Data processing method and apparatus

Abstract

A data processing method and apparatus. In the method, a first apparatus can associate data streams with identifiers, thereby facilitating a data receiving end or a data sending end to distinguish different data streams by means of identifiers. Moreover, when a training unit of the first apparatus performs AI training, different data streams can be distinguished by means of different identifiers, and thus during the training of the data streams, parameters can be shared, such as using a same neural network processing method.

Description

Data processing method and device Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a data processing method and device. Background Art

[0002] Currently, artificial intelligence (AI) technology can be applied to the physical layer of mobile communication networks (e.g., for channel coding, channel prediction, receivers, etc.). For example, for AI research at the physical layer, convolutional neural networks (CNNs) can be used to design receivers in multiple-input, multiple-output (MIMO) scenarios. For example, when performing MIMO pilotless transmission, to reduce the complexity of the neural network and improve its generalization / scalability, the same parameters can be reused across different layers of the neural network. However, this makes it impossible for the receiver to distinguish between data from different streams, limiting the training / inference performance of the neural network.

[0003] Summary of the Invention

[0004] The present application provides a data processing method and apparatus, which can distinguish and reuse multiple layers of data streams processed by the same neural network.

[0005] In the first aspect, the present application provides a data processing method, which is performed by a first device. The first device is a data sending end (such as a transmitter). For example, for uplink transmission, the first device can be a terminal, or a component of a terminal (such as a processor, a chip, or a chip system, etc.), or a logic module that can implement all or part of the terminal functions. For another example, for downlink transmission, the first device can be a network device, or a component of a network device (such as a processor, a chip, or a chip system, etc.), or a logic module that can implement all or part of the terminal functions. The first device determines one or more data streams, each of which corresponds to one or more identifiers. The first device inputs one or more data streams and their corresponding identifiers into a training unit to obtain output information; the training unit includes one or more artificial intelligence AI modules in the transmitter for signal processing.

[0006] In this method, the first device is the data transmitter and can associate data streams with identifiers, thereby facilitating the data receiver (e.g., a receiver) to distinguish different data streams through the identifiers. Furthermore, when performing AI training, the training unit of the first device can distinguish different data streams through different identifiers. This allows the training of these data streams to share parameters (e.g., reuse the same neural network processing method), which helps improve the generalization and scalability of the network.

[0007] In one possible implementation, a data stream includes multiple symbols, each of which corresponds to an identifier. For example, assuming a data stream includes M symbols, each of the M symbols corresponds to M identifiers (e.g., there is a one-to-one correspondence between the M symbols and the M identifiers). These M identifiers constitute an identifier sequence, and each data stream can be said to correspond to an identifier sequence. The M identifiers are M different identifiers, and M is a positive integer.

[0008] In one possible implementation, a data stream includes multiple symbols, and some or all of the multiple symbols correspond to an identifier. For example, assuming a data stream includes M symbols, some (assuming N symbols, where N is a positive integer less than M) or all (i.e., M symbols) of the M symbols correspond to an identifier. M is a positive integer. Optionally, other symbols of the M symbols other than the N symbols may correspond to another identifier, which is different from the identifier corresponding to the N symbols.

[0009] In a possible implementation, different data streams correspond to different identifiers.

[0010] In the above method, each symbol in a data stream corresponds to an identifier, or some or all symbols in a data stream may correspond to an identifier, and different data streams correspond to different identifiers, which is helpful for distinguishing data streams.

[0011] In one possible implementation, the first device concatenates each data stream with the corresponding identifier to obtain a concatenated vector; the dimension of the concatenated vector is determined based on the number of data streams and the number of symbols in each data stream. The first device inputs the concatenated vector into the training unit.

[0012] In this method, the data streams and identifiers can be associated by directly concatenating them to obtain a concatenated vector. The dimension of the concatenated vector is related to the number of data streams and the number of symbols in each data stream. The training unit can share parameters when training the data streams (with the concatenated vector as input), which is beneficial for improving the generalization and scalability of the network.

[0013] In one possible implementation, the first device concatenates each data stream with its corresponding identifier to obtain a concatenated vector; the dimension of the concatenated vector is determined based on the number of data streams and the number of symbols in each data stream. The first device then performs dimensionality-enhancing processing on the concatenated vector to obtain a high-dimensional vector; the high-dimensional vector is input into the training unit.

[0014] In this method, the method of associating data streams with identifiers can specifically be to directly perform cascade processing to obtain a cascade vector, and then perform dimensionality upgrading on the cascade vector to obtain a high-dimensional vector, which is beneficial for the training unit to perform feature extraction and / or signal processing in a high-dimensional space.

[0015] In one possible implementation, the first device performs dimensionality-increasing processing on the data stream to obtain a high-dimensional vector; the dimensionality of the high-dimensional vector is determined based on the number of data streams and the number of symbols in each data stream. The first device performs a first operation on the high-dimensional vector and one or more identifiers, and inputs the result of the first operation into the training unit. For example, the result of the first operation includes a data stream with an identifier, which is also a high-dimensional vector.

[0016] In this method, the data stream and the identifier are specifically associated by first performing dimensionality-upgrading on the data stream to obtain a high-dimensional vector, and then performing a first operation (such as addition, subtraction, multiplication, division, or concatenation) on the high-dimensional vector and the identifier to obtain the result of the first operation (also a high-dimensional vector). The training unit can share parameters when training the data stream (with the input being a high-dimensional vector), which is beneficial for improving the generalization and scalability of the network.

[0017] In a possible implementation, the one or more identifiers are determined based on a pseudo-random sequence, where the pseudo-random sequence is related to one or more of a cell identifier, a user identifier, and an index of a data stream.

[0018] In this method, the generation of the identifier can reuse the pseudo-random sequence generation method in the new wireless NR protocol, and the parameters in the pseudo-random sequence generation method are improved so that the generated identifier can accurately identify different data streams, which is conducive to distinguishing different data streams.

[0019] In a possible implementation manner, the index of the data stream is indicated by downlink control information DCI.

[0020] In this method, for uplink transmission, if the first device is a data transmitter (i.e., a terminal), the first device can receive the index of the data stream indicated by the DCI. For another example, for downlink transmission, if the first device is a data transmitter (i.e., a base station), the first device already knows the index of the data stream and does not need to indicate it.

[0021] In one possible implementation, one or more identifiers are determined based on a complex number sequence or a real number sequence, which is determined based on a random sequence and a mapping relationship. The mapping relationship includes a first mapping relationship between multiple complex number identifiers and multiple index values, and / or a second mapping relationship between multiple real number identifiers and multiple index values. The random sequence is determined based on a user identifier and / or an index of a data stream, and the values ​​in the random sequence are index values ​​in the first mapping relationship and / or the second mapping relationship.

[0022] In this method, the identifier can be generated by using the value of the random sequence as the index value in the mapping relationship, and the corresponding identifier can be determined by looking up the index value in the mapping relationship.

[0023] In one possible implementation, the first device determines a random sequence based on a user identifier and / or an index of a data stream, wherein the values ​​in the random sequence are index values ​​in a mapping relationship. The mapping relationship includes a first mapping relationship between a plurality of complex identifiers and a plurality of index values, and / or a second mapping relationship between a plurality of real identifiers and a plurality of index values. Based on the values ​​in the random sequence, the first device searches the mapping relationship for the complex identifiers or real identifiers in the mapping relationship corresponding to the values ​​in the random sequence to obtain a complex sequence or a real sequence. The first device determines one or more identifiers based on the complex sequence or the real sequence, for example, determining that the complex identifiers in the complex sequence are one or more identifiers.

[0024] In a possible implementation, the maximum value in the random sequence is the maximum index value in the first mapping relationship or the second mapping relationship.

[0025] In a possible implementation manner, the first mapping relationship or the second mapping relationship is predefined by the first device, or is obtained through radio resource control RRC configuration.

[0026] In one possible implementation, one or more identifiers are determined based on a first functional relationship, and parameters of the first functional relationship are related to one or more of the number of data streams, the maximum number of streams supported, the maximum dimension of the identifier, the dimension index of the identifier, and the user identifier.

[0027] In this method, the generation of the identifier can be based on the first functional relationship, and the parameters of the first function are designed to be related to the data stream, which is conducive to distinguishing different data streams.

[0028] In a possible implementation, the first functional relationship satisfies the following formula:

[0029] Among them, r(l,n RNTI, i) includes one or more identifiers, l represents the number of data streams, d represents the maximum dimension of the identifier, i represents the dimension index of the identifier, M represents the maximum number of streams supported, n RNTI Represents the user ID, cos represents the cosine function, and sin represents the sine function.

[0030] In a possible implementation, one or more identifiers belong to an identifier sequence of the data transmitter, or one or more identifiers belong to an identifier sequence of the data receiver; and there is a corresponding relationship between the identifier sequence of the data transmitter and the identifier sequence of the data receiver.

[0031] In one possible implementation, the data sending end sends output information, which is input into the data receiving end after being transmitted through the channel; or, the data receiving end sends output information, which is used by the data receiving end and the data sending end to update their respective training units.

[0032] In a second aspect, the present application provides a data processing method, which is performed by a second device. The second device is a data receiving end (such as a receiver). For example, for uplink transmission, the second device can be a network device, or a component of a network device (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the functions of the network device. For another example, for downlink transmission, the second device can be a terminal, or a component of a terminal (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the functions of the terminal. The second device determines one or more data streams, each of which corresponds to one or more identifiers. The second device inputs one or more data streams and their corresponding identifiers into a training unit to obtain output information; the training unit includes one or more artificial intelligence AI modules in the receiver for signal processing.

[0033] In this method, the second device is a data receiver that can associate received data streams with identifiers, allowing the data receiver to distinguish different data streams through the identifiers. Furthermore, when performing AI training, the training unit of the second device can distinguish different data streams through different identifiers. This allows the training of data streams to share parameters (e.g., reuse the same neural network processing method), which helps improve the generalization and scalability of the network.

[0034] Optionally, other possible implementations of the data processing method performed by the second device can refer to the possible implementations in the first aspect above, which will not be repeated here.

[0035] In a third aspect, the present application provides a data processing method, which is performed by a first device or a second device. The first device is a data transmitter, and the second device is a data receiver. For example, for uplink transmission, the first device can be a terminal, or a component of a terminal (e.g., a processor, chip, or chip system), or a logic module that can implement all or part of the terminal's functions; the second device can be a network device, or a component of a network device (e.g., a processor, chip, or chip system), or a logic module that can implement all or part of the network device's functions. For another example, for downlink transmission, the first device can be a network device, or a component of a network device (e.g., a processor, chip, or chip system), or a logic module that can implement all or part of the terminal's functions; the second device can be a terminal, or a component of a terminal (e.g., a processor, chip, or chip system), or a logic module that can implement all or part of the terminal's functions. Taking the first device as an example, the first device determines one or more data streams. The first device determines one or more identifiers corresponding to the data streams based on a pseudo-random sequence, where the pseudo-random sequence is related to one or more of a cell identifier, a user identifier, and a data stream index.

[0036] In this method, the first device can determine one or more data streams and improve the parameters in the pseudo-random sequence generation method based on the pseudo-random sequence generation method so that the generated identifier can accurately identify different data streams, which is conducive to distinguishing different data streams.

[0037] In a possible implementation, the first device determines a pseudo-random sequence based on one or more binary sequences.

[0038] In this method, the first device can reuse the method of determining a pseudo-random sequence based on a binary sequence (such as a real binary sequence and / or a complex binary sequence) described in the NR protocol, which is conducive to implementation.

[0039] In a fourth aspect, the present application provides a data processing method, which is performed by a first device or a second device. The first device is a data transmitter and the second device is a data receiver. For example, for uplink transmission, the first device can be a terminal, or a component of a terminal (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions; the second device can be a network device, or a component of a network device (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the network device functions. For another example, for downlink transmission, the first device can be a network device, or a component of a network device (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions; the second device can be a terminal, or a component of a terminal (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions. Taking the first device as an example, the first device determines a random sequence based on a user identifier and / or a data stream index. The values ​​in the random sequence are index values ​​in the mapping relationship. The mapping relationship includes a first mapping relationship between multiple complex identifiers and multiple index values, and / or a second mapping relationship between multiple real identifiers and multiple index values. The first device searches the mapping relationship for the complex identifiers or real identifiers corresponding to the values ​​in the random sequence based on the values ​​in the random sequence, and obtains the complex number sequence or real number sequence. The first device determines one or more identifiers based on the complex number sequence or real number sequence.

[0040] In this method, the identifier can be generated by using the value of the random sequence as the index value in the mapping relationship, and the corresponding identifier can be determined by looking up the index value in the mapping relationship.

[0041] In a possible implementation, the maximum value in the random sequence is the maximum index value in the first mapping relationship or the second mapping relationship.

[0042] In a possible implementation manner, the first mapping relationship or the second mapping relationship is predefined by the first device, or is obtained through radio resource control RRC configuration.

[0043] In a fifth aspect, the present application provides a data processing method, which is performed by a first device or a second device. The first device is a data transmitter and the second device is a data receiver. For example, for uplink transmission, the first device can be a terminal, or a component of a terminal (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions; the second device can be a network device, or a component of a network device (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the network device functions. For another example, for downlink transmission, the first device can be a network device, or a component of a network device (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions; the second device can be a terminal, or a component of a terminal (such as a processor, chip, or chip system, etc.), or a logic module that can implement all or part of the terminal functions. Taking the first device as an example, the first device determines one or more data streams. The first device determines one or more identifiers corresponding to the data streams based on the first functional relationship.

[0044] In this method, the generation of the identifier may be based on a first functional relationship, and the first device may design parameters of the first function to be related to the data stream, which is helpful for distinguishing different data streams.

[0045] In a possible implementation, the first functional relationship satisfies the following formula:

[0046] Among them, r(l,n RNTI , i) includes one or more identifiers, l represents the number of data streams, d represents the maximum dimension of the identifier, i represents the dimension index of the identifier, M represents the maximum number of streams supported, n RNTI Represents the user ID, cos represents the cosine function, and sin represents the sine function.

[0047] In a sixth aspect, the present application provides a communication device. The communication device may be a terminal, a device for a terminal, or a device capable of being used in conjunction with a terminal. In one possible implementation, the communication device may include a functional module, which may be implemented as a hardware circuit, software, or a combination of hardware circuits and software.

[0048] In one possible embodiment, the communication device includes a processing unit. The processing unit is configured to determine one or more data streams, each corresponding to one or more identifiers. The processing unit is further configured to input the one or more data streams and their corresponding identifiers into a training unit to obtain output information; the training unit includes one or more artificial intelligence (AI) modules in the transmitter that perform signal processing.

[0049] In one possible implementation, a data stream includes multiple symbols, each of which corresponds to an identifier. For example, assuming a data stream includes M symbols, each of the M symbols corresponds to M identifiers (e.g., there is a one-to-one correspondence between the M symbols and the M identifiers). These M identifiers constitute an identifier sequence, and each data stream can be said to correspond to an identifier sequence. The M identifiers are M different identifiers, and M is a positive integer.

[0050] In one possible implementation, a data stream includes multiple symbols, and some or all of the multiple symbols correspond to an identifier. For example, assuming a data stream includes M symbols, some (assuming N symbols, where N is a positive integer less than M) or all (i.e., M symbols) of the M symbols correspond to an identifier. M is a positive integer. Optionally, other symbols of the M symbols other than the N symbols may correspond to another identifier, which is different from the identifier corresponding to the N symbols.

[0051] In a possible implementation, different data streams correspond to different identifiers.

[0052] In one possible implementation, the processing unit is configured to concatenate each data stream with its corresponding identifier to obtain a concatenated vector; the dimension of the concatenated vector is determined based on the number of data streams and the number of symbols in each data stream. The processing unit is further configured to input the concatenated vector into the training unit.

[0053] In one possible implementation, the processing unit is configured to concatenate each data stream with its corresponding identifier to obtain a concatenated vector; wherein the dimension of the concatenated vector is determined based on the number of data streams and the number of symbols in each data stream. The processing unit is further configured to perform dimensionality-increasing processing on the concatenated vector to obtain a high-dimensional vector; and input the high-dimensional vector into the training unit.

[0054] In one possible implementation, the processing unit is configured to perform dimensionality-increasing processing on the data stream to obtain a high-dimensional vector; the dimensionality of the high-dimensional vector is determined based on the number of data streams and the number of symbols in each data stream. The processing unit is further configured to perform a first operation on the high-dimensional vector and one or more identifiers, and input the result of the first operation into the training unit. For example, the result of the first operation includes a data stream with an identifier, which is also a high-dimensional vector.

[0055] In a possible implementation, the one or more identifiers are determined based on a pseudo-random sequence, where the pseudo-random sequence is related to one or more of a cell identifier, a user identifier, and an index of a data stream.

[0056] In a possible implementation manner, the index of the data stream is indicated by downlink control information DCI.

[0057] In one possible implementation, one or more identifiers are determined based on a complex number sequence or a real number sequence, which is determined based on a random sequence and a mapping relationship. The mapping relationship includes a first mapping relationship between multiple complex number identifiers and multiple index values, and / or a second mapping relationship between multiple real number identifiers and multiple index values. The random sequence is determined based on a user identifier and / or an index of a data stream, and the values ​​in the random sequence are index values ​​in the first mapping relationship and / or the second mapping relationship.

[0058] In one possible implementation, the processing unit is configured to determine a random sequence based on a user identifier and / or an index of a data stream, wherein the values ​​in the random sequence are index values ​​in a mapping relationship. The mapping relationship includes a first mapping relationship between multiple complex identifiers and multiple index values, and / or a second mapping relationship between multiple real identifiers and multiple index values. The processing unit is configured to search the mapping relationship for the complex identifiers or real identifiers in the mapping relationship corresponding to the values ​​in the random sequence, based on the values ​​in the random sequence, to obtain a complex sequence or a real sequence. The processing unit determines one or more identifiers based on the complex sequence or the real sequence.

[0059] In a possible implementation, the maximum value in the random sequence is the maximum index value in the first mapping relationship or the second mapping relationship.

[0060] In a possible implementation manner, the first mapping relationship or the second mapping relationship is predefined by the first device, or is obtained through radio resource control RRC configuration.

[0061] In one possible implementation, one or more identifiers are determined based on a first functional relationship, and parameters of the first functional relationship are related to one or more of the number of data streams, the maximum number of streams supported, the maximum dimension of the identifier, the dimension index of the identifier, and the user identifier.

[0062] In a possible implementation, the first functional relationship satisfies the following formula:

[0063] Among them, r(l,n RNTI , i) includes one or more identifiers, l represents the number of data streams, n layer Indicates the index value of the data stream, d indicates the maximum dimension of the identifier, i indicates the dimension index of the identifier, M indicates the maximum number of streams supported, and n RNTI Represents the user ID, cos represents the cosine function, and sin represents the sine function.

[0064] In a possible implementation, one or more identifiers belong to an identifier sequence of the data transmitter, or one or more identifiers belong to an identifier sequence of the data receiver; and there is a corresponding relationship between the identifier sequence of the data transmitter and the identifier sequence of the data receiver.

[0065] In one possible embodiment, the communication device further includes a communication unit. The communication unit is configured to transmit output information from the data transmitting end, which is input to the data receiving end after being transmitted through a channel; or the communication unit is configured to transmit output information from the data receiving end, which is used by the data receiving end and the data transmitting end to update their respective training units.

[0066] In a seventh aspect, the present application provides a communication device. The communication device may be a network device, a device of a network device, or a device capable of being used in conjunction with a network device. In one possible implementation, the communication device may include a functional module, which may be implemented as a hardware circuit, software, or a combination of hardware circuit and software.

[0067] Optionally, other possible implementations of the data processing method performed by the communication device in the seventh aspect can refer to the possible implementations in the above-mentioned first aspect or sixth aspect, and will not be repeated here.

[0068] In an eighth aspect, the present application provides a communication device. The communication device may be a terminal or a network device, or a device of a terminal or a network device, or a device that can be used in conjunction with a terminal or a network device. In one possible implementation, the communication device may include a functional module, which may be a hardware circuit, software, or a combination of hardware circuit and software.

[0069] Optionally, other possible implementations of the data processing method performed by the communication device in the eighth aspect can refer to the above-mentioned third to fifth aspects, as well as the possible implementations in the third to fifth aspects, and will not be repeated here.

[0070] For aspects 6 to 8, as an example, the processing unit may be a processor, and the communication unit may be a transceiver unit, a transceiver, or a communication interface. It is understood that when the communication device is a communication device (such as a terminal or a network device), the communication unit may be a transceiver in the communication device (for example, a transceiver includes a transmitter and a receiver), for example, implemented by an antenna, a feeder, and a codec in the communication device, or, if the communication device is a chip provided in the device, the processing unit may be a processing circuit, a logic circuit, etc. of the chip, and the communication unit may be an input / output interface of the chip, such as an input / output circuit, a pin, etc.

[0071] In a ninth aspect, the present application provides a communication device, comprising: a processor configured to execute instructions; optionally, the communication device further comprising a memory configured to store the instructions, wherein when the instructions are executed by the processor, the communication device implements the method of any possible implementation of the first to fifth aspects, and any of the first to fifth aspects. Optionally, the processor and the memory are coupled.

[0072] In the tenth aspect, the present application provides a communication system, which includes multiple devices or equipment in the above-mentioned sixth to eighth aspects, so that the devices or equipment execute the methods in the first to fifth aspects, and any possible implementation methods of the first to fifth aspects.

[0073] In the eleventh aspect, the present application provides a computer-readable storage medium storing instructions, which, when the instructions are executed on a computer, enables the computer to execute the method of any possible implementation of the first to fifth aspects, and the first to fifth aspects.

[0074] In a twelfth aspect, the present application provides a computer program product comprising instructions, which, when executed on a computer, enable the computer to execute the method of any possible implementation of aspects 1 to 5, and aspects 1 to 5.

[0075] In a thirteenth aspect, the present application provides a chip comprising a processor (or a logic circuit). Optionally, the chip may further comprise a communication interface (or interface) for implementing the method in any one of the first to fifth aspects and the first to fifth aspects above. In one possible implementation, if the chip is the smallest processing unit in the entire machine, the chip may be a processor, or may comprise a processor and a memory, or may comprise a processor, a memory and a transceiver, for implementing the method in any one of the first to fifth aspects and the first to fifth aspects above.

[0076] In a fourteenth aspect, the present application provides a chip system. The chip system includes a processor and an interface. Optionally, it may also include a memory for implementing the method of any possible implementation of the first to fifth aspects above, and any of the first to fifth aspects. The chip system may be composed of a chip, or may include a chip and other discrete devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] FIG1 is a schematic diagram of a communication system provided by the present application;

[0078] FIG2 is a flow chart of a data processing method provided by the present application;

[0079] FIG3 is a schematic diagram of a data stream provided by the present application and a corresponding identifier being cascaded and input into a training unit;

[0080] FIG4 is a schematic diagram of another data stream provided by the present application and the corresponding identifier being cascaded and input into a training unit;

[0081] FIG5 is a schematic diagram of a method provided by the present application for performing dimensionality upgrading on a data stream, performing a first operation on the upgraded data stream and the corresponding identifier in a high-dimensional space and inputting the result into a training unit;

[0082] FIG6 is a schematic diagram of a communication device provided by the present application;

[0083] FIG7 is a schematic diagram of another communication device provided in this application. DETAILED DESCRIPTION

[0084] In the embodiments of this application, " / " can indicate that the associated objects are in an "or" relationship. For example, A / B can mean A or B. "And / or" can be used to describe the existence of three relationships between associated objects. For example, "A and / or B" can mean: A exists alone, A and B exists simultaneously, or B exists alone. A and B can be singular or plural. To facilitate the description of the technical solutions of the embodiments of this application, the words "first" and "second" may be used in the embodiments of this application to distinguish between technical features with the same or similar functions. The words "first" and "second" do not limit the number or order of execution, and the words "first" and "second" do not necessarily mean different. In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary" or "for example" should not be construed as preferred or advantageous over other embodiments or designs. The use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for easier understanding.

[0085] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0086] In order to enable a transmitter or a receiver to distinguish multiple layers of data streams processed by the same neural network, the present application provides a data processing method that can distinguish multiple layers of data streams.

[0087] The data processing method provided in this application can be applied to the communication system shown in Figure 1. For example, the communication system includes a network device and a terminal.

[0088] Among them, the communication system of the present application may include but is not limited to communication systems of various radio access technologies (RAT), for example, it may be: a narrowband Internet of Things system (NB-IoT), a long term evolution (LTE) communication system, or a 5G (or new radio (NR)) communication system, or a transition system between an LTE communication system and a 5G communication system, which may also be called a 4.5G communication system, and of course it may also be a future communication system, such as the sixth generation (6G) or even the seventh generation (7G) system. The network architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. It is known to those skilled in the art that with the evolution of the communication network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0089] Among them, a terminal, also known as a terminal device (terminal), user equipment (UE), mobile station (MS), mobile terminal (MT), etc., refers to a device that provides voice and / or data connectivity to users. For example, a handheld device with wireless connection function, a vehicle-mounted device, etc. Currently, some examples of terminals include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices, drones, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control (industrial control), wireless terminals in self-driving (self-driving), wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety (transportation safety), wireless terminals in smart cities (smart cities), wireless terminals in smart homes (smart homes), terminals in 5G networks, terminals in future evolved networks, or terminals in future communication systems.

[0090] The network device of the present application refers to a radio access network (RAN) node (or device) that connects a terminal to a wireless network, and may also be referred to as a base station. For example, some examples of RAN nodes include: a gNB, a transmission reception point (TRP), 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 (e.g., a home evolved NodeB, or a home Node B, HNB), a base band unit (BBU), or a wireless fidelity (Wifi) access point (AP), a satellite in a satellite communication system, a radio controller in a cloud radio access network (CRAN) scenario, a wearable device, a drone, or a device in an Internet of Vehicles (e.g., vehicle to everything (V2X)), or a communication device in device to device (D2D) communication.

[0091] In one possible implementation, the network device may include a centralized unit (CU) node, a distributed unit (DU) node, or a RAN device including a CU node and a DU node. The RAN device including the CU node and the DU node splits the protocol layer of the eNB in ​​the long term evolution (LTE) system, places the functions of some protocol layers in the CU for centralized control, and distributes the functions of the remaining part or all of the protocol layers in the DU, which is centrally controlled by the CU. In some deployments of network devices, the CU can also be divided into a CU-control plane (CP) and a CU-user plane (UP). In another possible implementation, the network device can also be an antenna unit (RU), etc. In another possible implementation, the network device can also be an open radio access network (ORAN) architecture, etc. This application does not limit the specific type of the network device. For example, when the network device is an ORAN architecture, the network device shown in the embodiment of the present application can be an access network device in the ORAN, or a module in the access network device, etc. In the ORAN system, CU can also be called an open centralized unit (open CU, O-CU), DU can also be called an open distributed unit (O-DU), CU-DU can also be called an open centralized unit-distributed unit (O-CU-DU), CU-UP can also be called an open centralized unit-control plane (O-CU-UP), and RU can also be called an open antenna unit (O-RU).

[0092] It should be noted that:

[0093] "Sending" and "receiving" in the embodiments of the present application indicate the direction of signal transmission. For example, "sending information to a terminal" can be understood as the destination end of the information being the terminal device, which can include direct sending through the air interface, and also includes indirect sending through the air interface by other units or modules. "Receiving information from a network device" can be understood as the source end of the information being the network device, which can include direct receiving from the network device through the air interface, and also includes indirect receiving from the network device through the air interface from other units or modules. "Sending" can also be understood as the "output" of the chip interface, and "receiving" can also be understood as the "input" of the chip interface.

[0094] In other words, sending and receiving can be performed between devices, for example, between a network device and a terminal device, or can be performed within a device, for example, sending or receiving between components, modules, chips, software modules or hardware modules within the device through a bus, wiring or interface.

[0095] It is understandable that information may be processed between the source and destination of information transmission, such as coding, modulation, etc., but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated.

[0096] In the embodiments of the present application, "indication" may include direct indication and indirect indication, and may also include explicit indication and implicit indication. The information indicated by a certain information (such as the indication information described below) is called information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, the information to be indicated can be directly indicated, such as the information to be indicated itself or the index of the information to be indicated. The information to be indicated can also be indirectly indicated by indicating other information, wherein there is an association between the other information and the information to be indicated; it is also possible to indicate only a part of the information to be indicated, while the other parts of the information to be indicated are known or agreed in advance, for example, the indication of specific information can be achieved with the help of the arrangement order of each information agreed in advance (such as predefined by the protocol), thereby reducing the indication overhead to a certain extent. The present application does not limit the specific method of indication. It is understandable that, for the sender of the indication information, the indication information can be used to indicate the information to be indicated, and for the receiver of the indication information, the indication information can be used to determine the information to be indicated.

[0097] 1. To facilitate understanding, the definitions of relevant terms involved in this application are introduced in detail below.

[0098] 1. Training units of this application:

[0099] At present, artificial intelligence (AI) technology can be applied to the network layer (such as network optimization, mobility management, resource allocation, etc.) and the physical layer (such as channel coding, channel prediction, etc.). For example, the more commonly used AI technologies include reinforcement learning, supervised learning and unsupervised learning. The training unit in this application includes one or more artificial intelligence AI modules that perform signal processing in a transmitter or receiver. For example, the training unit is one or more AI modules that perform signal processing in the physical layer (including a coding module, a modulation module, a MIMO module, a beamforming module, a radio frequency (RF) module, etc.), and the above one or more modules can be jointly optimized by an AI algorithm, and the whole of the joint optimization is a training unit.

[0100] 2. Data flow of this application:

[0101] For example, one or more data streams of the present application are multi-layer data streams. Assuming that the transmitter or receiver in the present application is designed using a convolutional neural network (CNN), such as the transmitter or receiver includes a multi-layer CNN neural network, then the one or more data streams input by the transmitter or receiver are multi-layer data streams. For another example, when the transmitter or receiver performs MIMO pilot-free transmission (such as without a demodulation reference signal (DMRS)), the one or more data streams input by the transmitter or receiver are multi-layer MIMO data streams.

[0102] 3. Identifier of this application:

[0103] For example, one or more identifiers of the present application are plural identifiers or real number identifiers, such as an identifier can be expressed in plural form (such as expressed as ), or can be expressed as a real number (e.g., 1 or -1). A set of one or more identifiers can be called an identifier sequence. For example, a set of one or more plural identifiers can be called a plural identifier sequence, and a set of one or more real identifiers can be called a real identifier sequence.

[0104] 2. Data processing method provided by this application:

[0105] FIG2 is a flow chart of a data processing method provided by the present application. The data processing method is applied to the communication system shown in FIG1 . For example, the data processing method can be executed by a first device. The first device can be a terminal or a device capable of implementing terminal functions, or a network device or a device capable of implementing network device functions. The data processing method includes the following steps:

[0106] S101: A first device determines one or more data streams, where each data stream corresponds to one or more identifiers.

[0107] (1) One or more data streams are multi-layer data streams, and each data stream includes multiple symbols.

[0108] For example, when the first device is a transmitter, assume that one or more data streams are represented as X (e.g., X can be a multi-layer modulated real symbol stream or a multi-layer modulated complex symbol stream, which can be collectively referred to as a multi-layer modulated data stream), and X includes multiple real symbols or multiple complex symbols. X can be represented in the form of a matrix, and the dimension of the X matrix is ​​represented as l×n; where l represents the number of data streams, n represents the number of symbols in each data stream, and l×n represents the total number of symbols in the transmitted data stream. Optionally, for a multi-layer modulated data stream X, l can represent the number of layers of the multi-layer modulated data stream, n can represent the number of symbols in each layer, and l×n represents the total number of symbols in the multi-layer modulated data stream of the transmitter.

[0109] For another example, when the first device is a receiver, assuming that one or more data streams are represented by Y, Y can be a multi-antenna reception symbol stream (such as a multi-layer MIMO data stream), such as Y including multiple real symbols or multiple complex symbols. Y can be represented in the form of a matrix, and the dimension of the Y matrix is ​​represented as l×n×r; wherein l represents the number of data streams, n represents the number of symbols of each data stream, and r represents the number of receiving antennas, then l×n×r represents the total number of symbols of the received data stream. Optionally, for a multi-layer modulated data stream Y, l can represent the number of layers of the multi-layer modulated data stream, n can represent the number of symbols of each layer, and r represents the number of receiving antennas, then l×n×r represents the total number of symbols of the multi-layer modulated data stream of the receiver.

[0110] (2) Identifiers are used to distinguish different data streams, or to distinguish different symbols in the same data stream.

[0111] In a possible implementation, a data stream includes multiple symbols, and each symbol corresponds to an identifier.

[0112] For example, suppose a data stream X includes l×n symbols, and each symbol corresponds to an identifier. Then, l×n symbols correspond to l×n identifiers. In this case, the number of identifiers is the same as the number of symbols in the data stream. Alternatively, a set of one or more identifiers can be called an identifier sequence. For example, a set of l×n identifiers is called an identifier sequence Z.

[0113] In a possible implementation, a data stream includes multiple symbols, and some or all of the multiple symbols correspond to an identifier.

[0114] For example, if some of the multiple symbols of a data stream correspond to an identifier, assuming that a data stream includes l×n symbols, some of the l×n symbols (such as a×n symbols, a<l) correspond to an identifier, and the other symbols of the l×n symbols (such as (la)×n symbols) correspond to another identifier, then the data stream corresponds to two identifiers. Optionally, the other symbols of the l×n symbols (such as (la)×n symbols) can also correspond to multiple other identifiers, such as each symbol corresponds to an identifier, or some symbols correspond to an identifier, etc., which is not limited in this application. For another example, if all of the multiple symbols of a data stream correspond to an identifier, assuming that a data stream includes l×n symbols, all of the l×n symbols correspond to the same identifier. In the above example, the number of identifiers is less than the number of symbols of the data stream.

[0115] In a possible implementation, different data streams correspond to different identifiers.

[0116] For example, assume that different data streams include a first data stream and a second data stream. Each symbol in the first data stream corresponds to an identifier, then the first data stream corresponds to multiple identifiers (or is called the first data stream corresponding to the first identifier sequence); each symbol in the second data stream corresponds to an identifier, then the second data stream corresponds to multiple identifiers (or is called the second data stream corresponding to the second identifier sequence). The first data stream and the second data stream correspond to different identifiers, then the first identifier sequence and the second identifier sequence are different. Specifically, all identifiers in the first identifier sequence are different from all identifiers in the second identifier sequence; or, some identifiers in the first identifier sequence are different from some identifiers in the second identifier sequence, which is not limited in this application.

[0117] For another example, assuming that the different data streams include a third data stream and a fourth data stream, all symbols in the third data stream correspond to a first identifier, and all symbols in the fourth data stream correspond to a second identifier. The first identifier and the second identifier are different, so the third data stream and the fourth data stream can be distinguished.

[0118] For another example, assuming that the different data streams include a fifth data stream and a sixth data stream, some symbols in the fifth data stream correspond to a first identifier, and the remaining symbols correspond to a second identifier; some symbols in the sixth data stream correspond to a third identifier, and the remaining symbols correspond to a fourth identifier. Then, the first identifier is different from the third identifier, and the first identifier is also different from the fourth identifier; the second identifier is different from the third identifier, and the second identifier is also different from the fourth identifier.

[0119] Optionally, the above examples may also be arranged and combined according to the correspondence between symbols and identifiers in different data streams, which will not be described in detail here.

[0120] S102: The first device inputs one or more data streams and their corresponding identifiers into a training unit to obtain output information.

[0121] Since the data streams have corresponding identifiers, even if the training unit includes a multi-layer CNN neural network and the multi-layer CNN neural network reuses the same neural network processing, the training unit can distinguish different data streams according to different identifiers and train the data streams.

[0122] For example, when the first device is a transmitter, the training unit may include one or more AI modules for signal processing of the transmitter, such as modules for channel coding, modulation, precoding, waveform generation, and other AI-coordinated optimization. The first device may input one or more data streams and their corresponding identifiers into the training unit (e.g., performing channel coding and modulation on the data streams and their corresponding identifiers), and obtain output information, which includes the data stream with the identifier information.

[0123] For another example, when the first device is a receiver, the training unit may include one or more AI modules for signal processing of the receiver, such as waveform analysis, channel estimation, channel equalization, demodulation, channel decoding, and other AI-co-optimized modules. The first device may input one or more data streams and their corresponding identifiers into the training unit (e.g., performing channel decoding and demodulation on the data streams and their corresponding identifiers), and obtain output information, which includes decoded information (e.g., log-likelihood ratio (LLR)).

[0124] In this embodiment, the first device can associate data streams with identifiers, and different data streams can be distinguished based on different identifiers. Furthermore, when performing AI training, the training unit of the first device can distinguish different data streams based on different identifiers. This allows the training of data streams to share parameters (e.g., reuse the same neural network processing method), which is beneficial for improving the generalization and scalability of the network.

[0125] 3. The following describes in detail the method for determining the identifier corresponding to the data stream.

[0126] Among them, when the first device in this application is a data sending end or a data receiving end, the method of determining the identifier corresponding to the data stream is similar, and the methods in the following examples can be used.

[0127] Example 1: One or more identifiers corresponding to a data stream are determined based on a pseudo-random sequence.

[0128] The pseudo-random sequence is related to one or more of a cell identifier, a user identifier, and an index of a data stream.

[0129] In one possible implementation, the first device may generate a pseudo-random sequence c(n) according to the method for generating a pseudo-random sequence in the NR protocol as shown in formula (1): c(n) = (x1(n+N c )+x2(n+N c ))mod2 (1)

[0130] Among them, c(n) represents a pseudo-random sequence, x1(n+N c ) and x2(n+N c ) represents a binary sequence, N c The value of N is 1600 (i.e. c =1600), mod represents the modulo operation. The pseudo-random sequence c(n) is an n-length sequence, where n starts at 0. For example, c(0) is expressed as c(0) = (x1(0+1600) + x2(0+1600)) mod 2, i.e., the value of c(0) is determined based on x1(1600) and x2(1600). The values ​​of x1(1600) and x2(1600) can be obtained by iterating through recursive formulas (2) and (3). For example, formula (2) is expressed as: x1(n+31) = (x1(n+3) + x1(n)) mod 2 (2)

[0131] Where x1(0) = 1, x1(n) = 0, and n = 1, 2, ..., 30. For example, based on formula (2), x1(1600) can be expressed as: x1(1569+31) = (x1(1569+31) + x1(1569)) mod 2, and so on, until x1(1600) can be expressed by x1(0) to x1(30). Similarly, when n>30, formula (3) can be used to express the subsequent binary sequence x2: x2(n+31) = (x2(n+3) + x2(n+2) + x2(n+1) + x2(n)) mod 2 (3)

[0132] The value of x2 is consistent with the parameter c defined in this application. init Related, the parameter c init Related to one or more of the cell ID, user ID, and data stream index. For example, assuming parameter c init As shown in formula (4): c init =A·n RNTI +B·n layer +n ID (4)

[0133] Among them, n RNTIIndicates parameters related to user identification (such as terminal identification) (such as cell-radio network temporary identifier (C-RNTI) etc.), A is n RNTI The coefficient of A, for example, can be 2 15 , or other values, which are not limited in this application. layer Indicates the index value of the data stream (optional, for multi-layer data streams, n layer Indicates the index value of the layer of the multi-layer data stream), B is n layer The coefficient of, for example, the value of B can be 2 10 , or other values, which are not limited in this application. ID represents a parameter related to the cell identity (such as cell ID or a parameter related to the cell identity configured by the base station). Optionally, the above formula (4) is only an example, and the parameter c init The expression of can also be a deformation of formula (4), such as the change of coefficients A and B, or n RNTI The present application does not limit the changes of parameters such as .

[0134] Optional, parameter c init It can also be obtained through radio resource control (RRC) configuration or downlink control information (DCI) indication. For example, when the first device is a terminal, the base station can indicate the parameter c to the terminal through RRC configuration or DCI. init ; The first device receives the parameter c init And based on the parameter c init , generate a pseudo-random sequence c(n).

[0135] Optional, n layer It can be indicated by DCI. For example, when the first device is a terminal, the base station can indicate the parameter n through DCI in the process of sending the reference signal. layer ; The first device receives the parameter n layer And based on the parameter n layer , generate a pseudo-random sequence c(n).

[0136] In one possible implementation, after the first device determines the pseudo-random sequence c(n), the first device may generate one or more complex identifiers or one or more real identifiers based on the pseudo-random sequence. For example, the first device may generate a complex identifier sequence r(n) as shown in formula (5):

[0137] Wherein, r(n) includes one or more identifiers, c(2n) represents the even-numbered sequence of the pseudo-random sequence c(n), constituting the real part of the complex identifier sequence, and c(2n+1) represents the odd-numbered sequence of the pseudo-random sequence c(n), constituting the imaginary part of the complex identifier sequence.

[0138] For another example, the first device may generate a complex identifier sequence r(n) as shown in formula (6):

[0139] Wherein, r(n) includes one or more identifiers, and c(n) represents a pseudo-random sequence.

[0140] Optionally, the correspondence between one or more identifiers included in r(n) and the data stream is described in the above S101 and will not be repeated here.

[0141] In this example, the generation of the identifier can reuse the pseudo-random sequence generation method in the NR protocol, and improve the parameter c in the pseudo-random sequence generation method. init , so that the generated identifier can accurately identify the MIMO data streams of different layers, which is conducive to distinguishing different data streams.

[0142] Example 2: One or more identifiers corresponding to a data stream are determined based on a complex number sequence or a real number sequence, and the complex number sequence or the real number sequence is determined based on a random sequence and a mapping relationship.

[0143] Wherein, the mapping relationship includes a first mapping relationship between multiple complex identifiers and multiple index values, and / or a second mapping relationship between multiple real identifiers and multiple index values. The first device can determine the corresponding real identifier or complex identifier based on the index value and the mapping relationship by determining the index value. Wherein, the first mapping relationship and / or the second mapping relationship can be predefined by the first device, or can be obtained through RRC configuration. For example, both the data sending end and the data receiving end can predefine the first mapping relationship, and / or predefine the second mapping relationship, which can be directly used when querying is needed. For another example, assuming that the first device is a terminal, the first mapping relationship and / or the mapping relationship can be configured by the base station for the terminal through RRC configuration.

[0144] In a possible implementation, the first mapping relationship between the complex sequence and the index value is shown in Table 1.

[0145] Table 1: The first mapping relationship table between complex number sequences and index values.

[0146] Each index value in Table 1 corresponds to a value of a complex identifier. For example, when the index value is 0, the corresponding complex identifier value is When the index value is 1, the corresponding plural identifier value is

[0147] In a possible implementation, the second mapping relationship between the real number sequence and the index value is shown in Table 2.

[0148] Table 2: Second mapping relationship table between real number sequence and index value.

[0149] Each index value in Table 2 corresponds to a real identifier value. For example, when the index value is 0, the corresponding complex identifier value is 1; when the index value is 1, the corresponding complex identifier value is -1.

[0150] Optionally, Table 1 or Table 2 above is merely an example, and this application does not limit the mapping relationship between real number sequences or complex number sequences and index values. For example, the mapping relationship between real number sequences and index values ​​may not be in the form of a table, but may be represented in the form of a set. For example, the first mapping relationship may be a mapping relationship between a first set of index values ​​and a second set of complex identifier values.

[0151] In one possible implementation, the first device determines one or more identifier sequences, including the following steps:

[0152] (1) The first device determines a random sequence based on the user identification and / or the index of the data stream.

[0153] The values ​​in the random sequence are index values ​​in the mapping relationship. Optionally, the maximum value in the random sequence is the maximum index value in the first mapping relationship or the second mapping relationship.

[0154] For example, assuming that the maximum index value in the above mapping relationship is L, the random sequence determined by the first device is as shown in formula (7): I(n)~U(0, L) (7)

[0155] Where I(n) represents a random sequence, ~U(0, L) represents a uniform distribution with parameter L, and L represents the maximum index value of the mapping relationship (e.g., the maximum index value in Table 1 is 3). According to formula (7), the random sequence I(n) satisfies the uniform distribution with parameter L. Optionally, the random sequence can also satisfy a truncated normal distribution with parameter L, where L represents the maximum index value of the mapping relationship.

[0156] Optionally, the random sequence I(n) is based on n layer and / or n RNTI Determined, or based on n layer and / or n RNTI The relevant function determines the layer and / or nRNTI It can be RRC configuration or DCI indication. The specific implementation method can refer to the corresponding description above and will not be repeated here.

[0157] (2) The first device searches for a complex identifier or a real identifier in a mapping relationship corresponding to the value in the random sequence based on the value of the random sequence, and obtains a complex number sequence or a real number sequence.

[0158] For example, assuming that the random sequence I(n) generated by the first device is {0, 1, 1, 2, 3}, according to the value of the random sequence, the identifier corresponding to the same index value is searched from the first mapping relationship or the second mapping relationship, such as searching the plural identifiers corresponding to the index values ​​{0, 1, 1, 2, 3} from the first mapping relationship. The above-mentioned multiple plural identifiers are multiple identifiers corresponding to the random sequence. The first device completes the search and obtains multiple identifiers.

[0159] In this second example, the identifier may be generated by using the value of the random sequence as the index value in the mapping relationship, and the corresponding identifier may be determined by searching the index value in the mapping relationship.

[0160] Example 3: One or more identifiers corresponding to the data stream are determined based on the first functional relationship.

[0161] The parameters of the first functional relationship are related to one or more of the number of data streams, the maximum number of streams supported, the maximum dimension of the identifier, the dimension index of the identifier, and the user identifier. The maximum dimension of the identifier can be equal to the dimension of the high-dimensional space, and the first device can perform dimensionality upscaling on the multi-layer data stream in the high-dimensional space to obtain the multi-layer data stream in the high-dimensional space.

[0162] In one possible implementation, the first functional relationship is shown in formula (8):

[0163] Among them, r(l,n RNTI , i) includes one or more identifiers, l represents the number of data streams, d represents the maximum dimension of the identifier, i represents the dimension index of the identifier (where i ranges from 0 to (d-1)), M represents the maximum number of streams supported, n RNTI Represents the user ID, cos represents the cosine function, and sin represents the sine function.

[0164] Optionally, the above-mentioned first functional relationship is only an example, and the first functional relationship may also be other variations, such as using other functions besides the sine function or the cosine function, which is not limited in this application.

[0165] In this third example, the identifier may be generated based on the first functional relationship, and the parameters of the first function may be designed to be related to the data stream, which is helpful for distinguishing different data streams.

[0166] Optionally, Examples 1 to 3 in this section may be performed individually or in combination with the steps in S101, and this application does not limit this.

[0167] 4. The following describes in detail the method for associating data streams with corresponding identifiers.

[0168] Since the data streams are associated with their respective identifiers, the data streams and their corresponding identifiers can be associated before being input into the training unit to obtain the input information of the training unit. The following describes several examples of how the data transmitter or receiver can associate data streams with corresponding identifiers.

[0169] Example 1: The dimensions of the data stream and the corresponding identifier remain unchanged, and they are directly concatenated to obtain the concatenated vector as the input information of the training unit.

[0170] For example, Figure 3 is a schematic diagram of a data stream provided by the present application, which is cascaded with the corresponding identifier and input into a training unit. The schematic diagram includes the processing flow of the data sending end and the data receiving end respectively, wherein the processing flow of the data sending end and the data receiving end respectively is similar, and the difference lies in the dimension of the data stream and the input or output information.

[0171] 1. In a possible implementation, for the data sending end, the specific cascade processing flow may include the following steps:

[0172] (1) The data transmitter concatenates each data stream with the corresponding identifier to obtain a concatenation vector. The dimension of the concatenation vector is determined based on the number of data streams and the number of symbols in each data stream.

[0173] (2) The data sending end inputs the cascade vector into the training unit.

[0174] For example, one or more data streams at a data transmitter can be represented as X, with the dimension of X being l×n. Here, l represents the number of data streams, n represents the number of symbols in each data stream, and l×n represents the total number of symbols in the transmitted data streams. In one possible implementation, assuming that each symbol in data stream X corresponds to an identifier, data stream X corresponds to multiple identifiers, and a set of multiple identifiers can be called an identifier sequence. For example, assuming that the identifier sequence at the data transmitter is represented as Z1, and assuming that Z1 includes multiple real symbols or multiple complex symbols. Z1 can be represented in matrix form. Since each symbol in data stream X corresponds to an identifier, l×n symbols correspond to l×n identifiers. The dimension of the Z1 matrix is ​​l×n, meaning that the dimension of the identifier sequence Z1 is the same as that of data stream X. In another possible implementation, assuming that some or all symbols in data stream X correspond to an identifier. For ease of description, assuming that each data stream in data stream X corresponds to an identifier, each of the l data streams corresponds to l identifiers. The dimension of the identifier sequence Z1 corresponding to data stream X can then be represented as l×1. Optionally, if at least two of the multiple data streams have different numbers of symbols, the data transmitter may perform zero padding on the data stream with fewer symbols so that the numbers of symbols of the multiple data streams are the same.

[0175] For example, the data transmitter concatenates each data stream with the corresponding identifier, and the resulting concatenation vector X′=CONCAT(X, Z1). Here, X′ represents the concatenation vector of the data transmitter (i.e., the input information of the training unit), and CONCAT(X, Z1) represents the concatenation of X and Z1 using the CONCAT function. The dimension of the concatenation vector X′ is l×n×2. Assume that the data stream X at the data transmitter is a complex data stream. ( represents a complex sequence), the identifier sequence Z1 corresponding to the data stream X is expressed as The concatenated vector is expressed as Similarly, assume that the data stream X at the data sending end is a real data stream ( represents a real number sequence), the identifier sequence Z1 corresponding to the data stream X is expressed as The concatenated vector is expressed as Alternatively, if the dimension of the identifier sequence Z1 corresponding to the data stream X is l×1, the identifier sequence Z1 corresponding to the data stream X can be expressed as The concatenated vector is expressed as Optionally, the data transmitter may also use other concatenation methods to concatenate each data stream with the corresponding identifier to obtain a concatenation vector. For example, other functions or methods may be used to concatenate the data stream X and its corresponding identifier sequence Z1 to obtain a concatenation vector. However, the output information of the training unit needs to be processed for dimensionality reduction, reducing the 2l dimensions to 1. The specific implementation method is not limited in this application.

[0176] The data transmitter inputs the concatenated vector X′ into the training unit to obtain output information. For example, assuming the training unit is a multi-layer neural network, the multi-layer neural network can use the same neural network processing. The training unit uses the multi-layer neural network to train the concatenated vector X′ to obtain a data stream with identifier information (such as multiple transmission symbols, where the multiple transmission symbols include data stream symbols and identifiers). The specific training process can use a training method such as CNN, which is not limited in this application.

[0177] 2. In another possible implementation, for the data receiving end, the specific cascade processing process is similar to the process of the data sending end. The difference lies in the dimension of the data flow at the data receiving end is different from the dimension of the data flow at the data sending end, and the input and output information are different.

[0178] For example, one or more data streams at a data receiver can be represented as Y, with the dimensions of Y being l×n×r. Here, l represents the number of data streams, n represents the number of symbols in each data stream, and r represents the number of receiving antennas. Thus, l×n×r represents the total number of symbols in the received data streams. In one possible implementation, assuming that each symbol in data stream Y corresponds to an identifier, data stream Y corresponds to multiple identifiers, and a set of multiple identifiers can be called an identifier sequence. For example, assuming that the identifier sequence at the data receiver is represented as Z2, and assuming that Z2 includes multiple real symbols or multiple complex symbols. Z2 can be represented in matrix form. Since each symbol in data stream Y corresponds to an identifier, l×n×r symbols correspond to l×n×r identifiers. The dimensions of the Z2 matrix are l×n×r, meaning that the dimensions of the identifier sequence Z2 are the same as those of data stream Y. In another possible implementation, assuming that some or all symbols in data stream Y correspond to an identifier. For ease of description, assuming that each data stream in data stream Y corresponds to an identifier, each of the l data streams corresponds to l identifiers. Then the dimension of the identifier sequence Z2 corresponding to the data stream Y can be expressed as l×1×r.

[0179] For example, the data receiving end concatenates each data stream with the corresponding identifier, and the resulting concatenation vector Among them, Y′ represents the cascade vector of the data receiving end (i.e., the input information of the training unit), represents the matrix obtained after Y is repeated l times, Indicates that the CONCAT function is used to and Z2 for cascade processing. The dimension of the cascade vector Y′ is expressed as l×n×(r+1). Assume that the data stream Y at the data receiving end is a complex data stream The identifier sequence Z2 corresponding to the data stream Y is expressed as Then the concatenated vector Y′ is expressed as Similarly, assuming that the data stream Y at the data receiving end is a real data stream The identifier sequence Z2 corresponding to the data stream Y is expressed as The concatenated vector is expressed as Alternatively, if the dimension of the identifier sequence Z2 corresponding to the data stream Y is l×1×1 (the last dimension is copied r times, and the dimension is updated to l×1×r), then the identifier sequence Z2 corresponding to the data stream Y can be expressed as The concatenated vector is expressed as Optionally, the data receiving end may also use other concatenation methods to concatenate each data stream with the corresponding identifier to obtain a concatenation vector. An example is given in the description of the reference data sending end, which will not be repeated here.

[0180] The data receiving end inputs the concatenated vector Y' into the training unit to obtain output information. For example, assuming the training unit uses a multi-layer neural network to train the concatenated vector Y', a data stream containing identifier information can be obtained. Optionally, the data receiving end can perform calculations based on the data stream containing identifier information to obtain LLRs. The specific training process can use training methods such as CNN, which is not limited in this application.

[0181] Example 2: The dimensions of the data stream and the corresponding identifier remain unchanged, and they are directly concatenated to obtain a concatenated vector for dimensionality upgrading. The high-dimensional vector obtained after dimensionality upgrading is used as the input information of the training unit.

[0182] For example, Figure 4 is a schematic diagram of another data stream provided by the present application and the corresponding identifier for cascade processing and input into the training unit. The schematic diagram includes the processing flows of the data sending end and the data receiving end respectively, wherein the processing flows of the data sending end and the data receiving end are similar, and the difference lies in the dimension of the data stream and the input or output information. In addition, this example 2 proposes that data can be identified in a high-dimensional space. Since AI usually performs feature extraction and / or signal processing in a latent space (also called a high-dimensional space), the data sending end or the data receiving end can identify the data in a high-dimensional space. The process shown in Figure 4 is similar to that of Figure 3, except that the cascade vector is subjected to dimensionality upgrade processing, and the high-dimensional vector obtained after dimensionality upgrade is used as the input information of the training unit. The following focuses on the difference between Figures 3 and 4.

[0183] 1. In a possible implementation, for the data sending end, the specific cascading and dimensionality upgrading process may include the following steps:

[0184] (1) The data transmitter concatenates each data stream with the corresponding identifier to obtain a concatenation vector. The dimension of the concatenation vector is determined based on the number of data streams and the number of symbols in each data stream.

[0185] For the specific implementation of this step, please refer to the detailed description of cascade processing in Example 1 of this section, which will not be repeated here.

[0186] (2) The data sending end performs dimensionality-upgrading on the concatenated vector to obtain a high-dimensional vector X d .

[0187] For example, for the data transmitter, the concatenation vector X′ can be expressed as The dimension-increasing unit at the data sending end performs dimension-increasing processing on the concatenated vector X′ to obtain a high-dimensional vector Xd expressed as Wherein, d represents the identifier of the data stream in the high-dimensional space, and the value of d can be configured by RRC or indicated by DCI. Wherein, the identifier d of the high-dimensional data stream in the high-dimensional space can be the same as the maximum dimension of the identifier or can be greater than the maximum dimension of the identifier, which is not limited in this application.

[0188] (3) The data sending end converts the high-dimensional vector X d Input the training unit and get the output information.

[0189] For example, different from Figure 3, the data sending end converts the high-dimensional vector X d Input the training unit and get the output information. For example, suppose the training unit uses a multi-layer neural network to train the concatenated vector X d The training can obtain a data stream with identifier information. The specific training process can adopt a training method such as CNN, which is not limited in this application.

[0190] 2. In another possible implementation, for the data receiving end, the specific cascade and dimensionality increase processing process is similar to the process of the data sending end. The difference lies in the dimension of the data stream at the data receiving end is different from the dimension of the data stream at the data sending end, and the input and output information are different.

[0191] For example, for the data receiving end, the concatenation vector Y′ can be expressed as The dimension-increasing unit at the data receiving end performs dimension-increasing processing on the concatenated vector Y′ to obtain a high-dimensional vector Y d Expressed as The data receiving end converts the high-dimensional vector Y d Input the training unit and get the output information. The implementation method can refer to the corresponding description in Example 1 of this section, and will not be repeated here.

[0192] Optionally, FIG4 also includes a dimensionality reduction unit for performing dimensionality reduction processing on the high-dimensional data stream output by the training unit. For example, the dimensionality reduction unit can reduce the dimensionality of the high-dimensional data stream to the dimension before the dimensionality increase processing, such as reducing the dimension to l×n×2. The specific dimensionality reduction method can refer to existing dimensionality reduction methods and is not limited in this application.

[0193] Example 3: Perform dimensionality-upgrading processing on the data stream, perform a first operation on the data stream after dimensionality upgrading and the corresponding identifier in the high-dimensional space, and use the result of the first operation as input information of the training unit.

[0194] For example, Figure 5 is a schematic diagram provided by the present application for performing dimensionality-upgrading processing on a data stream, performing a first operation on the upgraded data stream and the corresponding identifier in a high-dimensional space, and inputting the result into a training unit. The schematic diagram includes the processing flows for the data transmitter and the data receiver, respectively. The processing flows for the data transmitter and the data receiver are similar, differing only in the dimension of the data stream and the input or output information.

[0195] 1. In a possible implementation, for a data transmitting end, the specific dimension-upgrading process and the first operation flow of the dimension-upgraded data stream and the corresponding identifier may include the following steps:

[0196] (1) The data transmitter performs dimensionality-upgrading on the data stream to obtain a high-dimensional vector. The dimension of the high-dimensional vector is determined based on the number of data streams and the number of symbols in each data stream.

[0197] For example, one or more data streams at the data sending end can be represented as X, and the dimension of X is represented as l×n; the data sending end performs dimensionality-upgrading on each data stream to obtain a high-dimensional vector X d It can be expressed as Wherein, d represents the identifier of the data flow in the high-dimensional space, and the value of d may be configured by RRC or indicated by DCI.

[0198] (2) The data sending end performs a first operation on the high-dimensional vector and one or more identifiers, and inputs the result of the first operation into the training unit.

[0199] The first operation may include but is not limited to conventional mathematical operations such as addition, subtraction, multiplication, and division, cascade processing, etc. The result of the first operation includes a data stream with identifier information, and the data stream is also a high-dimensional vector.

[0200] For example, assuming that each symbol in the data stream X corresponds to an identifier, if the dimension of X is expressed as l×n, then the dimension of the identifier sequence Z1 corresponding to the data stream X is expressed as l×n. d Expressed as Then the identifier sequence Z1 can be expressed as (such as Z1 and X d If the dimensions are inconsistent, the final d dimensions can be obtained by l×n replication or by using a dimension-raising unit to perform dimension-raising processing. For example, if the first operation is a conventional mathematical operation such as addition, subtraction, multiplication, and division, then the high-dimensional vector X d The dimension of the identifier sequence Z1 needs to be consistent. Therefore, the data sending end can first perform dimension-upgrading on the identifier sequence Z1 to obtain the identifier sequence after dimension-upgrading. Then the high-dimensional vector X d The sum identifier sequence Z′1 will undergo the first operation.

[0201] 2. In another possible implementation, for the data receiving end, the specific cascade processing process is similar to the process of the data sending end. The difference lies in the dimension of the data flow at the data receiving end is different from the dimension of the data flow at the data sending end, and the input and output information are different.

[0202] For example, for the data receiving end, assuming that each symbol of the data stream Y corresponds to an identifier, if the dimension of Y is expressed as l×n×r, then the dimension of the identifier sequence Z2 corresponding to the data stream Y is expressed as l×n×r. d Expressed as Then the identifier sequence Z2 can be expressed as If the first operation is an addition operation, the result of the first operation can be expressed as Y d +Z2. Optionally, if the first operation is a conventional mathematical operation such as addition, subtraction, multiplication, or division, then the high-dimensional vector Y d The dimension of the identifier sequence Z2 needs to be consistent. For the specific implementation method, please refer to the similar description of the data sending end, which will not be repeated here.

[0203] Optionally, for Examples 1 to 3 of this subsection, the output information of the data transmitting end (such as a data stream with identifier information) is input into the data receiving end after being transmitted through the channel. For example, the received data stream Y of the data receiving end in Figure 3 can be expressed as Y=HX+N, where H represents the channel, X represents the transmitted data stream of the data transmitting end, and N represents noise. In one possible implementation, the data stream of the data transmitting end is associated with one or more identifiers, and the obtained output information includes a data stream with identifier information. The data receiving end may no longer associate the received data stream with one or more identifiers (such as the received data stream already carries identifier information, and the training unit of the data receiving end can distinguish different data streams). For example, the data receiving end in Figures 3 to 5 may only include a training unit and output information, and the input information of the training unit is the output information of the data transmitting end.

[0204] Optionally, the output information of the data receiving end is used by the data receiving end and the data transmitting end to update their respective training units. For example, the output information of the data receiving end can be used to solve the loss function with the transmitted data stream to obtain the gradient of each trainable parameter in the training unit. The parameters of the training units of the data transmitting end and the data receiving end are updated based on the gradient of the trainable parameters. The transmitted data stream can be training data (such as the transmitted data stream of the data transmitting end in Figure 3) or the correct decoding result obtained by the data receiving end.

[0205] Optionally, for Examples 1 to 3 of this section, the identifier sequence of the data transmitter and the identifier sequence of the data receiver may correspond to each other. For example, the identifier sequence Z1 of the data transmitter and the identifier sequence Z2 of the data receiver may be the same, or they may be different but related (e.g., Z2 is the identifier sequence obtained by performing a matrix transformation (e.g., transposition) on Z1). This relationship is not limited in this application.

[0206] Optionally, for Examples 1 through 3 of this section, the method for associating data streams with corresponding identifiers at the data transmitter and the method for associating identifiers corresponding to data streams at the data receiver may be different. For example, the data transmitter may use the method described in Example 1, while the data receiver may use the method described in Example 2, and there may be a corresponding relationship between the identifier sequence at the data transmitter and the identifier sequence at the data receiver.

[0207] Optionally, Examples 1 to 3 in this section may be performed individually or in combination with the steps in S102, and this application does not limit this.

[0208] In order to realize the various functions in the method provided by the present application, the device or equipment provided by the present application may include a hardware structure and / or a software module, and realize the above-mentioned various functions in the form of a hardware structure, a software module, or a hardware structure plus a software module. Whether a certain function among the above-mentioned functions is executed in the form of a hardware structure, a software module, or a hardware structure plus a software module depends on the specific application and design constraints of the technical solution. The division of modules in the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, the various functional modules in the various embodiments of the present application can be integrated into a processor, or they can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software functional modules.

[0209] Figure 6 is a schematic diagram of a communication device provided by the present application. The device may include a module corresponding to each of the methods / operations / steps / actions described in any of the embodiments shown in Figures 2 to 5. The module may be implemented as a hardware circuit, software, or a combination of hardware circuits and software.

[0210] The apparatus 600 includes a communication unit 601 and a processing unit 602, which are used to implement the methods executed by the devices in the above embodiments. The communication unit 601 is also called a transceiver unit, which includes a sending unit and a receiving unit. The sending unit is used to send signals, and the receiving unit is used to receive signals.

[0211] In one possible embodiment, the apparatus is a device or apparatus at a data transmitter. Specifically, the processing unit is configured to determine one or more data streams, each corresponding to one or more identifiers. The processing unit is further configured to input the one or more data streams and their corresponding identifiers into a training unit to obtain output information; the training unit includes one or more artificial intelligence (AI) modules in the transmitter that perform signal processing.

[0212] Optionally, the communication unit 601 is used to send output information, where the output information is input information of the data receiving end.

[0213] The specific execution process of the communication unit 601 and the processing unit 602 in this embodiment can refer to the description of the steps executed by the data sending end in the previous method embodiment, as well as the related description, which will not be repeated here. The data processing method implemented by the device can associate the data stream with the identifier, which is beneficial for the data receiving end (such as the receiver) to distinguish different data streams through the identifier. In addition, when the training unit is performing AI training, different data streams can be distinguished by different identifiers, so the training of the data stream can share parameters (such as reusing the same neural network processing method), which is beneficial to improving the generalization and scalability of the network.

[0214] In one possible embodiment, the apparatus is a device or apparatus at a data receiving end. Specifically, processing unit 602 is configured to determine one or more data streams, each corresponding to one or more identifiers. Processing unit 602 is further configured to input the one or more data streams and their corresponding identifiers into a training unit to obtain output information; the training unit includes one or more artificial intelligence (AI) modules in the receiver that perform signal processing.

[0215] Optionally, the communication unit 601 is used to receive output information, where the output information is output information of the data receiving end. The output information of the data receiving end is used by the data receiving end and the data sending end to update their respective training units.

[0216] The specific execution process of the communication unit 601 and the processing unit 602 in this embodiment can refer to the description of the steps performed by the data receiving end in the previous method embodiment, as well as the related description, which will not be repeated here. The data processing method implemented by the device can associate the received data stream with an identifier, so that the data receiving end can distinguish different data streams by the identifier. In addition, when the training unit performs AI training, different data streams can be distinguished by different identifiers, so that the training of the data streams can share parameters (such as reusing the same neural network processing method), which is beneficial to improving the generalization and scalability of the network.

[0217] In one possible implementation, when the communication device is a chip, the chip includes a transceiver unit and a processing unit. The transceiver unit may be an input / output circuit or a communication interface; the processing unit may be a processor, microprocessor, integrated circuit, or logic circuit integrated on the chip.

[0218] This application also provides a communication device. See Figure 7 for another schematic diagram of the structure of the communication device according to an embodiment of the present application. The communication device can be used to execute the steps performed by the data transmitter or receiver, or the first device or the second device, in the method embodiments described above. For details, see the relevant descriptions in the method embodiments described above.

[0219] The communication device includes a processor 701. Optionally, the communication device further includes a memory 702 and a transceiver 703.

[0220] In a possible implementation, the processor 701 , the memory 702 , and the transceiver 703 are connected via buses, and computer instructions are stored in the memory.

[0221] Optionally, the processing unit 602 in the aforementioned embodiment may specifically be the processor 701 in this embodiment, so the specific implementation of the processor 701 is not repeated. The communication unit 601 in the aforementioned embodiment may specifically be the transceiver 703 in this embodiment, so the specific implementation of the transceiver 703 is not repeated.

[0222] In this application, a processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or execute the methods, steps, and logic block diagrams disclosed in this application. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in this application may be directly executed by a hardware processor, or by a combination of hardware and software modules within the processor.

[0223] In the present application, the memory may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), or a volatile memory, such as a random-access memory (RAM). The memory is any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory in the present application may also be a circuit or any other device that can implement a storage function, for storing program instructions and / or data.

[0224] The present application provides another communication device, which includes a processor and an interface. Optionally, it also includes a memory, the processor is coupled to the memory, and the processor is configured to read and execute computer instructions stored in the memory to implement the data processing method in the embodiments shown in Figures 2 to 5.

[0225] The present application also provides a communication system including a first device and a second device. The first device is configured to execute all or part of the steps executed by the first device in the above embodiment. The second device is configured to execute all or part of the steps executed by the second device in the above embodiment.

[0226] The present application provides a computer-readable storage medium. The computer-readable storage medium stores a program or instruction. When the program or instruction is executed on a computer, the computer executes the data processing method in the embodiments shown in Figures 2 to 5.

[0227] The present application provides a computer program product. The computer program product includes instructions. When the instructions are executed on a computer, the computer executes the data processing method in the embodiments shown in Figures 2 to 5.

[0228] The present application provides a chip or a chip system, which includes at least one processor and an interface, the interface and the at least one processor are interconnected by lines, and the at least one processor is used to run computer programs or instructions to execute the data processing method in the embodiments shown in Figures 2 to 5.

[0229] The interface in the chip may be an input / output interface, a pin, or a circuit.

[0230] The chip system may be a system on chip (SOC) or a baseband chip, wherein the baseband chip may include a processor, a channel encoder, a digital signal processor, a modem, an interface module, and the like.

[0231] In one implementation, the chip or chip system described above in this application further includes at least one memory, in which instructions are stored. The memory may be a storage unit within the chip, such as a register, a cache, etc., or a storage unit of the chip (e.g., a read-only memory, a random access memory, etc.).

[0232] The technical solutions provided in this application can be implemented in whole or in part through software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in this application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a terminal, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium.

[0233] In this application, under the premise that there is no logical contradiction, the various embodiments may reference each other, for example, the methods and / or terms between method embodiments may reference each other, for example, the functions and / or terms between device embodiments may reference each other, for example, the functions and / or terms between device embodiments and method embodiments may reference each other.

[0234] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include these modifications and variations.

Claims

1. A data processing method, characterized in that: include: Determine one or more data streams, wherein each data stream corresponds to one or more identifiers; The one or more data streams and their respective corresponding identifiers are input into a training unit to obtain output information; the training unit includes one or more artificial intelligence AI modules for signal processing in a transmitter or a receiver.

2. The method according to claim 1, characterized in that A data stream includes multiple symbols, each of which corresponds to an identifier.

3. The method according to claim 1, characterized in that A data stream includes a plurality of symbols, and some or all of the plurality of symbols correspond to an identifier.

4. The method according to claim 2 or 3, characterized in that: Different data streams correspond to different identifiers.

5. The method according to any one of claims 1 to 4, characterized in that: The step of inputting the one or more data streams and their corresponding identifiers into a training unit comprises: Concatenate each data stream with the corresponding identifier to obtain a concatenation vector; the dimension of the concatenation vector is determined based on the number of data streams and the number of symbols of each data stream; The concatenated vector is input into the training unit.

6. The method according to any one of claims 1 to 4, characterized in that: The step of inputting the one or more data streams and their corresponding identifiers into a training unit comprises: Concatenate each data stream with the corresponding identifier to obtain a concatenated vector; wherein the dimension of the concatenated vector is determined based on the number of data streams and the number of symbols of each data stream; Perform dimension-up processing on the concatenated vector to obtain a high-dimensional vector; The high-dimensional vector is input into the training unit.

7. The method according to any one of claims 1 to 4, characterized in that: The step of inputting the one or more data streams and their corresponding identifiers into a training unit comprises: Performing dimensionality-upgrading processing on each data stream to obtain a high-dimensional vector; the dimension of the high-dimensional vector is determined based on the number of data streams and the number of symbols of each data stream; A first operation is performed on the high-dimensional vector and the one or more identifiers, and a result of the first operation is input into the training unit.

8. The method according to any one of claims 1 to 4, characterized in that: The one or more identifiers are determined based on a pseudo-random sequence, and the pseudo-random sequence is related to one or more of a cell identifier, a user identifier, and an index of a data stream.

9. The method according to claim 8, characterized in that The index of the data stream is indicated by downlink control information.

10. The method according to any one of claims 1 to 4, characterized in that: The one or more identifiers are determined based on a complex number sequence or a real number sequence, and the complex number sequence or the real number sequence is determined based on a random sequence and a mapping relationship; The mapping relationship includes a first mapping relationship between a plurality of complex identifiers and a plurality of index values, and / or a second mapping relationship between a plurality of real identifiers and a plurality of index values; The random sequence is determined based on a user identifier and / or an index of a data stream, and a value in the random sequence is an index value in the first mapping relationship and / or the second mapping relationship.

11. The method according to any one of claims 1 to 4, characterized in that: The one or more identifiers are determined based on a first functional relationship, and parameters of the first functional relationship are related to one or more of the number of data streams, the maximum number of streams supported, the maximum dimension of the identifier, the dimension index of the identifier, and the user identification.

12. The method according to claim 11, characterized in that The first functional relationship satisfies the following formula: Among them, the r(l,n RNTI ,i) includes one or more identifiers, wherein l represents the number of data streams, d represents the maximum dimension of the identifier, i represents the dimension index of the identifier, M represents the maximum number of streams supported, and n RNTI represents a user identifier, cos represents a cosine function, and sin represents a sine function.

13. The method according to claim 1, characterized in that The one or more identifiers belong to an identifier sequence of a data sending end, or the one or more identifiers belong to an identifier sequence of a data receiving end; there is a corresponding relationship between the identifier sequence of the data sending end and the identifier sequence of the data receiving end.

14. The method according to claim 13, characterized in that The method further comprises: The data transmitting end sends output information, and the output information is input into the data receiving end after being transmitted through the channel; or, The data receiving end sends output information, and the output information of the data receiving end is used by the data receiving end and the data sending end to update their respective training units.

15. A communication device, characterized in that: The method comprises a communication unit and a processing unit, wherein the communication unit and the processing unit are used to execute the method according to any one of claims 1 to 14.

16. A communication device, characterized in that: include: A processor and a memory, wherein the memory is used to store instructions, and when the instructions are executed by the processor, the communication device executes the method according to any one of claims 1 to 14.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the computer is caused to perform the method according to any one of claims 1 to 14.

18. A computer program product, characterized in that The method comprises instructions which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 14.

19. A chip, characterized in that: The chip includes a processor, or the chip includes a processor and an interface; the processor is used to execute a computer program, so that the chip implements the method according to any one of claims 1 to 14.

20. A chip system, characterized in that: The chip system comprises a processor and an interface, wherein the processor is configured to execute a computer program so that the chip system implements the method as claimed in any one of claims 1 to 14.

Images