple wireless technology fusion, the device-to-device (D2D) system, or the vehicle to everything (V2X) communication system.

[0229] The terminal device and the access network device related to the present application are introduced below.

[0230] Terminal device, also known as user equipment (UE), mobile station (MS), mobile terminal (MT), fixed wireless access (FWA), customer premise equipment (CPE), etc. The terminal device is a device including a wireless communication function (providing voice / data connectivity to users). For example, handheld devices with wireless connection function, vehicle-mounted devices, machine type communication (MTC) terminals, etc. At present, the terminal device can include: mobile phone, tablet computer, notebook computer, palm computer, mobile internet device (MID), wearable device, virtual reality (VR) device, augmented reality (AR) device, wireless terminal in industrial control, wireless terminal in self driving (e.g. unmanned aerial vehicle, vehicle), wireless terminal in remote medical surgery, wireless terminal in smart grid, wireless terminal in transportation safety, wireless terminal in smart city, or wireless terminal in smart home, etc. For example, the wireless terminal in self driving can be unmanned aerial vehicle, helicopter, or airplane, etc. For example, the wireless terminal in Internet of Vehicles can be vehicle-mounted device, whole vehicle device, vehicle-mounted module, vehicle, or ship, etc. The wireless terminal in industrial control can be camera, robot, or mechanical arm, etc. The wireless terminal in smart home can be television, air conditioner, sweeping machine, sound box, or set top box, etc. The terminal device can also be a device or module with corresponding communication function accessing the above-mentioned communication system. The terminal device is usually provided with a communication module, circuit or chip for executing corresponding communication function, and is also configured with program instructions for executing corresponding communication function.

[0231] It should be noted that the terminal device can be a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a module or a control unit in the above-mentioned device or apparatus, and the specific application is not limited. It should be noted that in this application, when referring to a terminal device, it can refer to the terminal device itself, or a chip, functional module or integrated circuit in the terminal device that completes the method provided in this application, and the specific application is not limited. The access network device is a device deployed in the wireless access network to provide wireless communication function for the terminal device. The access network device can access the terminal device to the radio access network (RAN) node of the wireless network, which can also be called access network device, RAN entity, access node, network node, or communication device, etc.

[0232] Specifically, the access network device can be an access network device for a 3rd generation partnership project (3GPP) related cellular system. For example, a 4G communication system, or a 5G communication system, or a future communication system. The access network device can also be an access network device in an open access network (open RAN, O-RAN or ORAN) or a cloud radio access network (CRAN). Alternatively, the access network device can also be an access network device in a communication system obtained by fusing two or more of the above communication systems.

[0233] The access network device includes, but is not limited to, an evolved Node B (eNB), a radio network controller (RNC), a Node B (NB), a base station controller (BSC), a base transceiver station (BTS), a home base station (for example, a home evolved NodeB or a home Node B, HNB), a baseband unit (BBU), an access point (AP) in a wireless fidelity (WiFi) system, a macro base station, a micro base station, a wireless relay node, a donor node, a wireless controller in a CRAN scenario, a wireless backhaul node, a transmission point (TP) or a transmission and reception point (TRP), and the like, and can also be an access network device in a 5G mobile communication system. For example, a next generation NodeB (gNB), a TRP or a TP in an NR system; or one or a group (including multiple antenna panels) of antenna panels of a base station in a 5G mobile communication system; or the access network device can also be a network node constituting a gNB or a transmission point. For example, a centralized unit (CU), a distributed unit (DU), a centralized unit control plane (CU-CP), a centralized unit user plane (CU-UP), or a radio unit (RU), and the like. The CU and the DU can be separately arranged or can be included in the same network element, for example, a BBU. The RU can be included in a radio frequency device or a radio frequency unit. For example, in a remote radio unit (RRU), an active antenna unit (AAU) or a remote radio head (RRH). Or the access network device can also be a server, a wearable device, a vehicle or a vehicle-mounted device, and the like. For example, the access network device in V2X technology can be a road side unit (RSU). It should be understood that the above-mentioned TRP can be a device or module located at the network side of the above-mentioned communication system and having corresponding communication functions. The TRP is usually provided with a communication module, circuit or chip for performing corresponding communication functions.The TRPs also have program instructions configured for respective communication functions.

[0234] It should be noted that in different systems, the CU (or CU-CP and CU-UP), DU or RU can also have different names, but those skilled in the art can understand their meanings. For example, in an open radio access network (ORAN) system, the CU can also be referred to as an open centralized unit (O-CU) or an open CU, the DU can also be referred to as an open distributed unit (O-DU), the CU-CP can also be referred to as an open centralized unit control plane (O-CU-CP), the CU-UP can also be referred to as an open centralized unit user plane (O-CU-UP), and the RU can also be referred to as an open radio unit (O-RU). The specific application is not limited. Any one of the CU, CU-CP, CU-UP, DU and RU in the present application can be realized by a software module, a hardware module, or a combination of a software module and a hardware module.

[0235] Optionally, for network elements in the ORAN system, each network element can implement the protocol layer functions shown in Table 1 below.

[0236] Table 1

[0237] It should be noted that in the ORAN system, the access network device in the present application can be one or more network elements in Table 1 above.

[0238] The architecture of the CU and the DU of the access network device will be introduced below. The access network device includes at least one CU and at least one DU. Optionally, the access network device also includes at least one RU.

[0239] The following is introduced by taking an access network device including a CU and a DU as an example. The CU has part of the function of the core network, and the CU can include a CU-CP and a CU-UP. The CU and the DU can be configured according to the protocol layer function of the wireless network they implement. For example, the CU is configured to implement the function of the packet data convergence protocol (PDCP) layer and the protocol layer above (for example, the function of the RRC layer and / or the SDAP layer). The DU is configured to implement the function of the protocol layer below the PDCP layer (for example, the function of the RLC layer, the MAC layer, and / or the physical (PHY) layer). For another example, the CU is configured to implement the function of the protocol layer above the PDCP layer (for example, the function of the RRC layer and / or the SDAP layer), and the DU is configured to implement the function of the protocol layer below the PDCP layer (for example, the function of the RLC layer, the MAC layer, and / or the PHY layer, etc.).

[0240] When the CU includes the CU-CP and the CU-UP, the CU-CP is used to implement the control plane function of the CU, and the CU-UP is used to implement the user plane function of the CU. For example, when the CU is configured to implement the function of the PDCP layer, the RRC layer and the SDAP layer, the CU-CP is used to implement the function of the RRC layer and the control plane function of the PDCP layer, and the CU-UP is used to implement the function of the SDAP layer and the user plane function of the PDCP layer.

[0241] The CU-CP can interact with a network element in the core network for implementing the control plane function. The network element in the core network for implementing the control plane function can be an access and mobility function network element, for example, an access and mobility management function (AMF) in a 5G system. The AMF is used to be responsible for the mobility management in the mobile network, such as the location update of the terminal device, the registration network of the terminal device, the handover of the terminal device, etc.

[0242] The CU-UP can interact with a network element in the core network for implementing the user plane function. The network element in the core network for implementing the user plane function, for example, a user function (UPF) in a 5G system, is used to be responsible for the forwarding and receiving of data in the terminal device.

[0243] Optionally, under the ORAN architecture, a RAN intelligent controller (RIC) module is also involved.

[0244] It should be noted that the access network device can be a device or apparatus with a chip, or a device or apparatus integrated with a circuit, or a chip, a chip system, a module or a control unit in the foregoing illustrated device or apparatus, and the specific application is not limited. It should be noted that in this application, when referring to the access network device, it can refer to the access network device itself, or refer to the chip, functional module or integrated circuit in the access network device that completes the method provided in the application, and the specific application is not limited.

[0245] In the embodiments of the present application, the functions of the access network device can also be performed by a module (such as a chip) in the access network device, or by a control subsystem containing the functions of the access network device. The control subsystem containing the functions of the access network device herein can be a control center in the application scenarios of the terminal devices such as smart grids, industrial control, intelligent transportation, and smart cities.

[0246] In combination with the above communication system, please refer to FIG. 5, which is a flow diagram of an embodiment of a communication method in the present application. In the embodiments of the present application, the first communication device acts as a receiver, and the second communication device acts as a transmitter or a transmitter. In one example scenario, the first communication device is a terminal device, and the second communication device is an access network device, and this example scenario is for a downlink scenario. In another example scenario, the first communication device is an access network device, and the second communication device is a terminal device, and this example scenario is for an uplink scenario. The communication method proposed in the embodiments of the present application includes:

[0247] S1, the second communication device sends first data to the first communication device, and the first data is carried in the physical layer.

[0248] In step S1, the second communication device sends first data to the first communication device, and the first data is carried in the physical layer, in other words, the first data can be a group of data transmitted in the form of a bit stream carried in the physical layer. The first data is used to determine the true value of the model training or model monitoring of the first model, and the first model refers to the model deployed in the first communication device. The first model is used to process the physical layer data of the first communication device.

[0249] Next, the first data is introduced, and first, the generation manner of the first data is introduced. For ease of understanding, refer to FIG. 6, which is a data processing schematic diagram involved in the communication method in the embodiment of the present application. The second communication device generates the first data by performing third processing on the sixth data, and the third processing refers to non-AI processing, that is, the third processing does not involve an AI model but adopts a traditional processing manner. The third processing includes any one or more of the following processing: channel coding, modulation (or modulation equalization), precoding, generation of the first reference signal, or resource mapping of the first reference signal. Among them, the first reference signal is used for channel estimation of the wireless resource carrying the first data. Since the first reference signal is a reference signal obtained by using non-AI processing, the first reference signal can also be called a legacy reference signal.

[0250] It should be noted that after the second communication device sends the first data, the first data reaches the first communication device through a channel. The first data through the channel can change, that is, the first data received by the first communication device can be inconsistent with the first data sent by the second communication device. However, for ease of description, the first data received by the first communication device and the first data sent by the second communication device are collectively referred to as the first data in the embodiment of the present application. Similarly, after the second communication device sends the second data, the second data reaches the first communication device through a channel. The second data through the channel can change, that is, the second data received by the first communication device can be inconsistent with the second data sent by the second communication device. However, for ease of description, the second data received by the first communication device and the second data sent by the second communication device are collectively referred to as the second data in the embodiment of the present application.

[0251] In an example, when the first reference signal is used for channel estimation, the first reference signal can be a demodulation reference signal (DMRS). It is noted that the first reference signal can be used for determination of channel state information (CSI), estimation of phase noise, estimation of time offset, estimation of frequency offset, location estimation, and / or environment target perception, etc. in addition to channel estimation. Accordingly, the first reference signal can also include a channel state information–reference signal (CSI-RS), a positioning reference signal (PRS), a sounding reference signal (SRS), a phase tracking reference signal (PTRS), a tracking reference signal (TRS), a perception signal, or a synchronization signal / physical broadcast channel block (SSB), a random access preamble, etc.

[0252] In an example, channel coding is performed using low-density parity-check code (LDPC) or polar code, for example. Modulation is performed using quadrature phase shift keying (QPSK), 16-quadrature amplitude modulation (16QAM), or 64-quadrature amplitude modulation (64QAM), for example. In an example, the reference signal is generated using a sequence such as a m-sequence, a gold sequence, a Zadoff-Chu (ZC) sequence, etc.

[0253] When the third processing includes different processing manners, the sixth data refers to different data. For example, when the third processing includes channel coding, the sixth data is data before channel coding, i.e., the sixth data is a transport block TB before channel coding. For another example, when the third processing includes modulation, the sixth data is data after channel coding processing and before modulation, i.e., the sixth data is a codeword after channel coding. For another example, when the third processing includes precoding, the sixth data is a codeword before precoding. For another example, when the third processing includes resource mapping of the first reference signal, the sixth data is a codeword after precoding processing and before resource mapping of the first reference signal.

[0254] Secondly, the transmission mode of the first data is introduced. Specifically, the first data can multiplex part of the transmission data, or the first data and the transmission data are independent of each other, and the transmission data is determined based on user plane data and / or control plane data. For example, the transmission data can come from the MAC layer above the physical layer (as shown in FIG. 2a), that is, the data in the MAC protocol data unit (PDU) issued by the MAC layer and received by the physical layer can be considered as the transmission data. The user plane data is, for example, service-related data, and the control plane data is, for example, signaling-related data. The transmission data can also be user plane data or control plane data dedicated for model training or model monitoring. The transmission data can be further processed by the third processing mentioned above, and the transmission data can also be processed by the physical layer. The two transmission modes of the first data are described below.

[0255] In a possible implementation, the second communication device sends transmission data to the first communication device, and part of the transmission data can be used as the first data. In other words, the first data multiplexes part of the transmission data. The specific mode is as follows:

[0256] S1-1, the second communication device determines the size of the transmission resource occupied by the second data and / or the transmission data.

[0257] In step S1-1, the second communication device can reserve part of the transmission resource for transmitting the second data. Based on the part of the transmission resource reserved for the second data, the second communication device can determine the transmission resource for transmitting the transmission data, and the transmission resource for transmitting the transmission data needs to subtract the part of the transmission resource reserved from the available transmission resource. For example, the second communication device determines the number of resource elements (REs) allocated for a physical downlink shared channel (PDSCH) in a physical resource block (PRB) through X, that is, the size or overhead of a transport block (TB) carrying the transmission data is determined according to X, and the X includes: the number of subcarriers in a physical resource block the overhead of DMRS occupied by each PRB in the scheduling duration the overhead configured by the information element “xoverhead” in the configuration information “PDSCH-ServingCellconfig” the overhead of the transmission resource reserved for the second data. The overhead of the transmission resource reserved for the second data can be represented by the number of REs.

[0258] S1-2, determining which data in the transmission data as the first data.

[0259] Optionally, the transmission data can be data before channel coding, i.e., the transmission data is TB before channel coding determined by S1-1, and the size of the transmission data can be the number of bits determined based on the above X, modulation mode, code rate, number of transmission layers and / or number of streams.

[0260] Optionally, the transmission data can be data after channel coding, i.e., the transmission data is codeword after channel coding determined by S1-1, and the size of the transmission data can be the number of bits determined based on the above X, modulation mode, number of transmission layers and / or number of streams.

[0261] In step S1-2, the first data can be a part of continuous data in the transmission data, for example, the first N bits of the transmission data as the first data, N being a positive integer greater than 0. The first data can also be data composed of multiple segments of discontinuous data in the transmission data, for example, the N1th bit to the N2th bit, the N3th bit to the N4th bit, and the N5th bit to the N6th bit of the transmission data as the first data, N1, N2, N3, N4, N5, and N6 being positive integers greater than 0, and N6>N5>N4>N3>N2>N1.

[0262] The second communication device can negotiate with the first communication device to determine the relative position of the first data in the transmission data; or the second communication device can itself determine the relative position of the first data in the transmission data and notify the first communication device of the relative position of the first data in the transmission data; or the second communication device can determine the relative position of the first data in the transmission data in a protocol predefined manner and indicate the relevant information (such as the above bit number N) to the first communication device; or the second communication device determines the relative position of the first data in the transmission data according to the configuration of other devices, such as a core network device or a network manager.

[0263] In the embodiments of the present application, the second communication device notifies the first communication device of the relative position of the first data in the transmission data through the third information. The third information includes one or more of the following information: identification information of a TB carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, an ending position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data, wherein the data amount can be the number of bits. If the first data is composed of multiple segments of discontinuous data in the transmission data, the third information can indicate the starting position and the ending position of the multiple segments of discontinuous data.

[0264] S1-3, the second communication device sends the transmission data to the first communication device, the transmission data comprising the first data.

[0265] In step S1-3, after the second communication device determines the overhead of the TB carrying the transmission data, the second communication device maps the transmission data into the RE corresponding to the transmission data. Optionally, if the transmission data is a TB before channel coding, the second communication device performs channel coding on the transmission data. The mapping process can be performed by using rate matching in channel coding, and the purpose of rate matching is to adjust the data stream after channel coding to a rate matching the actual channel resource and transmission requirement. The rate matching includes one or more of the following operations: bit selection, bit repetition or bit interleaving, wherein the bit selection refers to selecting a specific number of bits from the coded data stream to adapt to the target transmission rate, the bit repetition refers to repeating some bits in the data stream to increase the reliability of transmission, and the bit interleaving refers to disordering the order of bits in the data stream to improve the ability to resist burst errors.

[0266] Optionally, if the second communication device reserves part of the transmission resource (for example, RE) for the second data, the second communication device can bypass the RE reserved for the second data by rate matching in the process of mapping the transmission data to the RE. For example, the second communication device reserves RE1, RE2 and RE3 for the second data, and in the process of mapping the transmission data to the RE, the second communication device does not map the transmission data to the above-mentioned RE1, RE2 and RE3.

[0267] For example, the second communication device reserves part of the transmission resource for transmitting the second data, and based on the part of the reserved transmission resource, the second communication device determines the size of the transmission resource occupied by the first data. Then, the second communication device determines which part of the transmission data to intercept as the first data.

[0268] In another possible implementation, the first data can not be multiplexed with the transmission data, but directly generated by the physical layer of the second communication device, for example, the physical layer of the second communication device randomly generates some data, which are not determined based on the user plane data or the control plane data. In this case, the specific implementation is similar to steps S1-1 to S1-3, which will not be described here.

[0269] For ease of understanding, please refer to FIG. 7, which is a schematic diagram of the first data in an embodiment of the present application. The second communication device sends the transmission data to the first communication device, and a part of the transmission data can be used to determine the first data. After the first communication device receives the transmission data, the first communication device can determine the first data according to the part of the transmission data.

[0270] In another possible implementation, the second communication apparatus sends the transmission data and the first data to the first communication apparatus, and the transmission data and the first data are independent of each other. In this implementation, when the second communication apparatus calculates the overhead of the TB carrying the transmission data, the aforementioned X can further subtract the overhead occupied by the transmission resource reserved for the first data. For example, the first data can be data obtained from the transmission data, for example, the first data is data obtained by copying part of the transmission data. In another example, the first data can also be predefined data, for example, a bit stream of "010101···". In another example, the first data can also be random data generated by the second communication apparatus, for example, the physical layer of the second communication apparatus randomly generates some data, and the data is not determined based on user plane data or control plane data, and the second communication apparatus takes the random data as the first data.

[0271] Optionally, the first data further comprises a first reference signal. The embodiments of the present application do not limit the type of the first reference signal, for example, the first reference signal can be any one of the following reference signals: a channel state information-reference signal (CSI-RS), a positioning reference signal (PRS), a sounding reference signal (SRS), a phase tracking reference signal (PTRS), a tracking reference signal (TRS), a sensing signal, or a synchronization signal / physical layer broadcast channel block (SSB), a random access preamble, or a demodulation reference signal (DMRS), and the like.

[0272] Optionally, the second communication apparatus can further process the first data, for example, perform bit flipping, and then send the flipped first data to the first communication apparatus as part of the transmission data.

[0273] S2, the second communication apparatus sends second data to the first communication apparatus, and the second data is carried in a physical layer.

[0274] In step S2, the second communication apparatus sends second data to the first communication apparatus, and the second data is carried in a physical layer, in other words, the second data can be a set of data transmitted in a bit stream form and carried in a physical layer. The second data is used to determine the model input data of the first model.

[0275] The generation manner of the second data is introduced below. For ease of understanding, the data processing manner shown in FIG. 6 is used for illustration. In the process of generating the first data, the second communication apparatus first determines the sixth data. Then, the second communication apparatus performs copying processing on the sixth data to obtain the fifth data, which can be a subset of the sixth data. The second communication apparatus generates the second data according to the fifth data. Specifically, the second communication apparatus performs fourth processing on the fifth data to obtain the second data, and the fourth processing refers to AI processing. The model corresponding to the fourth processing can be the second model that has been trained, and correspondingly, the first model in the first communication apparatus is the first model to be trained or the first model to be monitored. The model corresponding to the fourth processing can also be the second model, which can be two independent models from the first model in the first communication apparatus, or the second model can be a model having a pairing relationship with the first model. The pairing relationship refers to that the first model and the second model jointly constitute a shared model, which can complete one or more corresponding sending and receiving functions, such as modulation and / or demodulation, channel encoding and / or channel decoding, reference signal and / or reference signal estimation, etc. For ease of description, the following description takes the second model used by the second communication apparatus as an example.

[0276] The fourth processing includes any one or more of the following processing: AI channel encoding, AI modulation, AI precoding, generation of a second reference signal, or AI resource mapping of the second reference signal. The second reference signal refers to a reference signal obtained by AI processing, or a reference signal obtained by AI training, or a reference signal generated by the second model, so the second reference signal can also be referred to as an AI reference signal. For example, after the first communication apparatus receives the second reference signal, the second reference signal is used for channel estimation of the wireless resource carrying the second data by the first model. The second reference signal can also be used for channel state information estimation of the wireless resource carrying the second data by the first model, or phase noise estimation, time offset estimation, frequency offset estimation, or target perception estimation during transmission of the second data, etc.

[0277] When the fourth processing includes different processing manners, the fifth data refers to different data. For example, when the fourth processing includes AI channel coding, the fifth data is the data before AI channel coding (correspondingly, the sixth data is the TB before channel coding). For another example, when the fourth processing includes AI modulation, the fifth data is the data after channel coding processing and before AI modulation (correspondingly, the sixth data is the codeword after channel coding). For another example, when the fourth processing includes AI precoding, the fifth data is the data after modulation processing and before AI precoding. For another example, when the fourth processing includes resource mapping of the second reference signal, the fifth data is the data after precoding processing and before resource mapping of the second reference signal. For another example, when the fourth processing includes AI channel coding, AI modulation, AI precoding, generation of the second reference signal, and AI resource mapping of the second reference signal, the fifth data is the original data input to the second model (correspondingly, the sixth data is the TB before channel coding), and the second data is the to-be-transmitted data transmitted to the first communication apparatus.

[0278] In steps S1 and S2, the third processing and the fourth processing have a corresponding relationship: for example, when the third processing includes channel coding, the fourth processing includes AI channel coding; for another example, when the third processing includes modulation, the fourth processing includes AI modulation; for another example, when the third processing includes precoding, the fourth processing includes AI precoding; for another example, when the third processing includes resource mapping of the first reference signal, the fourth processing includes resource mapping of the second reference signal; for another example, when the third processing includes channel coding and modulation, the fourth processing includes AI channel coding and AI modulation.

[0279] The second communication apparatus can configure, in response to a request of the first communication apparatus, a training task of the first model or a monitoring task of the first model for the first communication apparatus, or instruct the first communication apparatus to collect training data of the first model or collect monitoring data of the first model. The second communication apparatus can also actively configure a training task of the first model or a monitoring task of the first model for the first communication apparatus, or actively instruct the first communication apparatus to collect training data of the first model or collect monitoring data of the first model. The specific configuration process is as follows: the second communication apparatus determines the fourth processing according to the second processing executable by the first model in the first communication apparatus, and then determines the third processing according to the fourth processing. The second processing is AI processing, and the second processing can be regarded as the reverse processing of the fourth processing. For example, the second processing includes any one or more of the following processing: AI channel decoding, AI demodulation, or AI estimation based on the second reference signal (for example, AI channel estimation based on the second reference signal).

[0280] Optionally, the second data includes the second reference signal.

[0281] In the embodiments of the present application, the second reference signal refers to a reference signal having a correlation relationship with the first reference signal and being obtained based on AI processing or AI training. Taking the first reference signal and the second reference signal for channel estimation as an example, in model training or model monitoring of the first model, the channel estimation result based on the first reference signal can be used as the true value of the channel estimation result based on the second reference signal. This is because the true value of the estimation object of the reference signal is difficult to obtain directly or to agree in advance in some cases, for example, the true channel information in channel estimation, which needs to be obtained indirectly by actual measurement. The first reference signal and the second reference signal have a correlation relationship, which can be expressed as that the first reference signal and the second reference signal have a quasi co-location (QCL) relationship. The QCL relationship between the first reference signal and the second reference signal can be a newly defined QCL relationship. Four types of QCL relationships are currently defined to indicate the relationship between two reference signals, for example: the first reference signal and the second reference signal have a QCL relationship, which can indicate that one or more of the channel delay spread, average delay, Doppler spread, Doppler shift, or spatial reception parameters corresponding to the first reference signal are the same as one or more of the channel delay spread, average delay, Doppler spread, Doppler shift, or spatial reception parameters corresponding to the second reference signal. The first reference signal and the second reference signal having a correlation relationship can also be expressed as: the estimation result corresponding to the first reference signal is the same as the estimation result corresponding to the second reference signal, which is, for example: channel estimation result, channel state information estimation result, phase noise estimation result, time-frequency offset estimation result, or perception target estimation result. In other words, the estimation result corresponding to the first reference signal can be used as the true value of the estimation result corresponding to the second reference signal.

[0282] S3, the first communication device performs model training or model monitoring on the first model according to the first data and the second data.

[0283] In step S3, after the first communication device receives the first data and the second data from the second communication device, the first communication device performs model training or model monitoring on the first model according to the first data and the second data, or the first communication device determines training data for training the first model or monitoring data for monitoring the first model according to the first data and the second data.

[0284] First, the processing mode of the first communication device for the first data is introduced:

[0285] After the first communication device receives the first data, the first communication device performs a first processing on the first data to generate third data. The first processing is a non-AI processing, and the first processing includes any one or more of the following processing: channel decoding (e.g., using a minimum sum algorithm), demodulation (e.g., using a mean square error minimization algorithm), or channel estimation based on the first reference signal (e.g., using a Wiener filtering algorithm). The first communication device can determine the processing manner included in the first processing according to the first indication information of the second communication device, where the first indication information indicates the third processing adopted by the second communication device when generating the first data and / or the related parameter information (e.g., a modulation manner, a coding rate, etc.) in the third processing. For example, the first indication information indicates that the third processing includes channel coding and coding rate information, and the first processing is determined to include channel decoding and to perform channel decoding according to the corresponding coding rate. For another example, the first indication information indicates that the third processing includes modulation and a modulation manner, and the first processing is determined to include demodulation and to perform demodulation according to the corresponding modulation manner. That is, the first processing is the reverse processing of the third processing. It should be noted that, since the third processing is a non-AI processing, the first communication device can also know the processing manner (e.g., defined by a protocol) and / or the related parameter information in the processing that the third processing can include in advance, and therefore, the second communication device can also not indicate the processing manner of the third processing and / or the related parameter information in the processing through the first indication information. In some cases, the first processing includes the reverse processing of the third processing, and in addition, the first processing can also include other processing manners, for example, the first processing includes performing other processing manners after performing the reverse processing of the third processing.

[0286] Similarly to the third processing, when the first processing includes different processing manners, the third data refers to different data. For example, when the first processing includes channel decoding, the third data is the data obtained after channel decoding. For another example, when the first processing includes demodulation, the first data is the data after demodulation processing and before channel decoding. For another example, the first processing includes resource mapping of the first reference signal, and the third data is the channel estimation result based on the first reference signal. It should be noted that, the precondition for the third data to be valid is that the first data is correctly received, for example, the signal-to-noise ratio of the first data meets the requirement of the first processing, or the first data can be correctly decoded or demodulated.

[0287] In a possible implementation, after the first communication device receives the transmission data from the second communication device, the first communication device determines the part of data corresponding to the first data from the transmission data according to the third information. Then, the first communication device processes the part of data to recover the first data, and the processing includes physical layer processing. Alternatively, the first communication device directly determines the first data from the transmission data according to the third information.

[0288] In an example, the resource carrying the transmission data is a code word 1 corresponding to a transport block (TB) 1, the first N bits of the code word 1 are used to determine the first data, the third processing is modulation, and the fourth processing is AI modulation. After the first communication device receives the code word 1, the first communication device performs a first processing on the code word 1 (including the first data) to generate first intermediate data, which can be a TB determined after channel decoding. The first processing includes resource demapping, channel estimation, demodulation, and / or channel decoding, etc. When the first communication device detects that the first intermediate data satisfies the validity condition of the first data, for example, the first intermediate data passes the check (i.e., the transmission data successfully performs channel decoding), the first communication device determines that the first data can be used to determine the true value for model training or model monitoring of the first model. The first communication device determines that channel coding processing needs to be performed on the first intermediate data according to the third processing (modulation) to obtain the code word 1. Therefore, the first communication device performs channel coding processing on the first intermediate data to determine the code word 1 and part of the data in the code word 1 to obtain the third data. In other words, the part of the data refers to the first N bits of the code word 1 corresponding to the first intermediate data, i.e., the third data. For ease of distinction, the channel coding processing performed by the first communication device on the first intermediate data can also be referred to as channel recoding processing. It can be understood that the channel recoding processing used by the first communication device is the same as the channel coding used by the second communication device when generating the code word 1. For example, the first communication device can determine the corresponding channel coding according to the channel decoding processing used.

[0289] Optionally, the first communication device can perform the processing on the first data by itself, or the first communication device can perform the processing on the first data in cooperation with other devices, including but not limited to an over the top (OTT) server, an edge computing device, a cloud computing platform, or an external service board of the first communication device, etc. For example, in the above example, the first communication device performs the first processing on the code word 1, and the other device performs the channel recoding processing. In this way, the processing overhead and energy consumption of the first communication device can be reduced.

[0290] Secondly, the processing manner of the first communication device on the second data is introduced.

[0291] In a possible implementation, the second communication device can further send second information to the first communication device, where the second information is used to indicate resources carrying the second data. The second information is used to indicate one or more of the following resources: frequency domain resources carrying the second data, time domain resources carrying the second data, or spatial domain resources carrying the second data. For example, the frequency domain resources carrying the second data include: a frequency band, a subcarrier, a resource element (RE), or an RE pattern, where the RE pattern includes one or more REs, for example, two time-domain adjacent REs can form an RE pattern with a size of 1*2, and four time-domain and frequency-domain adjacent REs can form an RE pattern with a size of 2*2; the time domain resources carrying the second data include: a slot, a symbol, or a subframe; and the spatial domain resources carrying the second data include: an antenna port. In response to the second information, the first communication device determines the second data from the received data.

[0292] In another possible implementation, the first communication device and the second communication device agree on resources for transmitting the second data, for example, by means of a protocol agreement. According to the agreement, the first communication device determines the second data from the received data according to the agreed resources.

[0293] After the first communication device determines the second data, the first communication device performs second processing on the second data to generate fourth data. The second processing is AI processing, and the second processing includes one or more of the following: AI channel decoding, AI demodulation, or AI channel estimation based on the second reference signal. The first communication device can determine the processing manner included in the second processing according to second indication information of the second communication device, where the second indication information indicates a fourth processing adopted by the second communication device when generating the second data. For example, the second indication information indicates that the fourth processing includes AI channel encoding, and it is determined that the second processing includes AI channel decoding. For another example, the second indication information indicates that the fourth processing includes AI modulation, and it is determined that the second processing includes AI demodulation. That is, the second processing is the reverse processing of the fourth processing. In some cases, the second processing at least includes the reverse processing of the fourth processing, for example, the second processing can perform other processing manners after performing the reverse processing of the fourth processing.

[0294] Similar to the fourth processing, when the second processing includes different processing manners, the fourth data refers to different data. For example, when the second processing includes AI channel decoding, the fourth data is data obtained after AI channel decoding. For another example, when the second processing includes AI demodulation, the fourth data is data obtained after AI demodulation without channel decoding. For another example, when the second processing includes resource mapping of the second reference signal, the fourth data is a channel estimation result based on the second reference signal.

[0295] Optionally, after the first communication device receives and determines the second data, the first communication device can perform a buffering process on the second data. That is, the first communication device does not perform the second process on the second data immediately. When the first communication device determines that the first data and the second data satisfy the validity condition, the first communication device continues to perform the second process on the buffered second data; when the first communication device determines that the first data and the second data do not satisfy the validity condition, the first communication device clears the buffered second data, saving the storage and processing overhead of the first communication device.

[0296] Optionally, the second communication device can send fifth information to the first communication device, the fifth information being used to configure the validity condition of the first data and the second data, the validity condition including one or more of the following: the transmission data performs channel decoding successfully; the transmission data demodulates successfully; or, the condition of channel estimation performed by the first data satisfies a certain threshold requirement. The condition of channel estimation includes one or more of the following: signal to noise ratio (SNR), signal to interference plus noise ratio (SINR), or time delay spread, which refers to the maximum arrival time difference of signals transmitted between the second communication device and the first communication device through different paths, average time delay, Doppler spread, and Doppler offset.

[0297] Again, the specific way in which the first communication device performs model training or model monitoring on the first model is introduced.

[0298] Through the above method, the first communication device performs first processing on the first data to obtain third data; the first communication device performs second processing on the second data to obtain fourth data. The third data is used as the true value for model training or model monitoring of the first model, or the third data is used as the true value of the fourth data, and the fourth data is used as the model output data of the first model. Then the first communication device can use the third data and the fourth data to perform model training or model monitoring on the first model.

[0299] Optionally, in addition to using the third data and the fourth data to perform model training or model monitoring on the first model, the first communication device can also use other data to perform model training or model monitoring on the first model, the other data including but not limited to: channel information obtained by performing channel estimation on the channel between the first communication device and the second communication device in a non-AI processing manner, and / or channel information obtained by performing channel estimation on the channel between the first communication device and the second communication device in an AI processing manner.

[0300] The specific way in which the first communication device performs model training on the first model is as follows:

[0301] In a possible implementation, the first communication apparatus calculates a loss value of the third data and the fourth data, taking the third data obtained by the first communication apparatus based on the first data as a true value, where the loss value can be cross entropy, mean square error, normalized mean square error, cosine similarity, cosine similarity square, etc., and further trains the first model in the first communication apparatus according to the loss value.

[0302] In another possible implementation, the first communication apparatus can jointly perform training of the first model with other apparatuses, including but not limited to an over-the-top (OTT) server, an edge computing device, a cloud computing platform, or an external service board of the first communication apparatus, to reduce the processing overhead of the first communication apparatus. Taking the OTT server as an example, the training process is as follows: the first communication apparatus takes the third data obtained by the first communication apparatus based on the first data as a true value, determines the third data and the second data, and sends the third data and the second data to the OTT server associated with the first communication apparatus, so that the OTT server performs training of the first model. Optionally, the first communication apparatus sends the first intermediate data generated by the first processing and the second data, or the first data and the second data, to the OTT server associated with the first communication apparatus, so that the OTT server performs training of the first model.

[0303] The specific manner in which the first communication apparatus performs model monitoring on the first model is as follows:

[0304] In a possible implementation, after receiving the first data and the second data, the first communication apparatus processes the first data using the first processing to obtain a true value, that is, the third data, and processes the second data using the second processing to obtain the fourth data. Then, the first communication apparatus compares the true value and the fourth data to determine a model performance indicator, which can be prediction accuracy, block error rate, bit error rate, mean square error, normalized mean square error, cosine similarity, or cosine similarity square, to achieve model performance monitoring on the first model.

[0305] In another possible implementation, after receiving the first data, the first communication apparatus processes the first data using the first processing to obtain a true value, that is, the third data, and processes the first data using the second processing to obtain the seventh data. Then, the first communication apparatus compares the third data and the seventh data to determine a model performance indicator, which can be prediction accuracy, block error rate, bit error rate, mean square error, normalized mean square error, cosine similarity, or cosine similarity square, to achieve model performance monitoring on the first model.

[0306] In combination with the foregoing embodiments, an example scenario is as follows, please refer to FIG. 8, which is a schematic diagram of an application scenario of an embodiment of the present application. In the scenario shown in FIG. 8, the second communication device generates first data and second data at the physical layer, and the first data and the second data are carried at the physical layer. The first communication device receives the first data and the second data at the physical layer. Then, the first communication device performs data processing on the first data and the second data at the physical layer to obtain third data and fourth data. Further, the first communication device performs model training or model monitoring on the first model using the third data and the fourth data, and the first model is used to process the physical layer data of the first communication device. Correspondingly, the second model is used to process the physical layer data of the second communication device.

[0307] In the foregoing technical solution, since the first data and the second data are carried at the physical layer, the first data and the second data can be directly processed at the physical layer without the need for processing through a high-layer protocol, and therefore the first data and the second data have the characteristic of small data volume, thereby reducing the processing overhead of the first communication device for obtaining the first data and the second data and the processing overhead of performing model training or model monitoring based on the first data and the second data, and also reducing the communication overhead of the first communication device. Since the first model is used to process the physical layer data of the first communication device, after the first communication device receives the first data and the second data, the first data and the second data do not need to be processed by other transmission layers but are directly used for model training or model monitoring of the first model, which can effectively save the processing overhead of the first communication device and improve the efficiency of model training or model monitoring.

[0308] In combination with the foregoing embodiments, the information that can be interacted between the first communication device and the second communication device is introduced as follows. Please refer to FIG. 9, which is a schematic diagram of another embodiment of a communication method of an embodiment of the present application. The communication method proposed in the embodiment of the present application further includes the following steps:

[0309] D0, the first communication device sends the capability information of the first communication device to the second communication device.

[0310] In step D0, the first communication device can actively report the capability information of the first communication device to the second communication device, or the second communication device can request the first communication device to report the capability information of the first communication device, and the present application does not limit this.

[0311] The capability information of the first communication device indicates one or more of the following: whether the first communication device supports model training or model monitoring based on the first data and the second data; whether the first communication device supports reserving wireless resources for the second data; requirements of the first communication device on time domain resources of the first data and the second data, such as a transmission period, an interval of the first data and the second data; accuracy of the first communication device in performing channel estimation based on the first reference signal; conditions of the first communication device in performing channel estimation based on the first reference signal, the conditions including one or more of the following: a signal-to-noise ratio (SNR), a delay spread requirement, a mean delay requirement, a Doppler shift requirement, or a Doppler spread requirement; a data format of input data and / or output data supported by the first communication device, the data format including one or more of the following: a data length, or a size of the data; whether the first communication device supports feeding back gradient information related to the first model; a time length of the first communication device in updating the gradient information related to the first model; or a time length of the first communication device in feeding back the gradient information related to the first model. The following are described respectively.

[0312] The first communication device supports model training or model monitoring based on the first data and the second data, including: the first communication device supports model training or model monitoring of the first model based on the first data and the second data; the first communication device does not support model training or model monitoring of the first model based on the first data and the second data; a type of the first model that the first communication device supports model training or model monitoring based on the first data and the second data; a type of the first model that the first communication device does not support model training or model monitoring based on the first data and the second data, and the type of the first model can be replaced by a first process corresponding to the first model.

[0313] The second information element is that the first communication device supports reserving wireless resources for the second data, including: the first communication device can reserve wireless resources for the second data in each transmission, which can be PDSCH transmission, and the wireless resources include but are not limited to: the number of REs and / or the location of REs. Optionally, the wireless resources can also include the number of antenna ports and / or the number of antenna ports. Exemplarily, the capability information includes the number range of REs, which means that the first communication device supports reserving REs in the number range of REs for the second data, and the number range of REs includes the maximum value of the number of REs and the minimum value of the number of REs. In another example, the first communication device and the second communication device agree on the level corresponding to the number range of REs, and the capability information includes the level information currently supported by the first communication device, which indicates that the first communication device supports reserving the number range of REs for the second data. In another example, the first communication device and the second communication device agree on the maximum number of REs that can be reserved for the second data, and the capability information indicates whether the first communication device supports the maximum number of REs. The location of REs indicates the time domain resource, frequency domain resource and / or space domain resource where the first communication device supports reserving REs for the second data, for example, the location of REs indicates the frequency band, subcarrier, time slot, symbol and / or antenna port where the reserved REs are located.

[0314] The third information element is the time domain resource requirement of the first communication device for the first data and the second data, including: the time length required by the first communication device to process the first data and / or the second data, which can be in units of time slots, symbols, or subframes, etc. Alternatively, the first communication device requests the second communication device to periodically send the first data and / or the second data at a first time interval, or the first communication device requests the second communication device to periodically send the first data and / or the second data at a time interval greater than or equal to the first time interval, and the first time interval can also be in units of time slots, symbols, or subframes, etc.

[0315] The fourth information element is the accuracy of the first communication device performing channel estimation based on the first reference signal, which refers to the error range supported by the first communication device for performing channel estimation based on the first reference signal, or the quantized error range level. The error can be expressed in the form of estimation accuracy, mean square error, or normalized mean square error, etc.

[0316] The fifth information element is the condition of the first communication device performing channel estimation based on the first reference signal, which includes one or more of the following: signal-to-noise ratio (SNR), channel characteristic requirement, delay spread requirement, average delay requirement, Doppler shift requirement, average delay requirement, Doppler shift requirement, or Doppler spread requirement. The condition indicates that the first communication device needs to meet the above conditions, and the channel estimation result generated by the first communication device based on the first reference signal meets the accuracy requirement of model training or model monitoring of the first model.

[0317] The sixth information element is a data format of input data and / or output data supported by the first communication device, which includes one or more of the following: data length, or size of the data, etc. The data length or size of the data can refer to the data length or size of the data itself, for example, when the data is a bit stream, the data length or size can refer to the length of the bit stream or the amount of data composed by the bit stream. The data length or size can also refer to the length or size of the radio resource carrying the data, for example, the number of REs carrying the data. The input data and / or output data supported by the first communication device refers to the input data and / or output data supported by the first model in the first communication device.

[0318] The seventh information element is whether the first communication device supports feedback of gradient information related to the first model, which specifically includes: the first communication device supports feedback of gradient information related to the first model, or the first communication device does not support feedback of gradient information related to the first model. Optionally, when the first communication device supports feedback of gradient information related to the first model, it can further include: the first communication device supports real-time feedback of gradient information related to the first model, or the first communication device does not support real-time feedback of gradient information related to the first model. Whether the first communication device supports real-time feedback of gradient information related to the first model depends on the processing capability of the first communication device. For example, the second communication device periodically transmits first data and second data to the first communication device in one or more time slots in a time slot granularity. When the first communication device supports real-time feedback of gradient information related to the first model, the first communication device transmits gradient information related to the first model to the second communication device in one or more time slots in a time slot granularity. The time slots in which the first data and the second data are transmitted and the time slots in which the gradient information is reported can be configured to have an association relationship.

[0319] The eighth information element is a time length for the first communication device to update gradient information related to the first model, which indicates the time length required by the first communication device to update the gradient information related to the first model. The time length can be in various time units, for example: time slot, symbol, or subframe, etc.

[0320] The time length of the first communication device feeding back the gradient information related to the first model indicates a time length of the first communication device or other device assisting the first communication device in calculating the gradient information feeding back the gradient information, the other device being, for example, an OTT server, an edge computing node, a cloud computing platform, or the like. In the case where the time length indicates a time length of the other device assisting the first communication device in calculating the gradient information feeding back the gradient information, the time length can include a time length required by the other device assisting the first communication device in calculating the gradient information to update the gradient information related to the first model and / or a communication time length required by the other device assisting the first communication device in calculating the gradient information to interact with the first communication device. Similarly, the time length can be in various time units, for example, a time slot, a symbol, or a subframe, or the like.

[0321] Optionally, the capability information of the first communication device can also be predefined by a protocol, in which case the first communication device does not need to send the capability information of the first communication device to the second communication device.

[0322] D1, the second communication device sends first information to the first communication device, the first information indicating that the first data and the second data have an association relationship.

[0323] In step D1, the second communication device can send the first information to the first communication device through various messages, signaling or information, for example, the second communication device sends downlink control information (DCI) to the first communication device, and the DCI carries the first information.

[0324] The first information indicating that the first data and the second data have an association relationship can be replaced by: the first information indicating that the first data and the second data are used for model training or model monitoring of the first model; or the first information is used to indicate that the first data is used to determine the true value of the model training or model monitoring of the first model, and the first information is also used to indicate that the received second data is used to determine the model input data of the first model.

[0325] In addition to the direct indication manner, the first information can also indicate the association relationship between the first data and the second data in a multi-level indication manner. For example, the first information includes: first sub-information and second sub-information. The first sub-information is used to indicate one or more groups of association relationships, and the association relationship includes: the association relationship between the data used to determine the model input data of the first model and the data used to determine the true value. The second sub-information is used to indicate the target association relationship, and the target association relationship is that the first data and the second data have an association relationship. The target association relationship belongs to one or more groups of association relationships. The first sub-information and the second sub-information can be carried in the same message, signaling or information. The first sub-information and the second sub-information can also be carried in different messages, signaling or information. For example, the first sub-information is carried in a radio resource control (RRC) message, and the second sub-information is carried in a media access control element (MAC CE) message or DCI. For example, the first sub-information is shown in Table 2 or Table 3.

[0326] Table 2

[0327] Table 3

[0328] In combination with Table 2, if the target association relationship is the association relationship corresponding to ID1, that is, the target association relationship is that the first data is carried in TB1 and is the first N1 bits of TB1, and the second data is carried in TB2, the second sub-information includes ID1.

[0329] In combination with Table 3, if the target association relationship is the association relationship corresponding to ID5, that is, the target association relationship is that the first data is carried in the code word determined by TB1 and is the first N2 bits of the code word, and the second data is also derived from the code word, the second sub-information includes ID5.

[0330] Optionally, the first information can also include an index of the RE pattern carrying the second data.

[0331] When the first data comprises a first reference signal and the second data comprises a second reference signal, the first information can further be used to indicate that the first reference signal has a correlation relationship with the second reference signal. For example, the first information comprises: a RE position carrying the second reference signal, and / or an index of the first reference signal having the correlation relationship with the second reference signal. The first information indicating that the first reference signal has the correlation relationship with the second reference signal can be replaced by: the first information indicating that an estimation result determined by the first reference signal and the second reference signal are used for model training or model monitoring of the first model; or the first information being used to indicate that the estimation result determined by the first reference signal is used to determine a true value of the model training or the model monitoring of the first model, and the first information is further used to indicate that the received second reference signal is used to determine model input data of the first model.

[0332] Optionally, in addition to directly indicating that the first reference signal has the correlation relationship with the second reference signal, the first information can also indicate that the first reference signal has the correlation relationship with the second reference signal through a multi-level indication manner. For example, the first information comprises: third sub-information and / or fourth sub-information, the third sub-information is used to indicate one or more groups of correlation relationships, and the correlation relationship comprises: a reference signal in a non-AI processing manner has a correlation relationship with a reference signal in an AI processing manner; the fourth sub-information comprises a target correlation relationship, the target correlation relationship indicates that the first reference signal has the correlation relationship with the second reference signal, and the target correlation relationship belongs to one or more groups of correlation relationships. The third sub-information and the fourth sub-information can be carried in the same message, signaling or information; or the third sub-information and the fourth sub-information can be carried in different messages, signaling or information. For example, the third sub-information is carried in an RRC message, and the fourth sub-information is carried in a MAC CE message or a DCI. For example, the third sub-information is shown in Table 4.

[0333] Table 4

[0334] In combination with Table 4, if the target correlation relationship is the correlation relationship corresponding to ID5, i.e., the target correlation relationship is that the first reference signal is legacyRS2 and the second reference signal is AI RS2, the fourth sub-information comprises ID5.

[0335] Optionally, the first information can further comprise fifth sub-information, the fifth sub-information is used to indicate a subset of the one or more groups of correlation relationships. In this case, the target correlation relationship indicated by the second sub-information or the fourth sub-information belongs to the subset indicated by the fifth sub-information. For example, the fifth sub-information can be carried in a MAC CE message, the third sub-information is carried in an RRC message, and the fourth sub-information is carried in a DCI.

[0336] Optionally, the first information comprises one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data. Optionally, the first information can further comprise: a RE position carrying the first data, or a RE position carrying the second data.

[0337] Further optionally, the first information can further comprise one or more of the following: identification information of a data set comprising the first data and / or the second data; identification information of the first model; identification information of a second model used by the second communication device, the second model being used to generate the second data; or identification information of a model set comprising one or more models, the one or more models comprising the first model.

[0338] For example, the first data and the second data are obtained through multiple interactions between the first communication device and the second communication device, and the first data and the second data can be regarded as a data set. Then, the first communication device or other devices assisting the first communication device in training can use data of the data set to uniformly perform model training or model monitoring of the first model. At this time, the second communication device needs to send identification information of the data set to the first communication device when sending the first data and the second data.

[0339] For another example, when the first model and the second model are paired models, i.e., the first communication device and the second communication device use the same shared model, the identification information of the first model is equivalent to the identification information of the second model, and the identification information of the first model can be replaced by identification information of the paired model or identification information of the shared model.

[0340] D2, the second communication device sends second information to the first communication device, the second information being used to indicate resources carrying the second data.

[0341] In step D2, the second communication device sends second information to the first communication device, the second information being used to indicate resources carrying the second data, and the second information is specifically used to indicate one or more of the following resources: frequency domain resources carrying the second data, time domain resources carrying the second data, or space domain resources carrying the second data. The frequency domain resources carrying the second data comprise: a frequency band, a subcarrier, a resource element (RE), or a RE pattern, wherein the RE pattern comprises one or more REs; the time domain resources carrying the second data comprise: a slot, a symbol, or a subframe; and the space domain resources carrying the second data comprise: an antenna port.

[0342] Exemplarily, when the second communication device is an access network device and the first communication device is a terminal device, the second information can be carried in the DCI.

[0343] D3, the second communication device sends third information to the first communication device, the third information being used to indicate the relative position of the first data in the transmission data.

[0344] In step D3, the second communication device informs the first communication device of the relative position of the first data in the transmission data through the third information. The third information includes one or more of the following information: identification information of a TB carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data. If the first data is composed of multiple discontinuous data segments, the third information can indicate the starting position and the ending position of the multiple discontinuous data segments. For example, the third information includes TB1 and the first 100 bits, and the meaning of the third information is that the transmission data or the first data is carried in TB1, and the first 100 bits of the TB1 are used to determine the first data.

[0345] It can be understood that step D3 is an optional step, and when the first information is also used to indicate the relative position of the first data in the transmission data, step D3 is not performed.

[0346] D4-1, the second communication device sends fourth information to the first communication device, the fourth information being used to instruct the first communication device to feed back gradient information related to the first model.

[0347] In step D4-1, the second communication device can also send fourth information to the first communication device, the fourth information being used to instruct the first communication device to feed back gradient information related to the first model. The gradient information refers to gradient information generated in the training process of the first model, which can be determined based on the loss function value in the training of the first model, for example, determined through a back propagation method.

[0348] Optionally, the fourth information includes time domain resources carrying the gradient information. For example, the fourth information indicates that the time slots carrying the gradient information are uplink time slots 1 and 2. The first communication device sends the gradient information to the second communication device in the uplink time slots 1 and 2.

[0349] Optionally, the fourth information further comprises an association relationship between the time domain resource carrying the gradient information and the time domain resource of the second data on which the gradient information is based. For example, the fourth information can be used to configure the time slots carrying the gradient information, and the fourth information can also be used to configure the association relationship between the time slots carrying the gradient information and the time slots carrying the first data and the second data. For example, the fourth information indicates that the time slots carrying the first data and the second data are downlink time slots 1, 2 and 3, and the fourth information further indicates that the time slot carrying the gradient information corresponding to the downlink time slots 1 and 2 is uplink time slot 1, and the time slot carrying the gradient information corresponding to the downlink time slot 3 is uplink time slot 2, then the downlink time slots 1, 2 and the uplink time slot 1 have an association relationship, and the downlink time slot 3 and the uplink time slot 2 have an association relationship.

[0350] Optionally, the fourth information can also not indicate the time domain resource carrying the gradient information, and / or the association relationship between the time domain resource carrying the gradient information and the time domain resource of the second data on which the gradient information is based. For example, the association relationship between the time domain resource carrying the gradient information and the time domain resource of the second data on which the gradient information is based is preconfigured or pre-defined by a protocol. Then the first communication device determines the time domain resource carrying the first data and the second data according to the preconfigured or pre-defined information, and / or determines the time domain resource carrying the gradient information.

[0351] Optionally, the fourth information is further used to indicate the first data and the second data on which the gradient information is based. For example, the second communication device can configure the first communication device to feed back one gradient information based on a group of first data and second data. The second communication device can also configure the first communication device to feed back one gradient information based on multiple groups of first data and second data, so as to save communication overhead and improve model training efficiency and accuracy.

[0352] Optionally, for one or more downlink time slots carrying the first data and the second data corresponding to the same uplink time slot carrying the gradient information, for example, the downlink time slots 1, 2 and the uplink time slot 1 have an association relationship, then when the second data corresponding to the downlink time slots 1 and 2 is generated, the second communication device uses the same second model to generate the second data, or the first communication device can assume that the second model does not change when the second communication device generates the second data in the downlink time slots 1 and 2. By the above method, it is ensured that the gradient information of the first model generated by the first communication device corresponds to the same second model, and the accuracy of the gradient information is ensured.

[0353] In a possible implementation, the second communication device receives the capability information of the first communication device, and the capability information indicates that the first communication device supports feeding back the gradient information related to the first model. Based on the capability information, the second communication device sends the fourth information to the first communication device.

[0354] In another possible implementation, when the second communication device needs to update the second model, the second communication device sends fourth information to the first communication device.

[0355] D4-2, in response to the fourth information, the first communication device sends gradient information related to the first model to the second communication device.

[0356] In step D4-2, in response to the fourth information, the first communication device sends gradient information related to the first model to the second communication device. The gradient information is generated according to one or more sets of first data and second data. According to the fourth information, the first communication device can feed back one gradient information according to one set of first data and second data, or the first communication device can feed back one gradient information according to multiple sets of first data and second data.

[0357] When the fourth information indicates the time domain resource carrying the gradient information, and the first communication device supports real-time feedback of the gradient information, the first communication device feeds back the gradient information on the time domain resource. When the fourth information does not indicate the time domain resource carrying the gradient information, a predefined association relationship between the time domain resource carrying the gradient information and the time domain resource of the second data on which the gradient information is based is predefined, and the first communication device supports real-time feedback of the gradient information, the first communication device feeds back the gradient information on the predefined time domain resource carrying the gradient information.

[0358] When the first communication device does not support real-time feedback of the gradient information, the first communication device can also feed back the identification information of the first data and / or the identification information of the second data corresponding to the gradient information when feeding back the gradient information. The identification information of the first data can be the identification information of the data packet carrying the first data, the timestamp of the first data, or the sequence number of the data packet carrying the first data. Similarly, the identification information of the second data can be the identification information of the data packet carrying the second data, the timestamp of the second data, or the sequence number of the data packet carrying the second data. In this way, the second communication device can determine the first data and / or the second data corresponding to the gradient information according to the identification information of the first data and / or the identification information of the second data.

[0359] For example, the first communication device is a terminal device, and the second communication device is an access network device. The first communication device can feed back the gradient information to the second communication device through uplink control information (UCI). The first communication device can also feed back the gradient information to the second communication device through an RRC message. The first communication device can further feed back the gradient information to the second communication device through a logged minimization drive test (Logged MDT) message. In addition, the first communication device can also send the gradient information to the second communication device through a non-access stratum (NAS) message or a MAC CE message.

[0360] Optionally, when any one of the first data and the second data received by the first communication device does not satisfy the validity condition of the first data and the second data, for example, decoding of the first data fails, the first communication device does not use the first data and the second data to obtain the gradient information. For all the first data and all the second data corresponding to the same time domain resource carrying the gradient information, when all the first data and all the second data received by the first communication device do not satisfy the validity condition of the first data and the second data, the first communication device can feed back, for the first data and the second data, the gradient information in the time domain resource carrying the gradient information, which can be an invalid value agreed between the first communication device and the second communication device. For example, downlink time slots 1 and 2 and uplink time slot 1 have an association relationship. When all the first data and all the second data received in the downlink time slots 1 and 2 do not satisfy the validity condition of the first data and the second data, the first communication device can feed back, in the uplink time slot 3, the gradient information, which can be an invalid value agreed between the first communication device and the second communication device.

[0361] Through the method in steps D4-1 to D4-2, the first communication device can realize reverse updating of the gradient information, and improve the training effect of the second model by the second communication device under the condition of lower communication overhead.

[0362] D5, the second communication device sends fifth information to the first communication device, and the fifth information is used to configure the validity condition of the first data and the second data.

[0363] In step D5, the second communication apparatus can send fifth information to the first communication apparatus, the fifth information being used for configuring an effective condition of the first data and the second data, the effective condition including one or more of the following: successful channel decoding of the transmitted data; successful demodulation of the transmitted data; or, a condition of channel estimation of the first data satisfying a certain threshold requirement, the condition of channel estimation including one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), or delay spread, which refers to the maximum time difference of arrival of signals transmitted between the second communication apparatus and the first communication apparatus via different paths, average delay, Doppler spread, and / or Doppler shift.

[0364] It should be noted that the execution order of steps D1 to D5 is not limited in the embodiments of the present application. Steps D1 to D5 are optional steps.

[0365] In combination with the foregoing embodiments, the following describes various application scenarios related to the embodiments of the present application. Taking the second processing as AI demodulation, the fourth processing as AI modulation; the second processing as channel estimation based on the second reference signal, the fourth processing as inserting the second reference signal; the second processing as AI channel decoding, and the fourth processing as AI channel encoding as examples, the corresponding application scenarios are described respectively. It should be noted that in the following application scenarios, the first communication apparatus is taken as a terminal device, and the second communication apparatus is taken as an access network device for illustration.

[0366] Application scenario one: the second processing is AI demodulation, and the fourth processing is AI modulation.

[0367] Please refer to FIG. 10, which is a schematic diagram of an application scenario in the embodiments of the present application. In this application scenario, the second model of the second communication apparatus performs the fourth processing as AI modulation, and the first model of the first communication apparatus performs the second processing as AI demodulation. At present, the modulation modes of non-AI processing include QPSK or 16QAM, etc. The access network device can notify the terminal device of the specific modulation mode, and the terminal device demodulates the received data according to the modulation mode. After AI modulation is adopted, the AI modulation can change with the continuous change of the channel, for example, the modulated constellation points / constellation symbols are no longer uniformly distributed constellation points similar to QPSK or 16QAM, thus having better performance. Since the constellation points or constellation symbols formed by AI modulation are dynamically changed, the first model performing AI demodulation in the first communication apparatus needs to be matched with the second model performing AI modulation in the second communication apparatus, so that the first communication apparatus can successfully demodulate the data processed by AI modulation using the first model.

[0368] To achieve the above purpose, the second communication apparatus processes as follows:

[0369] Referring to FIG. 11, FIG. 11 is a schematic diagram of one scenario of a data processing manner in an embodiment of the present application. The original data is "11000···11", which can be user plane data or control plane data, for example, MAC PDU data delivered by a MAC layer to a physical layer. After channel coding processing on the original data, the second intermediate data "001111····0011" is obtained, which can be non-AI channel coding, for example, LDPC coding or Polar coding. The first 6 bits of the second intermediate data "001111····0011" are copied to obtain the fifth data "001111". The AI modulation is performed on the fifth data to generate the second data. The conventional modulation is performed on the second intermediate data "001111····0011" to generate the transmission data, wherein the first 6 bits of the second intermediate data "001111····0011" are used as the sixth data, and therefore the transmission data includes the first data, which refers to the data generated by the conventional modulation processing on the sixth data. The conventional modulation can also be referred to as non-AI modulation, for example, 16QAM. The second data shown in FIG. 11 is carried in an RE pattern composed of 2*2 squares, and the RE pattern is configured with corresponding time domain positions and frequency domain positions.

[0370] Referring to FIG. 12, FIG. 12 is a schematic diagram of another scenario of model training in an embodiment of the present application. The second communication device generates the transmission data using the conventional modulation processing, and the transmission data includes the first data. Then, the second communication device sends the transmission data and the second data to the first communication device. The second communication device sends the first information, the second information and the third information to the first communication device. According to the second information and the third information, the first communication device determines the second data and the first data in the transmission data in the received data. According to the first information, the first communication device determines that the first data and the second data have a correlation relationship. The first communication device performs the conventional demodulation on the first data, that is, adopts the non-AI processing demodulation manner to demodulate the first data, and then performs channel decoding to obtain the first intermediate data. In the case that the first intermediate data passes the check (that is, the transmission data successfully performs the channel decoding), the first intermediate data is processed by channel re-encoding, and the third data is obtained according to the first information or the third information, which can be used as the true value or label value of the model training of the first model. The first communication device performs the AI demodulation on the second data to obtain the fourth data, which is used as the model output data of the first model. The first communication device determines the loss value of the first model according to the third data and the fourth data, and then trains the first model, wherein the loss value can be cross entropy.

[0371] Application scenario two: the second processing is the generation of the second reference signal, and the fourth processing is the channel estimation based on the second reference signal.

[0372] Please refer to FIG. 13, which is a schematic diagram of another application scenario in the embodiments of the present application. In this application scenario, the fourth processing performed by the second model of the second communication device is to generate a second reference signal, and the second processing performed by the first model of the first communication device is channel estimation based on the second reference signal. At present, the non-AI reference signal (for example, the reference signal sequence content, the generation method) of this transmission can be notified in advance between the first communication device and the second communication device, so that the first communication device can perform channel estimation based on the non-AI reference signal to obtain channel information, but the design of the non-AI reference signal does not combine the estimation process of the reference signal. The second reference signal based on AI processing adopts an end-to-end optimization method to jointly optimize the reference signal itself and the estimation process, so it has better performance. In order for the first communication device to obtain correct channel information based on the channel estimation of the second reference signal, the first model in the first communication device that performs channel estimation based on the second reference signal needs to be matched with the second model in the second communication device that generates the second reference signal, so that the first communication device can use the first model to complete the channel estimation based on the second reference signal.

[0373] In order to achieve the above-mentioned purpose, the processing of the second communication device is as follows:

[0374] Please refer to FIG. 14, which is a schematic diagram of one scenario of data processing method in the embodiments of the present application. After the second communication device generates the second reference signal using the second model, the second communication device embeds the second reference signal into the reserved RE. Optionally, the second communication device embeds the first reference signal near the RE of the second reference signal to improve the accuracy of obtaining the true value. The second communication device sends the first information to the first communication device, and the first information indicates that the first reference signal and the second reference signal have a correlation relationship, and the correlation relationship can be that the channel estimation result based on the first reference signal can be used as the true value of the channel estimation result based on the second reference signal.

[0375] The second communication device sends transmission data to the first communication device, the transmission data including first data (first reference signal) and second data (second reference signal). For ease of understanding, please refer to FIG. 15, which is a schematic diagram of model training in an embodiment of the present application. After the first communication device receives the transmission data, the first communication device performs channel estimation based on the first reference signal on the received transmission data according to the first reference signal to obtain third data, for example, the third data is a channel estimation result corresponding to a wireless resource carrying the transmission data; and performs channel estimation based on the second reference signal on the received transmission data using the first model according to the second reference signal to obtain fourth data, for example, the fourth data is a channel estimation result corresponding to a wireless resource carrying the transmission data estimated using the first model. Then the first communication device determines a loss value of the first model according to the third data and the fourth data, and further trains the first model, wherein the loss value can be a normalized mean square error.

[0376] Optionally, the validity condition of the first reference signal and the second reference signal includes that the SNR or SINR in the channel estimation process based on the first reference signal is greater than or equal to a first threshold, or the delay spread in the channel estimation process based on the first reference signal is less than or equal to a second threshold, or the average delay in the channel estimation process based on the first reference signal is less than or equal to a third threshold, or the Doppler shift in the channel estimation process based on the first reference signal is less than or equal to a fourth threshold, or the Doppler spread in the channel estimation process based on the first reference signal is less than or equal to a fifth threshold. Wherein the SNR, SINR, delay spread, average delay, Doppler shift, and Doppler spread in the channel estimation process based on the first reference signal can be determined according to information such as the environment, motion state of the first communication device, or estimation result of other reference signals. When the validity condition is met, that is, the condition of performing channel estimation on the first data meets certain threshold requirements, the association relationship between the first reference signal and the second reference signal is established, and the third data obtained based on the channel estimation of the first reference signal can be used as the true value for model training or model monitoring of the first model.

[0377] Application scenario three: the second processing is AI channel decoding, and the fourth processing is AI channel encoding.

[0378] Please refer to FIG. 16, which is a schematic diagram of another application scenario in the embodiments of the present application. In this application scenario, the second model of the second communication device performs the fourth process as AI channel encoding, and the first model of the first communication device performs the second process as AI channel decoding. The existing channel encoding mode includes low-density parity-check (LDPC) code or Polar code. In the existing channel encoding mode, the receiver performs channel decoding on the demodulated data according to different code block lengths and channel encoding modes. After AI, the AI channel encoding mode is no longer agreed by the protocol. AI channel encoding and AI channel decoding have better performance than the existing channel encoding and channel decoding.

[0379] For ease of understanding, please refer to FIG. 17, which is a schematic diagram of model training in the embodiments of the present application. In the second communication device, the sixth data carried in TB1 is subjected to LDPC channel encoding to obtain first data, which is carried in code word 1, which is the data obtained by subjecting TB1 to LDPC channel encoding; the fifth data carried in TB2 is subjected to AI channel encoding to obtain second data, which is carried in code word 2, wherein code word 2 is the data obtained by subjecting TB2 to AI channel encoding. After the first communication device receives the first data and the second data, it performs LDPC channel decoding on the first data to obtain third data, and performs AI channel decoding on the second data to obtain fourth data. Then, in the case that the third data passes the check (i.e., the transmission data performs channel decoding successfully), the first communication device determines the loss value of the first model according to the third data and the fourth data, and then trains the first model.

[0380] It can be understood that, when the above application scenario is used for model monitoring, the first communication device can determine the model performance index of the first model according to the third data and the fourth data, and then monitor the first model. The type of the model performance index of the first model can be indicated by the second communication device, or determined in a pre-defined or pre-configured manner through the protocol.

[0381] In the above application scenario, the specific information exchanged between the second communication device and the first communication device, the processing mode of the first communication device for the first data and the second data, and the specific manner in which the first communication device determines the first data in the transmission data are described in the foregoing embodiments, and will not be repeated here.

[0382] In the application scenario, the second communication device can normally transmit the transmission data to ensure normal operation of the service, and the first communication device can obtain the first data required for model training of the first model through the transmission data, thereby saving communication overhead. In addition, the second communication device uses the existing data transmission mode to assist the first communication device to complete the model training or model monitoring of the first model, thereby reducing air interface overhead and air interface complexity.

[0383] It can be understood that the above application scenario can also be combined for implementation. For example, the AI processing of the first communication device includes AI demodulation, channel estimation based on the second reference signal, and / or AI channel decoding; and correspondingly, the AI processing of the second communication device includes AI modulation, generation of the second reference signal, and / or AI channel encoding.

[0384] In combination with the foregoing embodiments, another method embodiment related to the present application will be introduced next. Please refer to FIG. 18, which is a flowchart of an embodiment of a communication method according to the present application. The communication method proposed in the embodiment of the present application includes the following steps:

[0385] F0, the first communication device sends capability information of the first communication device to the second communication device.

[0386] Step F0 is an optional step.

[0387] For step F0, please refer to the foregoing step D0, which will not be repeated here.

[0388] F1, the second communication device sends fourth information to the first communication device, the fourth information being used to instruct the first communication device to feed back gradient information related to the first model.

[0389] F2, the first communication device sends the gradient information related to the first model to the second communication device.

[0390] For steps F1-F2, please refer to the foregoing steps D4-1-D4-2, which will not be repeated here.

[0391] It should be noted that in the embodiment shown in FIG. 18, the first data and the second data can be carried in a high layer, such as an application layer, in addition to being carried in a physical layer, and the embodiments of the present application do not limit this.

[0392] For ease of understanding, in combination with the corresponding embodiment of FIG. 18, an application scenario related to the embodiment shown in FIG. 18 is introduced. Please refer to FIG. 19, which is a schematic diagram of another application scenario in the embodiment of the present application. Take an example in which the fourth processing includes AI modulation and the second processing includes AI demodulation. The second communication device performs channel encoding on the original data to obtain second intermediate data after channel encoding, and the second intermediate data includes the sixth data. The second communication device performs traditional non-AI modulation processing on the second intermediate data to obtain transmission data, wherein the second communication device performs traditional non-AI modulation processing on the sixth data in the second intermediate data to obtain the first data, and the transmission data includes the first data. The second communication device performs copy processing on the sixth data in the second intermediate data to obtain the fifth data, and then performs AI modulation processing on the fifth data to generate the second data.

[0393] In a possible implementation, the second communication device implements the transfer of the true value by sending the first data to the first communication device. After receiving the first data and the second data, the first communication device performs corresponding processing to obtain the third data and the fourth data. Then, the first communication device performs model training of the first model using the third data and the fourth data to obtain gradient information related to the first model. The first communication device feeds back the gradient information to the second communication device.

[0394] In another possible implementation, the second communication device sends the sixth data to the first communication device to implement the transfer of the true value, for example, sending the sixth data in a high layer or an application layer. The first communication device performs model training of the first model according to the sixth data and the fourth data to obtain gradient information related to the first model, and the fourth data is data obtained by the first communication device by processing the second data from the second communication device. The first communication device feeds back the gradient information to the second communication device.

[0395] In the above technical solution, the first communication device can implement reverse update of the gradient information, and under the condition of lower communication overhead, the training effect of the second communication device on the second model is improved.

[0396] In combination with the foregoing embodiments, the following describes another method embodiment involved in the present application. Please refer to FIG. 20, which is a schematic diagram of a communication method in the embodiment of the present application. The communication method proposed in the embodiment of the present application includes:

[0397] Step G1, the first communication device reports capability information to the second communication device. The capability information indicates whether the first communication device has the capability of receiving a data set (y, x) over the air, y is output data of an AI transmitter in the second communication device, y is similar to the second data in the foregoing embodiments, x is output data of an AI receiver in the second communication device, and x is similar to the true value of the second data determined according to the first data in the foregoing embodiments.

[0398] Optionally, the capability information can also indicate a size of a data set supported by the first communication device to receive in each transmission.

[0399] Step G1 is an optional step.

[0400] Step G2, the second communication device generates a data set (y, x). The AI transmitter shown in the second communication device can perform any one or more AI processes shown in FIG. 20. Correspondingly, the AI receiver shown in the second communication device can perform any one or more AI processes shown in FIG. 20.

[0401] It should be noted that the channel between the AI transmitter and the AI receiver can be a channel generated based on real data acquisition, or a channel generated by simulation, and the embodiments of the present application do not limit this.

[0402] Step G3, the second communication device sends the data set (y, x) to the first communication device.

[0403] For example, the data set (y, x) is sent to the first communication device through an air interface, such as a high layer signaling carrying the data set (y, x), which can be an RRC message or a NAS message. The high layer signaling can also include channel information.

[0404] For another example, the second communication device sends the data set (y, x) to the first communication device in the form of application layer data.

[0405] Optionally, the second communication device can also send the data set (y, x) to the first communication device through a non-air interface.

[0406] Next, the communication device related to the embodiments of the present application is introduced. The communication device can be used as the first communication device and / or the second communication device in the foregoing embodiments.

[0407] FIG. 21 is a structural schematic diagram of a communication device according to an embodiment of the present application. Please refer to FIG. 21, the communication device 2100 includes a transceiver module 2101 and a processing module 2102.

[0408] The communication device 2100 includes an access network device, which can be the first communication device and / or the second communication device. Alternatively, the communication device 2100 includes components (for example, chips), modules or units in a terminal device, and the access network device can be the first communication device and / or the second communication device. Alternatively, the communication device 2100 includes components (for example, chips), modules or units in a core network device, and the core network device can be the core network device.

[0409] The communication apparatus 2100 can be configured to perform all or part of the steps executed by the first communication apparatus in the embodiments of FIGS. 5-20. For details, refer to related description in the foregoing embodiments of FIGS. 5-20.

[0410] The communication apparatus 2100 can be configured to perform all or part of the steps executed by the second communication apparatus in the embodiments of FIGS. 5-20. For details, refer to related description in the foregoing embodiments of FIGS. 5-20.

[0411] The processing module 2102 is configured to perform data processing. The transceiver module 2101 is configured to implement corresponding communication functions.

[0412] Optionally, the transceiver module 2101 can include a sending module and a receiving module. The sending module is configured to perform the sending operations in the above method embodiments. The receiving module is configured to perform the receiving operations in the above method embodiments.

[0413] It should be noted that the communication apparatus 2100 can include the sending module and not include the receiving module. Alternatively, the communication apparatus 2100 can include the receiving module and not include the sending module. Specifically, whether the sending module and the receiving module are included in the communication apparatus 2100 can depend on whether the sending action and the receiving action are included in the above schemes executed by the communication apparatus 2100.

[0414] Optionally, the communication apparatus 2100 can further include a storage module, which can be configured to store instructions and / or data. The processing module 2102 can read the instructions and / or data in the storage module, so that the communication apparatus 2100 implements the foregoing method embodiments.

[0415] The communication apparatus 2100 can be configured to perform the actions performed by the first communication apparatus in the embodiments of FIGS. 5-20. The processing module 2102 is configured to perform processing-related operations of the first communication apparatus in the embodiments of FIGS. 5-20. The transceiver module 2101 is configured to perform receiving- or sending-related operations of the first communication apparatus in the embodiments of FIGS. 5-20.

[0416] The communication apparatus 2100 can be configured to perform the actions performed by the second communication apparatus in the embodiments of FIGS. 5-20. The processing module 2102 is configured to perform processing-related operations of the second communication apparatus in the embodiments of FIGS. 5-20. The transceiver module 2101 is configured to perform receiving- or sending-related operations of the second communication apparatus in the embodiments of FIGS. 5-20.

[0417] For example, the communication apparatus 2100 is configured to implement the following schemes.

[0418] In an example, when the communication apparatus 2100 is applied to the first communication apparatus, the communication apparatus 2100 includes:

[0419] The transceiver module 2101 is configured to receive first data, the first data being carried in a physical layer.

[0420] The transceiver module 2101 is further configured to receive second data, the second data being carried in the physical layer, the first data and the second data having a correlation relationship, the first data being used to determine a true value of model training or model monitoring of a first model, the second data being used to determine model input data of the first model, and the first model being used to process physical layer data of the first communication device.

[0421] In a possible implementation manner,

[0422] The processing module 2102 is configured to perform model training or model monitoring on the first model according to the first data and the second data.

[0423] In a possible implementation manner,

[0424] The transceiver module 2101 is further configured to receive transmission data from the second communication device, the transmission data being determined based on user plane data and / or control plane data, and the first data being determined based on the transmission data.

[0425] In a possible implementation manner,

[0426] The transceiver module 2101 is further configured to receive first information, the first information being used to indicate that the first data and the second data have a correlation relationship.

[0427] In a possible implementation manner, the first information includes one or more of the following information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data.

[0428] In a possible implementation manner,

[0429] The transceiver module 2101 is further configured to receive second information, the second information being used to indicate one or more of the following resources: frequency domain resources carrying the second data, time domain resources carrying the second data, or spatial domain resources carrying the second data.

[0430] In a possible implementation manner,

[0431] The transceiver module 2101 is further configured to receive third information, the third information being used to indicate a relative position of the first data in the transmission data.

[0432] In a possible implementation, the third information includes one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data.

[0433] In a possible implementation, the first data further includes a first reference signal, and the first reference signal is used for channel estimation of a wireless resource carrying the first data; and the second data further includes a second reference signal, and the second reference signal is used for channel estimation of a wireless resource carrying the second data by the first model.

[0434] In a possible implementation, the first information is further used to indicate that the first reference signal and the second reference signal have an association relationship.

[0435] In a possible implementation,

[0436] The transceiver 2101 is further configured to receive fourth information, where the fourth information is used to indicate that the first communication device feeds back gradient information related to the first model.

[0437] The processing module 2102 is configured to send the gradient information according to the fourth information.

[0438] In a possible implementation,

[0439] The transceiver 2101 is further configured to receive fifth information, where the fifth information is used to configure an effective condition of the first data and the second data, and the effective condition includes one or more of the following: successful execution of channel decoding on the transmission data; successful demodulation of the transmission data; or a condition of performing channel estimation on the first data, where the condition of performing channel estimation includes one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), average delay, Doppler spread, Doppler shift, or delay spread.

[0440] In a possible implementation,

[0441] The processing module 2102 is configured to perform first processing on the first data to generate third data.

[0442] The processing module 2102 is configured to perform second processing on the second data using the first model to generate fourth data, where the third data is a true value of the fourth data.

[0443] The processing module 2102 is configured to determine a loss value of the first model based on the third data and the fourth data.

[0444] In a possible implementation manner,

[0445] The transceiver 2101 is further configured to send capability information of the first communication apparatus, the capability information of the first communication apparatus indicating one or more of the following information: whether the first communication apparatus supports model training or model monitoring based on the first data and the second data; whether the first communication apparatus supports frequency domain resources reserved for the second data; time domain resource requirements of the first communication apparatus for the first data and the second data; accuracy of channel estimation of the first communication apparatus based on the first reference signal; conditions of channel estimation of the first communication apparatus based on the first reference signal, the conditions including one or more of the following: signal-to-noise ratio (SNR), delay spread requirement, average delay requirement, Doppler shift requirement, or Doppler spread requirement; data format of input data and / or output data supported by the first communication apparatus; whether the first communication apparatus supports feedback of gradient information related to the first model; time length of the first communication apparatus for updating the gradient information related to the first model; or time length of the first communication apparatus for feeding back the gradient information related to the first model.

[0446] In yet another example, the communication apparatus 2100 is applied to a second communication apparatus, and the communication apparatus 2100 includes:

[0447] The transceiver 2101 is further configured to send, to the first communication apparatus, first data, the first data being carried in a physical layer.

[0448] The transceiver 2101 is further configured to send, to the first communication apparatus, second data, the second data being carried in the physical layer, the first data and the second data having a correlation relationship, the first data being used to determine a true value of model training or model monitoring of a first model, the second data being used to determine input data of the first model, and the first model being a model of the first communication apparatus.

[0449] In a possible implementation manner,

[0450] The transceiver 2101 is further configured to send, to the first communication apparatus, transmission data, the transmission data being determined based on user plane data and / or control plane data, and the first data being determined based on the transmission data.

[0451] In a possible implementation manner,

[0452] The transceiver 2101 is further configured to send, to the first communication apparatus, first information, the first information being used to indicate that the first data and the second data have a correlation relationship.

[0453] In a possible implementation, the first information comprises one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data.

[0454] In a possible implementation,

[0455] The transceiver 2101 is further configured to send, to the first communication apparatus, second information used to indicate one or more of the following resources: frequency domain resources carrying the second data, time domain resources carrying the second data, or spatial domain resources carrying the second data.

[0456] In a possible implementation,

[0457] The transceiver 2101 is further configured to send, to the first communication apparatus, third information used to indicate a relative position of the first data in the transmission data.

[0458] In a possible implementation, the third information comprises one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or a data amount occupied by the first data in the transmission data.

[0459] In a possible implementation, the first data further comprises a first reference signal used for channel estimation of a wireless resource carrying the first data, and the second data further comprises a second reference signal used for channel estimation of a wireless resource carrying the second data by the first model.

[0460] In a possible implementation, the first information is further used to indicate that the first reference signal and the second reference signal have an association relationship.

[0461] In a possible implementation,

[0462] The transceiver 2101 is further configured to send, to the first communication apparatus, fourth information used to indicate that the first communication apparatus feeds back gradient information related to the first model.

[0463] The transceiver 2101 is further configured to receive the gradient information from the first communication apparatus.

[0464] In a possible implementation,

[0465] The transceiver module 2101 is further configured to send fifth information to the first communication device, where the fifth information is used to configure an effective condition of the first data and the second data, and the effective condition comprises one or more of the following: successful channel decoding of the transmission data; successful demodulation of the transmission data; or, a condition of performing channel estimation on the first data, where the condition of performing channel estimation comprises one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), average delay, Doppler spread, Doppler shift, or delay spread.

[0466] In a possible implementation manner,

[0467] The transceiver module 2101 is further configured to send capability information of the first communication device to the first communication device, where the capability information of the first communication device indicates one or more of the following: whether the first communication device supports model training or model monitoring based on the first data and the second data; whether the first communication device supports frequency domain resources reserved for the second data; a requirement of the first communication device on time domain resources of the first data and the second data; an accuracy of the first communication device in performing channel estimation based on the first reference signal; a condition of the first communication device in performing channel estimation based on the first reference signal, where the condition comprises one or more of the following: SNR, delay spread requirement, average delay requirement, Doppler shift requirement, or Doppler spread requirement; a data format of input data and / or output data supported by the first communication device; whether the first communication device supports feedback of gradient information related to the first model; a time length of the first communication device in updating the gradient information related to the first model; or, a time length of the first communication device in feeding back the gradient information related to the first model.

[0468] In a possible implementation manner,

[0469] The processing module 2102 is further configured to perform third processing on the sixth data to generate the first data.

[0470] The processing module 2102 is further configured to determine fifth data from the sixth data, where the fifth data belongs to the sixth data.

[0471] The processing module 2102 is further configured to perform fourth processing on the fifth data to generate the second data.

[0472] For other implementation manners, refer to related descriptions in the foregoing embodiments shown in FIGS. 5-20, which will not be repeated here.

[0473] It should be understood that specific processes in which each module performs the corresponding processes described above have been described in detail in the foregoing method embodiments, which will not be repeated here for the sake of brevity.

[0474] The processing module 2102 in the above embodiment can be implemented by at least one processor or processor-related circuit. The transceiver module 2101 can be implemented by a transceiver or transceiver-related circuit. The transceiver module 2101 can also be referred to as a communication module or a communication interface. The storage module can be implemented by at least one memory.

[0475] The application further provides another communication apparatus. FIG. 22 is another structural schematic diagram of a communication apparatus according to an embodiment of the application. Referring to FIG. 22, the communication apparatus 2200 includes a processor 2201.

[0476] Optionally, the communication apparatus 2200 further includes a memory 2202.

[0477] Optionally, the communication apparatus 2200 further includes a transceiver 2203.

[0478] In a possible implementation, the processor 2201, the memory 2202 and the transceiver 2203 are connected through a bus respectively, and the memory 2202 stores computer instructions.

[0479] In a possible implementation, when the communication apparatus 2200 includes an access network device, or a CU or a DU included in the access network device, or a component (for example, a chip), a module or a unit in the access network device, the communication apparatus 2200 can be used to execute steps performed by the first communication apparatus and / or the second communication apparatus in the above method embodiments, and reference can be made to related descriptions in the above method embodiments.

[0480] In another possible implementation, when the communication apparatus 2200 includes a core network device, or a component (for example, a chip), a module or a unit in the core network device, the communication apparatus 2200 can be used to execute steps performed by the core network device in the above method embodiments, and reference can be made to related descriptions in the above method embodiments.

[0481] Optionally, the processing module 2102 in the above embodiment shown in FIG. 21 can be the processor 2201, and the transceiver module 2101 in the above embodiment shown in FIG. 21 can be the transceiver 2203.

[0482] The application further provides a communication apparatus. FIG. 23 is another structural schematic diagram of a communication apparatus according to an embodiment of the application. Referring to FIG. 23, the communication apparatus 2300 can be a terminal device in the above method embodiments, or a component (for example, a chip), a module or a unit of the terminal device in the above method embodiments. The communication apparatus 2300 can be used to execute steps performed by the first communication apparatus and / or the second communication apparatus in the above method embodiments, and reference can be made to related descriptions in the above method embodiments.

[0483] The processor is mainly used for processing data or signals, controlling the communication device, executing corresponding software programs, processing data of the software programs, and the like.

[0484] It should be noted that the signal processing algorithm of the processor has weak capability and cannot perform complex signal processing algorithms.

[0485] The memory is mainly used for storing software programs and data. The radio frequency circuit is mainly used for conversion between a baseband signal and a radio frequency signal and processing of the radio frequency signal.

[0486] The antenna is mainly used for receiving and transmitting radio frequency signals in the form of electromagnetic waves.

[0487] Optionally, the communication device 2300 further includes an input and output device, for example, a touch screen, a display screen, a keyboard, and the like, which are mainly used for receiving data input by a user and outputting data to the user.

[0488] When data needs to be transmitted, the processor performs baseband processing on the data to be transmitted, and outputs a baseband signal to the radio frequency circuit. The radio frequency circuit performs radio frequency processing on the baseband signal, and transmits a radio frequency signal in the form of electromagnetic waves through the antenna. When data is transmitted to the communication device, the radio frequency circuit receives a radio frequency signal through the antenna, converts the radio frequency signal into a baseband signal, and outputs the baseband signal to the processor. The processor converts the baseband signal into data and processes the data.

[0489] For ease of illustration, only one memory and one processor are shown in FIG. 23. In actual products of the communication device, one or more processors and one or more memories can exist. The memory can also be referred to as a storage medium or a storage device, and the like. The memory can be independent of the processor or integrated with the processor, and the embodiments of the present application do not limit this.

[0490] In the embodiments of the present application, the antenna and the radio frequency circuit having the functions of receiving and transmitting can be regarded as a transceiving unit of the communication device, and the processor having the processing function can be regarded as a processing unit of the communication device. As shown in FIG. 23, the communication device 2300 includes a transceiving unit 2310 and a processing unit 2320. The transceiving unit can also be referred to as a transceiver, a transceiver machine, a transceiving device, and the like. The processing unit can also be referred to as a processor, a processing board, a processing module, a processing device, and the like.

[0491] Optionally, the devices for implementing the receiving function in the transceiving unit 2310 can be regarded as a receiving unit, and the devices for implementing the transmitting function in the transceiving unit 2310 can be regarded as a transmitting unit, that is, the transceiving unit 2310 includes the receiving unit and the transmitting unit. The transceiving unit can also be referred to as a transceiver, a transceiver, or a transceiving circuit, and the like. The receiving unit can also be referred to as a receiver, a receiver, or a receiving circuit, and the like. The transmitting unit can also be referred to as a transmitter, a transmitter, or a transmitting circuit, and the like.

[0492] It should be understood that the transceiver unit 2310 is configured to perform the transmitting operation and the receiving operation of the first communication device and / or the second communication device in the above method embodiments, and the processing unit 2320 is configured to perform other operations of the first communication device and / or the second communication device in the above method embodiments, in addition to the transceiving operation.

[0493] When the communication device is a chip, the chip includes a transceiver unit and a processing unit. The transceiver unit can be an input / output circuit or a communication interface, and the processing unit is a processor or a microprocessor or an integrated circuit or a logic circuit integrated on the chip. In the above method embodiments, the transmitting operation corresponds to the output of the input / output circuit, and the receiving operation corresponds to the input of the input / output circuit.

[0494] The present application also provides a communication system including a first communication device and a second communication device, the first communication device is configured to perform all or part of the steps performed by the first communication device in the embodiments shown in FIGS. 5-20, and the second communication device is configured to perform all or part of the steps performed by the second communication device in the embodiments shown in FIGS. 5-20.

[0495] The embodiments of the present application also provide a computer program product including computer instructions, which, when executed on a computer, cause the computer to perform the method of the embodiments shown in FIGS. 5-20.

[0496] The embodiments of the present application also provide a computer-readable storage medium including computer instructions, which, when executed on a computer, cause the computer to perform the method of the embodiments shown in FIGS. 5-20.

[0497] The embodiments of the present application also provide a chip device including a processor, which is configured to invoke computer programs or computer instructions stored in a memory to cause the processor to perform the method of the embodiments shown in FIGS. 5-20.

[0498] Optionally, the processor is coupled to the memory through an interface.

[0499] Optionally, the chip device further includes a memory, and the memory stores computer programs or computer instructions.

[0500] The processor mentioned in any of the above can be a general central processing unit, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the programs of the methods of the embodiments shown in FIGS. 5-20. The memory mentioned in any of the above can be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM), and the like.

[0501] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. For example, the division of the units is only a logical function division. There can be another division manner for actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0502] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. In actual implementation, some or all of the units can be selected according to the actual needs to achieve the purposes of the embodiments.

[0503] In addition, each functional unit in the embodiments of the present application can be integrated in a processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware, or in the form of software functional units.

[0504] When the integrated unit is implemented in the form of software functional units and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such an understanding, the technical solutions of the present application essentially or the entire or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application.

[0505] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application; although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A communication method characterized by comprising: The method is applied to a first communication device, and the method comprises: receiving first data, the first data being carried in a physical layer; receiving second data, the second data being carried in a physical layer, the first data and the second data having a correlation relationship, the first data being used to determine a true value of model training or model monitoring of a first model, the second data being used to determine model input data of the first model, the first model being used to process physical layer data of the first communication device.

2. The method of claim 1, wherein, The method further comprises: performing model training or model monitoring on the first model according to the first data and the second data.

3. The method according to claim 1 or 2, characterized in that, The receiving of the first data comprises: receiving transmission data from a second communication device, the transmission data being determined based on user plane data and / or control plane data, the first data being determined based on the transmission data.

4. The method according to any one of claims 1-3, characterized in that, The method further comprises: receiving first information, the first information being used to indicate that the first data and the second data have a correlation relationship.

5. The method of claim 4, wherein, The first information comprises one or more of the following information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or, a data amount occupied by the first data in the transmission data.

6. The method according to any one of claims 1-5, characterized in that, The method further comprises: receiving second information, the second information being used to indicate one or more of the following resources: a frequency domain resource carrying the second data, a time domain resource carrying the second data, or a space domain resource carrying the second resource.

7. The method according to any one of claims 3-6, characterized in that, The method further comprises: receiving third information, the third information being used to indicate a relative position of the first data in the transmission data.

8. The method of claim 7, wherein, The third information comprises one or more of the following information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or, a data amount occupied by the first data in the transmission data.

9. The method of any one of claims 1-8, wherein: the first data further comprises a first reference signal, the first reference signal being used for channel estimation of a radio resource carrying the first data; the second data further comprises a second reference signal, the second reference signal being used for channel estimation of a radio resource carrying the second data by the first model.

10. The method of claim 9, wherein, The first information is further used to indicate that the first reference signal and the second reference signal have a correlation relationship.

11. The method according to any one of claims 1-10, characterized in that, The method further comprises: receiving fourth information, the fourth information being used to indicate that the first communication device feeds back gradient information related to the first model; sending the gradient information according to the fourth information.

12. The method according to any one of claims 1-11, characterized in that, The method further comprises: receiving fifth information, the fifth information being used to configure an effective condition of the first data and the second data, the effective condition comprising one or more of the following: the transmission data successfully performs channel decoding; the transmission data is successfully demodulated; Or, the first data performs channel estimation under a condition, wherein the condition of the channel estimation comprises one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), average delay, Doppler spread, Doppler shift, or delay spread.

13. The method according to any one of claims 1-12, characterized in that, The method further comprises: performing first processing on the first data to generate third data; performing second processing on the second data using the first model to generate fourth data, the third data serving as ground truth for the fourth data; determining a loss value of the first model based on the third data and the fourth data.

14. The method of any one of claims 1-13, wherein, The method further comprises: sending capability information of the first communication device, the capability information of the first communication device indicating one or more of the following: whether the first communication device supports model training or model monitoring based on the first data and the second data; frequency domain resources reserved by the first communication device for the second data; time domain resource requirements of the first communication device for the first data and the second data; accuracy of channel estimation by the first communication device based on the first reference signal; conditions of channel estimation by the first communication device based on the first reference signal, the conditions comprising one or more of the following: signal-to-noise ratio (SNR), delay spread requirement, average delay requirement, Doppler shift requirement, or Doppler spread requirement; data format of input data and / or output data supported by the first communication device; whether the first communication device supports feedback of gradient information related to the first model; time length for the first communication device to update gradient information related to the first model; or time length for the first communication device to feedback gradient information related to the first model.

15. A method of communication, comprising: The method is applied to a second communication device, and the method comprises: sending first data to a first communication device, the first data being carried in a physical layer; sending second data to the first communication device, the second data being carried in a physical layer, the first data and the second data having an association relationship, the first data being used to determine ground truth for model training or model monitoring of a first model, the second data being used to determine model input data for the first model, the first model being a model of the first communication device.

16. The method of claim 15, wherein, Sending the first data to the first communication device comprises: sending transmission data to the first communication device, the transmission data being determined based on user plane data and / or control plane data, the first data being determined based on the transmission data.

17. The method according to claim 15 or 16, characterized in that, The method further comprises: sending first information to the first communication device, the first information being used to indicate that the first data and the second data have an association relationship.

18. The method of claim 17, wherein, The first information comprises one or more of the following: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, identification information of a TB carrying the second data, a starting position of the first data in the transmission data, or a data amount occupied by the first data in the transmission data.

19. The method according to any one of claims 15-18, characterized by, The method further comprises: sending second information to the first communication device, the second information being used for indicating one or more of the following resources: a frequency domain resource carrying the second data, a time domain resource carrying the second data, or a spatial domain resource carrying the second data.

20. The method of any one of claims 16-19, wherein, The method further comprises: sending third information to the first communication device, the third information being used for indicating a relative position of the first data in the transmission data.

21. The method of claim 20, wherein, The third information comprises one or more of the following information: identification information of a transport block (TB) carrying the transmission data, identification information of a TB carrying the first data, a starting position of the first data in the transmission data, and / or, a data amount occupied by the first data in the transmission data.

22. The method of any of claims 15-21, wherein the first data further comprises a first reference signal, the first reference signal being used for channel estimation of a wireless resource carrying the first data; the second data further comprises a second reference signal, the second reference signal being used for channel estimation of a wireless resource carrying the second data by the first model.

23. The method of claim 22, wherein, The first information is further used for indicating that the first reference signal and the second reference signal have a correlation relationship.

24. The method of any one of claims 15-23, wherein, The method further comprises: sending fourth information to the first communication device, the fourth information being used for indicating that the first communication device feeds back gradient information related to the first model; receiving the gradient information from the first communication device.

25. The method of any one of claims 15-24, wherein, The method further comprises: sending fifth information to the first communication device, the fifth information being used for configuring an effective condition of the first data and the second data, the effective condition comprising one or more of the following: the transmission data successfully performs channel decoding; the transmission data successfully performs demodulation; or, a condition of performing channel estimation on the first data, wherein the condition of performing channel estimation comprises one or more of the following: signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), average delay, Doppler spread, Doppler shift, or delay spread.

26. The method of any one of claims 15-25, wherein, The method further comprises: sending capability information of the first communication device to the first communication device, the capability information of the first communication device indicating one or more of the following information: whether the first communication device supports model training or model monitoring based on the first data and the second data; whether the first communication device supports frequency domain resources reserved for the second data; a requirement of the first communication device on time domain resources of the first data and the second data; an accuracy of the first communication device in performing channel estimation based on the first reference signal; a condition of the first communication device in performing channel estimation based on the first reference signal, the condition comprising one or more of the following: signal-to-noise ratio (SNR), delay spread requirement, average delay requirement, Doppler shift requirement, or Doppler spread requirement; a data format of input data and / or output data supported by the first communication device; whether the first communication device supports feeding back gradient information related to the first model; The first communication device updates the time length of gradient information related to the first model; Or, the first communication device feeds back the time length of gradient information related to the first model.

27. The method of any one of claims 15-26, wherein, The method further comprises: performing third processing on the sixth data to generate the first data; determining fifth data according to the sixth data, the fifth data belonging to the sixth data; performing fourth processing on the fifth data to generate the second data.

28. A communications device, characterized by A module for performing the method of any one of claims 1 to 27.

29. A communications device, characterized by At least one processor coupled with a memory; the at least one processor is configured to perform the method of any one of claims 1 to 27.

30. The communication apparatus according to claim 29, wherein, The communication device is a chip or a chip system.

31. A readable storage medium, characterized by, The storage medium stores a computer program or instructions, when the computer program or instructions are executed by the communication device, the method of any one of claims 1 to 27 is realized.

32. A computer program product, characterised in that, When the computer program product runs on the computer, the computer is caused to perform the method of any one of claims 1 to 27.

Images