Communication method, communication device and storage medium

Abstract

The embodiment of the invention provides a communication method, a communication device and a storage medium, relates to the technical field of communication, and is used for improving the data transmission efficiency in a communication process. The method comprises the following steps: receiving a first communication signal without a pilot symbol; processing the first communication signal through a first model to obtain a second communication signal; and performing signal analysis processing on the second communication signal to obtain communication data.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technology, and in particular to a communication method, a communication device, and a storage medium. Background Art

[0002] In wireless mobile communication systems, the fading of wireless channels can cause serious distortion of transmitted signals. In order to accurately restore the transmitted signals of the transmitter at the receiving end, wireless mobile communication systems usually perform channel estimation by sending pilot symbols to obtain accurate channel state information (CSI), and then use the obtained channel state information to perform channel equalization to eliminate the distortion of the signal waveform caused by the channel. Among them, pilot-based channel estimation methods include maximum likelihood (ML) estimation algorithm, least squares (LS) channel estimation algorithm, minimum mean square error (MMSE) channel estimation algorithm, and compressed sensing (CS) channel estimation algorithm. In this way, the transmitter needs to send a pilot signal, which generates pilot overhead and leads to a decrease in communication efficiency. Summary of the invention

[0003] The present disclosure provides a communication method, a communication device and a storage medium, which are used to improve the data transmission efficiency during the communication process.

[0004] In order to achieve the above objectives, the present disclosure adopts the following technical solutions:

[0005] In a first aspect, the present disclosure provides a communication method, the method comprising:

[0006] receiving a first communication signal without a pilot symbol;

[0007] Processing the first communication signal through the first model to obtain a second communication signal;

[0008] The second communication signal is subjected to signal analysis processing to obtain communication data.

[0009] In a second aspect, a communication device is provided, the device comprising: a receiving module and a processing module.

[0010] A receiving model, configured to receive a first communication signal without a pilot symbol;

[0011] A processing module, configured to process the first communication signal through the first model to obtain a second communication signal;

[0012] The processing module is further used to perform signal analysis processing on the second communication signal to obtain communication data.

[0013] In a third aspect, a communication device is provided, comprising: a memory and a processor; the memory and the processor are coupled; the memory is used to store computer program instructions executable by the processor; and the processor implements a method of any embodiment of the first aspect when executing the computer program instructions.

[0014] In a fourth aspect, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed on a computer (such as a communication device), the method of any embodiment of the first aspect is implemented.

[0015] According to a fifth aspect, a computer program product is provided, which includes computer program instructions, and when the computer program instructions are executed, the method of any embodiment of the first aspect is implemented.

[0016] Based on the technical solution provided by the present disclosure, all symbols can be used to transmit data during the communication process, thereby improving the data transmission efficiency. In addition, the technical solution adopts a low-speed receiver (first model) architecture based on a cascaded neural network, based on which the received communication signal without pilot symbols can be processed, which can more accurately reflect the channel impact and further improve the receiver signal detection performance.

[0017] In addition, the technical solution provided by the present invention adopts a receiver (first model) with a neural network structure that can be driven by both data and model, and can use linearly calculated data as training data input into the first model, thereby reducing the sample demand and training time required. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of the architecture of a communication system provided in an embodiment of the present disclosure;

[0019] Figure 2 A flow chart of a communication method provided in an embodiment of the present disclosure;

[0020] Figure 3 A schematic diagram of a symbol provided for an embodiment of the present disclosure;

[0021] Figure 4 A schematic diagram of a data processing process provided for an embodiment of the present disclosure;

[0022] Figure 5 A flowchart of another communication method provided for an embodiment of the present disclosure;

[0023] Figure 6 A flowchart of another communication method provided for an embodiment of the present disclosure;

[0024] Figure 7 A schematic diagram of a data processing process of a receiver provided in an embodiment of the present disclosure;

[0025] Figure 8 A flowchart of another communication method provided for an embodiment of the present disclosure;

[0026] Fig. 9 A schematic diagram of another symbol provided for an embodiment of the present disclosure;

[0027] Fig.10 A flowchart of another communication method provided for an embodiment of the present disclosure;

[0028] Fig.11 A schematic diagram of the composition of a communication device provided in an embodiment of the present disclosure;

[0029] Fig.12 A schematic diagram of the structure of a communication device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.

[0031] In the description of the present disclosure, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, "at least one" means one or more, and "plurality" means two or more. The words "first", "second", etc. do not limit the quantity and execution order, and the words "first", "second", etc. do not limit them to be different.

[0032] It should be noted that in the present disclosure, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the present disclosure should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0033] Since wireless channels are not fixed and predictable like wired channels, but have great randomness, the performance of wireless communication systems is greatly affected by wireless channels, such as shadow fading and frequency selective fading. In order to accurately restore the transmitted signal of the transmitter at the receiving end, the receiving end usually uses channel equalization technology to compensate for channel fading. At present, orthogonal frequency division multiplexing (OFDM) receiver systems usually use mathematical formulas for signal detection. For example, linear detection using zero-forcing algorithms and minimum mean square error detection algorithms has low detection accuracy, and this method is not applicable due to the complex channel environment, which is difficult to express with mathematical formulas. The receiver complexity corresponding to the nonlinear detection method is high. In addition, it is proposed to use wireless communication technology based on neural network architecture to more accurately fit the channel impact and further improve the detection performance. However, the transmitting end needs to send a pilot signal to assist in obtaining accurate channel state information, thereby generating pilot overhead and causing the communication efficiency of the wireless mobile communication system to decrease.

[0034] In view of this, the present disclosure provides a communication method, which includes: a receiving end receives a first communication signal without a pilot symbol, processes the communication signal through a first model to obtain a second communication signal, and performs signal analysis processing on the second communication signal to obtain communication data. In this way, based on the low-speed receiver (first model) architecture of the cascaded neural network, the received communication signal without a pilot symbol can be processed, so that all symbols can be used to transmit data, thereby improving data transmission efficiency.

[0035] The method provided by the present disclosure can be applied to various communication systems. For example, the communication system can be a long term evolution (LTE) system, a fifth generation (5G) communication system, a Wi-Fi system, a communication system related to the third generation partnership project (3GPP), a future evolution communication system (such as a sixth generation (6G) communication system, etc.), or a system integrating multiple systems, etc., without limitation. Figure 1 Taking the communication system 100 as an example, the method provided by the embodiment of the present disclosure is described. Figure 1 It is only a schematic diagram and does not constitute a limitation on the applicable scenarios of the technical solution provided by the present disclosure.

[0036] Figure 1 The following is a schematic diagram of the architecture of a communication system provided by an embodiment of the present disclosure. Figure 1As shown, the communication system 100 may include one or more network devices 11 and one or more terminal devices 12. The terminal device 12 may be communicatively connected with the one or more network devices 11.

[0037] In some embodiments, the network device 11 can be used to implement functions such as resource scheduling, wireless resource management, and wireless access control of terminal devices. Specifically, the network device can be any node in a base station transceiver, a wireless base station, a wireless transceiver, a small base station, a wireless access point, a transmission receive point (TRP), a transmission point (TP), an evolved node B (eNB), a home node B, a home evolved node B, a reader, and some other access nodes. In some embodiments, the communication system 100 may also include different types of base stations, such as a macro cell base station and / or a small cell base station.

[0038] The terminal device 12 may also be referred to as a terminal, user equipment (UE), mobile station, mobile terminal, etc. The terminal device may be an Internet of Things (IoT) device, which is used for various information collection and then forward error correction coding of the data information before sending the data to the base station. The terminal device may also be a mobile phone, a tablet computer, a computer with wireless transceiver function, a car, a tram, and other mobile devices. In addition, the terminal device may be fixed or mobile. Various types of terminal devices may also include or be referred to by those skilled in the art as mobile stations, user stations, mobile units, user units, wireless units, remote units, mobile devices, wireless devices, wireless communication devices, remote devices, mobile user stations, access terminals, mobile terminals, wireless terminals, remote terminals, handheld devices, user agents, mobile clients, clients, passive tags, or some other possible devices. In addition, various types of UEs may also be cellular phones, personal digital assistants (PDAs), wireless modems, wireless communication devices, handheld devices, tablet computers, laptop computers, cordless phones, wireless local loop (WLL) stations, and the like. Various types of UEs may communicate with various types of base stations and network devices (including macro eNBs, small cell eNBs, relay base stations, etc.) The wireless communication system 100 may also include an IoT system or be a part of an IoT system.

[0039] In some embodiments, during the communication process, the network device sends data to the terminal device, and the terminal device receives the data sent by the network device. Thus, the network device can be called a transmitting end. Correspondingly, the terminal device can be called a receiving end. Alternatively, the terminal device sends data to the network device, and the network device can be called a receiving end. Correspondingly, the terminal device can be called a transmitting end.

[0040] It should be noted that Figure 1 This is just an exemplary framework diagram. Figure 1 The number of devices or nodes included in the Figure 1 In addition to the functional nodes shown, the communication system may also include other nodes or devices, such as core network devices.

[0041] The application scenarios of the embodiments of the present disclosure are not limited. The system architecture and business scenarios described in the embodiments of the present disclosure are intended to more clearly illustrate the technical solutions of the embodiments of the present disclosure, and do not constitute a limitation on the technical solutions provided by the embodiments of the present disclosure. It is known to those skilled in the art that with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present disclosure are also applicable to similar technical problems.

[0042] The embodiments provided by the present disclosure are described in detail below in conjunction with the accompanying drawings.

[0043] like Figure 2 As shown, an embodiment of the present disclosure provides a communication method, which is applied to a receiving end and includes the following steps:

[0044] S101. Receive a first communication signal without a pilot symbol.

[0045] In some embodiments, all symbols included in the first communication signal may be used to carry data.

[0046] It should be noted that the transmitting end can send the first communication signal without pilot symbols to the receiving end, and correspondingly, the receiving end receives the first communication signal. There are no pilot symbols in the communication process, that is, all symbols can be used to transmit data, which improves data transmission efficiency.

[0047] In one example, Figure 3 As shown, the frequency domain data received by the receiving end has been normalized, including the corresponding data pilot symbol layer 1 and data symbol layer 1.

[0048] S102: Process the first communication signal using the first model to obtain a second communication signal.

[0049] In some embodiments, the sub-model of the first model can be constructed based on a convolutional neural network, a residual neural network, or a residual neural network combined with an attention mechanism.

[0050] In some embodiments, the first model may include a pre-demodulation sub-model, a channel estimation sub-model, and a channel equalization sub-model. That is, the pre-demodulation sub-model, the channel estimation sub-model, and the channel equalization sub-model may be constructed based on a convolutional neural network, a residual neural network, or a residual neural network combined with an attention mechanism. For example, Figure 4 As shown, the receiver at the receiving end may include a pre-demodulation neural network Net1, a channel estimation neural network Net2, and a channel equalization neural network Net3.

[0051] In some embodiments, Figure 4 The transmitter of the pilot-free communication system can use channel state information (CSI) to precode and pre-equalize the transmitted signal to further improve the detection performance of the pilot-free receiver. The channel state information can be obtained through the channel sounding reference signal (SRS) in the time division duplex (TDD) mode, and can be obtained through the UE's channel state information reference signal (CSI-RS) measurement feedback in the frequency division duplex (FDD) mode.

[0052] Thus, the receiving end can process the first communication signal through the pre-demodulation sub-model, the channel estimation sub-model and the channel equalization sub-model to obtain the second communication signal. Figure 5 As shown, the receiving end may process the first communication signal by the following steps S1021 to S1023:

[0053] S1021. Process a first symbol in a first communication signal based on a pre-demodulation sub-model to obtain a second symbol.

[0054] The first symbol may be used to carry data and perform channel estimation. In some embodiments, the first symbol may also be referred to as a data pilot symbol. In addition, other symbols in the first communication signal except the first symbol may be referred to as data symbols, and the data symbols are only used to carry data.

[0055] In some embodiments, the first symbol is the first received symbol in the first communication signal.

[0056] It should be noted that the first symbol can be understood as the first symbol transmitted by the service channel. It can be a symbol preset by the receiving end and the transmitting end. The first symbol received in the first communication signal is taken as the first symbol, so that after receiving the first symbol, the receiving end can first perform channel estimation based on the first symbol to obtain channel estimation parameters, so as to process other symbols after the first symbol in the first communication signal based on the channel estimation parameters.

[0057] In addition, the second symbol can be obtained by processing the first symbol. It should be understood that the second symbol obtained here can be understood as the predicted value output by the pre-demodulation sub-model for the information data sent by the transmitter carried by the first symbol.

[0058] In some embodiments, the receiving end may first perform real number processing on the first symbol to determine a first real number array based on the first symbol, then input the first real number array into a pre-demodulation sub-model to obtain second real number data, and then perform modulation processing on the second real number array to obtain a second symbol.

[0059] In some embodiments, the pre-demodulation sub-model is trained based on the first sample set, wherein the first sample set includes a plurality of first samples and labels corresponding to the plurality of first samples, a first sample is a first symbol, and the label corresponding to the first sample is a third symbol corresponding to the first symbol.

[0060] It should be understood that the second symbol is the predicted value of the information data sent by the transmitter in the data pilot symbol, and the third symbol is the actual value of the information data sent by the transmitter in the data pilot symbol.

[0061] In some embodiments, the first sample set can be determined based on relevant data generated in a simulation platform for a communication system. The simulation platform can simulate the communication process under various communication environments, that is, it can obtain simulation data under a communication channel based on multiple channel characteristics. For example, the simulation data may include the transmission data of the transmitting end and the reception data of the receiving end. Thus, the reception data of the receiving end can be used as a training sample, and the transmission data corresponding to the reception data can be used as a label of the training sample to train the model. Among them, in an embodiment of the present disclosure, a first symbol (data pilot symbol) can be extracted from the received data as the above-mentioned first sample, and a third symbol can be extracted from the transmission data as a label corresponding to the first sample to train the pre-demodulation sub-model. Among them, the third symbol obtained here is the actual value of the information data sent by the transmitting end carried by the first symbol.

[0062] In some embodiments, multiple sets of training samples and labels may be generated under different fading channels.

[0063] In some embodiments, the first symbol extracted from the received data may be converted to a real number and subjected to data rearrangement processing to obtain a first sample.

[0064] For example, the sender can obtain the Figure 4 The processing process in generates the bit stream data X to be modulated bit , and the data Y of the data pilot symbol received by the receiving end through the wireless fading channel PW . Then, we can analyze the data Y PW Perform real number processing and data rearrangement based on the processing subband or processing granularity of the pre-demodulation sub-model to obtain N*X real number arrays. A real number array may include [Re(Y PW ), Im(Y PW )]. The processing subband of the pre-demodulation sub-model is X REs, that is, Y PW , rearranged into N sub-band data. And, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part. Data Y PW The corresponding N*X real number array is a first sample.

[0065] In addition, the sender data X bit The real number processing is performed and the data is rearranged based on the processing subband or processing granularity of the pre-demodulation sub-model to obtain N*X real number arrays. Among them, the j-th real number array can be expressed as Among them, Q m is the modulation order. Then the data X bit The corresponding N*X*Q m An array of real numbers X′ pw,bit That is, the label corresponding to the first sample. Therefore, the pre-demodulation sub-model can be trained based on the obtained first sample and the label corresponding to the first sample.

[0066] In some embodiments, the first sample set may also be determined based on actual communication data in various communication systems. For example, the received data of the receiving end in the historical communication data actually transmitted in the communication system may be used as a training sample, and the sent data of the sending end in the historical communication data may be used as a label corresponding to the training sample to train the model. Similarly, the first symbol (data pilot symbol) may be extracted from the received data in the historical communication data as the above-mentioned first sample, and the third symbol (actually sent data of the sending end) may be extracted from the sent data in the historical communication data as a label corresponding to the first sample to train the pre-demodulation sub-model.

[0067] It should be understood that the above-mentioned method for determining the first sample set is only an example, and the first sample set may also be determined in other possible ways, which is not limited in the present disclosure.

[0068] In some embodiments, the process of training the pre-demodulation sub-model according to the first sample set includes:

[0069] S1. Obtain an initial pre-demodulation sub-model to be trained and a first sample set.

[0070] S2. Input any first sample in the first sample set as a target sample into the initial pre-demodulation sub-model to obtain an output result corresponding to the target sample, that is, a predicted value of the information data carried by the first symbol.

[0071] S3. Compare the output result of the initial pre-demodulation sub-model with the label corresponding to the target sample to determine the loss value.

[0072] S4. Adjust the model parameters of the initial pre-detuning sub-model according to the loss value.

[0073] S5. Determine a new target sample based on the first sample set, and repeat steps S1-S4 until the model converges.

[0074] Furthermore, whether the model has converged can be determined based on the loss value outputted by the initial pre-demodulation sub-model each time during the training process, or whether the sample has converged can be determined based on the number of times the model has been trained, which is not limited in the present disclosure.

[0075] For example, when the loss value is less than a loss value threshold, it can be determined that the model has converged.

[0076] For another example, when the number of training times exceeds a threshold, it can be determined that the model has converged.

[0077] In this way, the converged model can be determined as the pre-detuning sub-model that has been trained.

[0078] In some embodiments, the loss function used in the above training process can be a binary cross entropy loss function, Euclidean distance, cosine distance, a combination of Euclidean distance and cosine distance as a loss function, or other possible loss functions.

[0079] In some embodiments, the pre-demodulation sub-model or the pre-demodulation neural network may be trained using an Adam optimizer.

[0080] Furthermore, after the pre-demodulation sub-model training is completed, the pre-demodulation sub-model can be deployed at the receiving end in the communication system for online signal detection, wherein the first symbol obtained by detection also needs to be processed similar to the above-mentioned first sample so that the processed first symbol can be input into the pre-demodulation sub-model to obtain the second symbol.

[0081] S1022. Process the first symbol and the second symbol based on the channel estimation sub-model to obtain a channel estimation parameter.

[0082] In some embodiments, a third real number array may be determined based on the first symbol and the second symbol, and then the third real number array is input into a channel estimation sub-model to obtain channel estimation parameters.

[0083] In some embodiments, the first symbol and the second symbol may be processed into real numbers and data may be rearranged based on the processing granularity of the channel estimation sub-model to obtain a third real number array.

[0084] In some embodiments, the channel estimation submodel is trained based on the second sample set, wherein the second sample set includes a plurality of second samples and labels corresponding to the plurality of second samples, a second sample is a first symbol and a third symbol corresponding to the first symbol, and the label corresponding to the second sample is a measured channel parameter.

[0085] In some embodiments, the second sample set can also be determined based on relevant data generated in a simulation platform for the communication system. Similarly, the simulation platform can simulate the communication process under various communication environments, that is, it can obtain simulation data under a communication channel based on multiple channel characteristics. Thus, it is possible to extract the first symbol based on the received data of the receiving end and the third symbol based on the sent data of the sending end, as well as the ideal channel H for this communication process in the simulation platform. pw Thus, the first symbol and the third symbol can be used as the second sample, and the ideal channel H pw As labels, we train the channel estimation sub-model.

[0086] In some embodiments, multiple sets of second samples and labels may be generated under different fading channels.

[0087] In some embodiments, the first symbol extracted from the received data and the third symbol extracted from the transmitted data may be converted to real numbers and rearranged to obtain a second sample.

[0088] For example, the sender can obtain the Figure 4 The processing process generates the data pilot symbol modulation data X at the transmitting end PW , and the data Y of the data pilot symbol received by the receiving end through the wireless fading channel PW . Then, we can analyze the data X PW And data Y PWPerform real number processing and data rearrangement based on the processing subband or processing granularity of the channel estimation submodel to obtain N*X real number arrays. A real number array may include [Re(Y PW ), Im(Y PW ), Re(X PW ), Im(X PW )]. The processing subband of the channel estimation submodel is X REs, that is, Y PW , rearranged into N sub-band data. And, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part. Data X PW And data Y PW The corresponding N*X real number array is a second sample.

[0089] In addition, the ideal channel H through which the data pilot symbol (first symbol) passes during the communication process can also be obtained. pw , perform real number processing and data rearrangement based on the processing subband or processing granularity of the channel estimation submodel to obtain N*X real arrays H′ pw Among them, the j-th real number array can be expressed as X′ pw,j =[Re(H pw ), Im(H pw )]. Data H pw The corresponding N*X*2 real number array H′ pw That is, the label corresponding to the second sample. Therefore, the channel estimation sub-model can be trained based on the obtained second sample and the label corresponding to the second sample.

[0090] In some embodiments, the second sample set may also be determined based on actual communication data in various communication systems. For example, the received data of the receiving end and the sent data of the sending end in the historical communication data actually transmitted in the communication system may be used as training samples, and the actual channel parameters may be used as labels to train the model.

[0091] It should be understood that the above-mentioned method for determining the second sample set is only an example, and the second sample set may also be determined in other possible ways, which is not limited in the present disclosure.

[0092] In some embodiments, when the measured channel parameters include channel parameters in multiple antenna dimensions, the channel parameters in multiple antenna dimensions are arranged based on a preset order of the channel estimation submodel, and the channel estimation parameters output by the channel estimation submodel are also arranged based on a preset order.

[0093] For example, in a low-speed scenario, a pilotless communication system is used for transmission using multiple-input multiple-output (MIMO) technology, with L = 2 layers of data streams, P = 2 antenna ports and T = 2 physical transmit antennas, R = 2 physical receive antennas. At this time, the preset order arrangement of the channel based on the channel estimation submodel can be expressed as The channel estimation parameters output by the channel estimation sub-model are also arranged based on the preset order.

[0094] In addition, the process of training the channel estimation sub-model according to the second sample set can refer to the training process of the pre-demodulation sub-model in the above steps S1-S5, which will not be repeated here.

[0095] Furthermore, after the channel estimation sub-model is trained, the channel estimation sub-model can be deployed at the receiving end in the communication system for online signal detection, wherein the detected first symbol and second symbol also need to be processed similar to the above-mentioned second sample so that the processed data can be input into the channel estimation sub-model to obtain channel estimation parameters.

[0096] S1023. Process the first communication signal and the channel estimation parameter based on the channel equalization sub-model to obtain a second communication signal.

[0097] In some embodiments, the first communication signal and the channel estimation parameter may be preprocessed to obtain a third real number array, and the third real number array may be input into the channel equalization sub-model to obtain the second communication signal.

[0098] In some embodiments, the first communication signal and the channel estimation parameters may be processed into real numbers and data may be rearranged based on the processing granularity of the channel equalization sub-model to obtain a fourth real number array.

[0099] In some embodiments, the channel equalization submodel is trained based on a third sample set, wherein the third sample set includes a plurality of third samples and labels corresponding to the plurality of third samples, a third sample is a first communication signal and a channel parameter, and the label corresponding to the third sample is a third communication signal corresponding to the first communication signal.

[0100] It should be understood that the second communication signal can be understood as an estimated value of the transmitted data, while the third communication signal is an actual value of the transmitted data, and thus can be used as a label for model training.

[0101] In some embodiments, the third sample set can be determined based on relevant data generated in a simulation platform for a communication system. The simulation platform can simulate the communication process under various communication environments, that is, it can obtain simulation data under a communication channel based on multiple channel characteristics. Thus, the received data, that is, the first communication signal and the channel parameters can be extracted from the simulation data as the third sample. And the transmitted data, that is, the second communication signal, can be extracted from the simulation data as a label corresponding to the third sample.

[0102] In some embodiments, multiple sets of third samples and labels may be generated under different fading channels.

[0103] In some embodiments, the first communication signal and the channel parameters may be processed into real numbers and data may be rearranged to obtain a third sample.

[0104] For example, the sender can obtain the Figure 4 The processing in generates the modulation data X at the transmitting end Data , and the data Y of the data symbol received by the receiving end through the wireless fading channel Data , and obtain the channel H through which the data symbol passes Data Then we can use the data Y Data and channel H Data Perform real number processing and data rearrangement based on the processing subband or processing granularity of the channel equalization submodel to obtain N*X real number arrays. A real number array may include [Re(Y Data ), Im(Y Data ), Re(H Data ), Im(H Data )]. The processing subband of the channel equalization submodel is X REs, that is, Y PW , rearranged into N sub-band data. And, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part. Data X PW and channel H Data The corresponding N*X real number array is a third sample.

[0105] In addition, the data X can be sent Data Perform real number processing and data rearrangement to obtain N*X real number arrays, among which the jth real number array can be expressed as X′ Data,j =[Re(X Data ), Im(X Data )]. The N*X real number array is the label of the third sample. Thus, the channel equalization sub-model can be trained based on the obtained third sample and the label corresponding to the third sample.

[0106] In some embodiments, the third sample set may also be determined based on actual communication data in various communication systems. For example, the training samples and labels may be determined based on the received data of the receiving end, the sent data of the sending end, and the channel parameters in the historical communication data actually transmitted in the communication system to train the model.

[0107] It should be understood that the above-mentioned method for determining the third sample set is only an example, and the third sample set may also be determined in other possible ways, which is not limited in the present disclosure.

[0108] In addition, the process of training the channel equalization sub-model according to the third sample set can refer to the training process of the pre-demodulation sub-model in the above steps S1-S5, which will not be repeated here.

[0109] Furthermore, after the channel equalization sub-model is trained, the channel equalization sub-model can be deployed at the receiving end in the communication system for online signal detection, wherein the detected data also needs to be processed similar to the third sample so that the processed data can be input into the channel equalization sub-model to obtain a second communication signal.

[0110] It should be noted that the technical solution provided by the present invention adopts a receiver (first model) with a neural network structure that can be driven by both data and model, and can use linearly calculated data as training data input into the first model, thereby reducing the sample demand and training time required.

[0111] In some embodiments, the receiving end may perform cyclic prefix (CP) removal and Fourier transform (FFT) processing on the first communication signal, and then process the first communication signal through the first model to obtain the second communication signal.

[0112] For example, Figure 4 As shown, the sender can encode the bit stream to be sent to obtain X bit Then, X bit Modulation processing is performed to obtain X Date , and then perform layer mapping processing and inverse Fourier transform (IFFT) processing. In some embodiments, it may also include introducing a cyclic prefix CP, also known as adding a CP process. That is, after the time domain OFDM signal to be transmitted is obtained through IFFT processing, the CP corresponding to the subcarrier spacing corresponding to the time domain OFDM signal may be added, so that when sending the data frame, the transmitter may send the time domain OFDM signal and the CP.

[0113] It should be noted that the transmitter introduces a cyclic prefix between each subcarrier to prevent subcarrier interference (ICI).

[0114] Correspondingly, after the transmitted data passes through the wireless channel, the receiving end corresponding to the transmitting end can receive the received data. The receiving end can first perform CP removal and FFT processing on the received data (i.e., the first communication signal), and then process the first communication signal through the first model. In addition, the receiving end can also perform mediation processing and decoding processing on the output data of the first model to obtain a received bit stream.

[0115] S103: Perform signal analysis processing on the second communication signal to obtain communication data.

[0116] In some embodiments, the signal analysis process includes complex processing, demodulation processing, and decoding processing.

[0117] Exemplarily, due to obtaining the second communication signal is a real number array, and the second communication signal can also be processed into a complex number to obtain a processed second communication signal

[0118] It can also be expressed as

[0119] Perform complex processing to obtain the processed second communication signal

[0120] Among them, complex() represents the processing process of forming a complex value data.

[0121] In some embodiments, complex value data can also be The subsequent mediation and decoding processing of this model will not be described in detail here.

[0122] Based on the technical solution provided by the present disclosure, all symbols can be used to transmit data during the communication process, thereby improving the data transmission efficiency. In addition, the technical solution adopts a low-speed receiver (first model) architecture based on a cascaded neural network, based on which the received communication signal without pilot symbols can be processed, which can more accurately reflect the channel impact and further improve the receiver signal detection performance.

[0123] In addition, the technical solution provided by the present invention adopts a receiver (first model) with a neural network structure that can be driven by both data and model, and can use linearly calculated data as training data input into the first model, thereby reducing the sample demand and training time required.

[0124] In some embodiments, Figure 6 As shown, the receiving end processes the first symbol in the first communication signal based on the pre-demodulation sub-model to obtain the second symbol, which may include the following steps:

[0125] S201. Determine a first real number array according to a first symbol.

[0126] It should be noted that the receiving end can construct the input data of the pre-demodulation sub-model, such as the pre-demodulation neural network Net1, based on the first symbol. Since the structure of the pre-demodulation sub-model is usually a real neural network, the receiving end can first perform real number processing on the first symbol to obtain a first real number array, so that the first real number array can be input into the pre-demodulation sub-model for subsequent data processing.

[0127] In some embodiments, a first initial real number array may be determined according to the first symbol, and then a second real number array may be determined based on the first initial real number array.

[0128] In some embodiments, the elements in the first initial array of real numbers are determined based on a value of the real part of the first symbol and a value of the imaginary part of the first symbol.

[0129] Exemplarily, the receiving end can extract the first symbol Y from the received first communication signal (frequency domain signal) Y PW , and Y PW The real part and the imaginary part of are separated to form a set of real number arrays, that is, the first initial real number array corresponding to the first symbol. For example, the first initial real number array can be expressed as [Re(Y PW ), Im(Y PW )]. Wherein, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part.

[0130] Furthermore, the size of the first subband, the number of the first subbands and the number of elements of the first initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain the first real number array.

[0131] The size of the first subband is determined based on the processing granularity of the pre-demodulation sub-model.

[0132] For example, the processing granularity of the neural network model (pre-demodulation sub-model) can be defined as X REs as a processing sub-band, and the data pilot symbols received on all antennas are The data is rearranged based on the processing sub-band as the granularity, and the received data is rearranged into N sub-bands. For example, the first real number array Net is obtained. 1,in =[N×X×2] ka .

[0133] Example 1: In a low-speed scenario, in a pilot-free communication system, a single-input single-output (SISO) technology is used for transmission, with T = 1 physical transmitting antenna and R = 1 physical receiving antenna. The frequency domain data Y received by the receiving end occupies 24 RBs in the frequency domain, a total of 288 REs. And it occupies 14 OFDM symbols in the time domain, and the data dimension of Y is 14*288. The first model processing granularity X = 72 is 6RB, that is, 72 REs. The data pilot symbol adopts QPSK modulation, Q m =2.

[0134] Thus, the data pilot symbol Y can be extracted from the received frequency domain signal Y PW , which is the first symbol mentioned above, has a dimension of 288*1. Since the neural network composed of the pre-demodulation sub-model is a real neural network, it is necessary to convert Y PW The real and imaginary parts of are separated to form a set of real number arrays, namely the first initial real number array mentioned above, for example [Re(Y PW ), Im(Y PW )], whose dimension is 288*2. Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part.

[0135] Then, the first real number array can be determined based on the first initial real number array. When the processing granularity X of the neural network model of the pre-demodulation sub-model is 72 and the frequency domain data is divided into N=4 sub-bands, each sub-band There are 6 RBs and 72 REs in total.

[0136] The dimension of Y is 72*1, PW It only has the value on R = 1 receiving antenna. Therefore, Y after rearrangement according to the subband PW A set of real number arrays, namely the first real number array, can be formed, such as the following Net 1,in , whose dimension is 4*72*2.

[0137]

[0138] In addition, the training sample of the pre-demodulation sub-model provided in the above embodiment, ie, the first sample, can be determined based on the bit stream data to be modulated by the data pilot symbol of the transmitting end and the received data pilot symbol of the receiving end.

[0139] Exemplarily, the bit stream data X to be modulated by the data pilot symbol of the transmitting end bit , whose dimension is 288*Q m , Q m=2 is the modulation order, after passing through the wireless fading channel, the received data pilot symbol data Y is obtained at the receiving end PW , the dimension is 288*1. Then for Y PW Perform the real number processing and data reorganization processing in step S201 to obtain 4*72 real number arrays. Among them, a single real number array is composed of [Re(Y PW ), Im(Y PW )], which can be used as the first sample of the pre-demodulation sub-model. In addition, the bit stream data X to be modulated by the data pilot symbol of the transmitting end bit , real number processing and data reorganization can also be performed to obtain 4*72 real number arrays, where the jth real number array can be Among them, the modulation order is Q m =2, this 4*72*Q m An array of real numbers X′ pw,bit It can be used as the label of the first sample. Then, the pre-demodulation sub-model can be trained based on the obtained first sample and the label.

[0140] S202. Input the first real number array into the pre-demodulation sub-model to obtain a second real number array.

[0141] The receiving end can construct the input data of the pre-demodulation sub-model based on the first symbol extracted from the received data, that is, the first real number array, and then input the first real number array into the pre-demodulation sub-model to obtain the second real number array.

[0142] For example, the receiving end may convert the first real number array Net obtained in step 201 into 1,in Input the neural network Net1 of the pre-demodulation sub-model, and after online detection of the neural network Net1 of the pre-demodulation sub-model, the output of the neural network Net1 can be obtained. That is, the second real number array.

[0143] In some examples, taking Example 1 above as an example, the first real number array Net 1,in Input into the neural network Net1 of the pre-demodulation sub-model to obtain the output of the neural network Net1 It is a 4*72 real number array, of which the j-th real number array is as follows:

[0144] For example, the following Net 1,Out , which can also be called the second real number array.

[0145]

[0146] S203. Perform modulation processing on the second real number array to obtain a second symbol.

[0147] In some embodiments, the second array of real numbers may be modulated based on the modulation order of the data pilot symbol to obtain a second symbol.

[0148] For example, Figure 7 As shown, the receiving end can output the second real number array of the neural network Net1 of the pre-demodulation sub-model Taking the communication scenario in Example 1 above as an example, based on the modulation order Q m =2 for modulation processing to obtain the second symbol, that is, the estimated value of the data pilot signal Its dimensions are 4*72*2.

[0149] Based on the above embodiments, the communication signal of the received data pilot symbol can be processed to improve the accuracy of determining the channel influence, thereby improving the signal detection performance of the receiver.

[0150] In some embodiments, Figure 8 As shown, the receiving end processes the first symbol and the second symbol based on the channel estimation sub-model to obtain the channel estimation parameter, which may include the following steps:

[0151] S301. Determine a third real number array based on the first symbol and the second symbol.

[0152] It should be noted that the receiving end can construct the input data of the channel estimation sub-model, such as the channel estimation neural network Net2, based on the first symbol and the second symbol. Since the structure of the channel estimation sub-model is usually a real neural network, the receiving end can first perform real number processing on the first symbol and the second symbol to obtain a third real number array, so that the third real number array can be input into the channel estimation sub-model for subsequent data processing.

[0153] In some embodiments, a third initial real number array may be determined based on the first symbol and the second symbol.

[0154] The elements in the third initial real number array are determined based on the value of the real part of the first symbol, the value of the imaginary part of the first symbol, the value of the real part of the second symbol, and the value of the imaginary part of the second symbol.

[0155] Exemplarily, the receiving end can extract the first symbol Y from the received first communication signal (frequency domain signal) Y PW , and the second symbol A set of real number arrays can be formed, that is, the third initial real number array Among them, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part.

[0156] Then, the size of the second subband, the number of the second subbands, and the number of elements of the third initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain a third real number array. In some embodiments, the size of the second subband is determined based on the processing granularity of the channel estimation submodel.

[0157] For example, the processing granularity of the neural network model (channel estimation submodel) can be defined as X REs as a processing subband, and the data pilot symbols received on all antennas are The data is rearranged based on the processing sub-band as the granularity, and the received data is rearranged into N sub-bands. For example, the third real number array Net is obtained. 2,in =[N×X×4] ka .

[0158] In one example, the scenario in Example 1 above is taken as an example, that is, in a low-speed scenario without a pilot communication system, SISO technology is used for transmission, with T = 1 physical transmitting antenna, R = 1 physical receiving antenna, etc. The data pilot symbol Y can be extracted from the received frequency domain signal Y PW , that is, the first symbol mentioned above, whose dimension is 288*1, and the second symbol Then we get the third initial real number array, for example Then, the third real number array can be determined based on the third initial real number array. When the processing granularity X of the neural network model constituting the channel estimation sub-model is 72 and the frequency domain data is divided into N=4 sub-bands, the data pilot signals received by all antennas can be Data is rearranged based on the processing sub-band granularity, and the data is received by N = 4 sub-bands Net 2,in , which is the third real number array, has a dimension of 4*72*2. and The dimensions are all 72*1.

[0159]

[0160] In addition, the training samples of the channel estimation sub-model provided in the above embodiment, ie, the second samples, can be determined based on the bit stream data to be modulated by the data pilot symbols of the transmitting end, the received data pilot symbols of the receiving end, and the channel parameters.

[0161] Exemplarily, the transmitting end data pilot symbol modulates data X PW , whose dimension is 288*1, after passing through the wireless fading channel, the received data pilot symbol data Y at the receiving end is obtained PW , the dimension is 288*1. Then for the data X PW And data YPW After the real number processing and data reorganization, a 4*72 real number array is obtained. Among them, a single real number array is composed of [Re(Y PW ), Im(Y PW ), Re(X PW ), Im(X PW )], which can be used as the training of the channel estimation sub-model, i.e., the second sample. In addition, the channel parameter H can also be obtained PW Its dimension is 288*1. After the real number processing and data reorganization processing, a 4*72 real number array is obtained, of which the jth real number array can be H′ pw,j =[Re(H PW ), Im(H PW )]. It can be used as the label of the second sample. Then, the channel estimation sub-model can be trained based on the obtained second sample and the label.

[0162] Example 2: In a low-speed scenario, a pilot-free communication system is used for transmission using MIMO technology, with L = 2 layers of data streams, P = 2 antenna ports and T = 2 physical transmitting antennas, and R = 2 physical receiving antennas. The frequency domain signal Y received by the receiving end can be:

[0163] Y=HX+N, that is

[0164] Where X indicates the modulated transmit signal of two data streams, and its dimension is L×1, i.e. 2×1. H is the wireless channel matrix of the transmit and receive channels, and its dimension is T×R, i.e. 2×2. N is the additive white Gaussian noise, and its dimension is R×1, i.e. 2×1. Frequency domain received data Y Data The dimension of is also R×1, that is, 2×1.

[0165] The received signal Y occupies 12 RBs in the frequency domain, a total of 144 REs, and 4 OFDM symbols in the time domain, of which the data pilot Y PW The first OFDM symbol in the time domain, the pilots of different streams occupy different REs in the frequency domain, so the data starts from the second OFDM in the time domain and lasts for 3 OFDM symbols. The data of different streams occupy the same RE in the frequency domain, such as Fig. 9 As shown. The processing granularity of the neural network model is X=36. In some embodiments, the received frequency domain data has been normalized, including data symbol layer 1 / 2, data pilot symbol layer 1 and data pilot symbol layer 2.

[0166] The data pilot symbol Y can be extracted from the received frequency domain signal Y PW , that is, the first symbol mentioned above, whose dimension is 288*1, and the second symbol The data pilot includes the data pilot of the first antenna port of the first receiving antenna (R=1, P=1), the data pilot of the first antenna port of the second receiving antenna (R=2, P=1), the data pilot of the second antenna port of the first receiving antenna (R=1, P=2), and the data pilot of the second antenna port of the second receiving antenna (R=2, P=2). PW and the predicted second symbol To obtain a third initial real number array, such as the following real number array, whose dimension is 4*72*4.

[0167]

[0168] Then, the third real number array can be determined based on the third initial real number array. When the processing granularity X of the neural network model constituting the channel estimation sub-model is 36 and the frequency domain data is divided into N=2 sub-bands, the data pilot signals received by all antennas can be Data is rearranged based on the processing sub-band granularity, and the data is received by N = 2 sub-bands Net 2,in , which is the third real number array, has a dimension of 8*36*4. The dimension is 36*1.

[0169]

[0170]

[0171] In addition, the training samples of the channel estimation sub-model provided in the above embodiment, ie, the second samples, can be determined based on the bit stream data to be modulated by the data pilot symbols of the transmitting end, the received data pilot symbols of the receiving end, and the channel parameters.

[0172] Exemplarily, the transmitting end data pilot symbol modulates the data Its dimension is 2*72. After passing through the wireless fading channel, the received data pilot symbol data is obtained at the receiving end. The dimension is 4*72. Then for the data X PW And data Y PW Perform real number processing and data reorganization processing, N=2, and the dimension is 8*36*4.

[0173] For example

[0174]

[0175] And, get the channel parameters Its dimension is 4*72. After real number processing and data reorganization, N=2, we get H′ pw . Thus, the 8*36*2 real number array H′pw As the training sample, that is, the label of the second sample. Then, the channel estimation sub-model can be trained based on the obtained second sample and the label.

[0176]

[0177] S302: Input the third real number array into the channel estimation sub-model to obtain channel estimation parameters.

[0178] For example, Figure 7 As shown, the receiving end can use the third real number array Net 2,in Input the channel estimation sub-model Net2, and you can get the output data of the channel estimation sub-model Net2, that is, the channel estimation parameters

[0179] Taking the communication scenario of Example 1 above as an example, further, the receiving end may convert the third real number array Net obtained in step 301 above into 2,in The neural network Net2 of the channel estimation sub-model is input, and after the online detection of the neural network Net2 of the channel estimation sub-model, the output of the neural network Net2 can be obtained, such as the following Net2 Out , which is the channel estimation parameter

[0180]

[0181] Alternatively, taking the communication scenario of Example 2 above as an example, the receiving end may convert the third real number array Net obtained in step 301 above into 2,in The neural network Net2 of the channel estimation sub-model is input, and the output of the neural network Net2 can be obtained through online detection of the neural network Net2 of the channel estimation sub-model, such as the following Net 2,Out ,

[0182]

[0183] That is, the channel estimation parameter Its dimension is 8*72*2, including 2 sub-bands, 2 receiving antennas, and 2 antenna ports.

[0184] Based on the above embodiments, channel estimation can be performed based on received data pilot symbols and predicted data pilot symbol data without transmitting pilot symbols, which can reduce transmission overhead and improve receiver signal detection performance.

[0185] In some embodiments, Fig.10 As shown, the receiving end processes the first communication signal and the channel estimation parameter based on the channel equalization sub-model to obtain the second communication signal, which may include the following steps:

[0186] S401. Perform preprocessing based on the first communication signal and the channel estimation parameter to determine a fourth real number array.

[0187] It should be noted that the receiving end can construct the input data of the channel equalization sub-model, such as the channel equalization neural network Net3, based on the first communication signal and the channel estimation parameters determined above. Since the structure of the channel equalization sub-model is usually a real neural network, the receiving end can first perform real number processing on the first communication signal and the channel estimation parameters to obtain a fourth real number array, so that the fourth real number array can be input into the channel equalization sub-model for subsequent data processing.

[0188] In some embodiments, a fourth initial array of real numbers may be determined based on the first communication signal and the channel estimation parameters.

[0189] The elements in the fourth initial real number array are determined based on the value of the real part of the first communication signal, the value of the imaginary part of the first communication signal, the value of the real part of the channel estimation parameter, and the value of the imaginary part of the channel estimation parameter.

[0190] For example, the receiving end may receive data Y based on Data and channel estimation parameters Construct a set of real number arrays, that is, the fourth initial real number array Among them, Re() represents the operation of taking the real part, and Im() represents the operation of taking the imaginary part.

[0191] Then, the size of the third subband, the number of the third subbands, and the number of elements of the fourth initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first communication signal is rearranged based on the dimensions to obtain a fourth real number array. In some embodiments, the size of the third subband is determined based on the processing granularity of the channel equalization submodel.

[0192] For example, the processing granularity of the neural network model (channel estimation submodel) can be defined as X REs as a processing subband, and the data symbols received on all antennas are The data is rearranged based on the processing sub-band as the granularity, and the received data is rearranged into N sub-bands. For example, the fourth real number array Net is obtained. 3,in =[N×X×4] ka .

[0193] In one example, the scenario in Example 1 above is taken as an example, that is, in a low-speed scenario without a pilot communication system, SISO technology is used for transmission, with T = 1 physical transmitting antenna, R = 1 physical receiving antenna, etc., which will not be repeated. The data symbol Y can be extracted from the received frequency domain signal Y Data = {YData,1 ,Y Data,2 ,…Y Data,13}, a total of 13 symbols. Each symbol has 288 REs, and the dimension is 13*288.

[0194] The data Y of each symbol can be extracted based on the symbol order Data,sym ,sym=1,2,…13, the data Y of the current symbol Data,sym And the channel estimation parameters Construct the fourth initial real number array, such as When the processing granularity X of the neural network model of the channel equalization sub-model is 72 and the frequency domain data is divided into N=4 sub-bands, the data symbols Y received by all antennas can be Data Data is rearranged based on the processing sub-band granularity, and the data is received by N = 4 sub-bands Net 3,in , which is the fourth real number array, its dimension is 4*72*2.

[0195]

[0196] In addition, the training sample of the channel equalization sub-model provided in the above embodiment, ie, the third sample, can be determined based on the data symbols received by the receiving end and the channel estimation parameters.

[0197] Exemplarily, the transmitting end modulates the data X Data , whose dimension is 288*1, after passing through the wireless fading channel, the received data symbol data Y at the receiving end is obtained Data , the dimension is 288*1, the ideal channel H Data Then, for data X Data And data Y Data After the real number processing and data reorganization, a 4*72 real number array is obtained. Among them, a single real number array is composed of [Re(Y Data ), Im(Y Data ), Re(H Data ), Im(H Data ], which can be used as the training of the channel equalization sub-model, i.e., the third sample. In addition, the modulation data X at the transmitting end can also be obtained. Data , whose dimension is 288*1, after the real number processing and data reorganization, a 4*72 real number array is obtained, where the jth real number array can be X′ Data,j =[Re(X data ), Im(X data )]. It can be used as a label of the third sample. Then, the channel equalization sub-model can be trained based on the obtained third sample and the label.

[0198] S402: Input the fourth real number array into the channel equalization sub-model to obtain a second communication signal.

[0199] The second communication signal may be understood as an estimated value of the transmitted data.

[0200] For example, Figure 7 As shown, the receiving end can use the fourth real number array Net 3,in Input the channel equalization sub-model Net3 to obtain the output data of the channel equalization sub-model Net3, that is, the second communication signal

[0201] Taking the communication scenario of Example 1 above as an example, further, the receiving end may convert the fourth real number array Net obtained in step 401 above into 3,in The neural network Net3 of the channel equalization sub-model is input, and the output of the neural network Net3 can be obtained through online detection of the neural network Net3 of the channel equalization sub-model, such as the following Net 3,Out , that is, the second communication signal corresponding to the data symbol

[0202]

[0203] In some embodiments, the above processing can be performed on each symbol to obtain the estimated value of the transmitted data obtained by calculating the data of 13 symbols, that is, the second communication signal

[0204] In some embodiments, due to the second communication signal is a real number array, and the second communication signal can also be processed into a complex number to obtain a processed second communication signal

[0205] It can also be expressed as

[0206] Perform complex processing to obtain the processed second communication signal

[0207] In some embodiments, complex value data can also be The subsequent modulation and decoding processing will not be described in detail here.

[0208] Based on the above embodiment, the estimated value of the transmitted data can be determined through the trained model based on the received data symbols and the channel estimation parameters, which can improve the signal detection performance of the receiver.

[0209] The above mainly introduces the solution provided by the present disclosure from the perspective of interaction between various nodes. It is understandable that each node, such as a device or equipment, includes a hardware structure and / or software module corresponding to the execution of each function in order to realize the above functions. It should be easily appreciated by those skilled in the art that, in combination with the algorithm steps of each example described in the embodiments disclosed herein, the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0210] The embodiments of the present disclosure may divide the functional modules of the communication device according to the above method embodiments. For example, each functional module may be divided corresponding to each function, or two or more functions may be integrated into one functional module. The above integrated modules may be implemented in the form of hardware or software. It should be noted that the division of modules in the embodiments of the present disclosure is schematic and is only a logical function division. There may be other division methods in actual implementation. The following is an example of dividing each functional module corresponding to each function.

[0211] Fig.11 FIG. 1 is a schematic diagram showing the composition of a communication device provided by an embodiment of the present disclosure. Fig.11 As shown, the communication device 110 includes a receiving module 1101 and a processing module 1102 .

[0212] In some embodiments, the receiving module 1101 may be used to receive a first communication signal without a pilot symbol. The processing module 1102 is used to process the first communication signal through a first model to obtain a second communication signal, and perform signal analysis processing on the second communication signal to obtain communication data.

[0213] In some embodiments, the first model includes a pre-demodulation sub-model, a channel estimation sub-model, and a channel equalization sub-model.

[0214] In some embodiments, the sub-model of the first model is constructed based on a convolutional neural network, a residual neural network, or a residual neural network combined with an attention mechanism.

[0215] In some embodiments, the processing module 1102 is specifically configured to process the first symbol in the first communication signal based on the pre-demodulation submodel to obtain the second symbol, process the first symbol and the second symbol based on the channel estimation submodel to obtain the channel estimation parameter, and process the first communication signal and the channel estimation parameter based on the channel equalization submodel to obtain the second communication signal.

[0216] In some embodiments, the first symbol is used to carry data and perform channel estimation.

[0217] In some embodiments, the first symbol is the first received symbol in the first communication signal.

[0218] In some embodiments, the processing module 1102 is specifically configured to: determine a first real number array based on the first symbol, input the first real number array into a pre-demodulation sub-model to obtain a second real number array, and perform modulation processing on the second real number array to obtain a second symbol.

[0219] In some embodiments, the processing module 1102 is specifically configured to: determine a first initial real number array according to the first symbol; the elements in the first initial real number array are determined based on the value of the real part of the first symbol and the value of the imaginary part of the first symbol. The size of the first subband, the number of the first subbands, and the number of elements in the first initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain the first real number array.

[0220] In some embodiments, the first subband size is determined based on a processing granularity of the pre-demodulation submodel.

[0221] In some embodiments, the processing module 1102 is specifically configured to: determine a third real number array based on the first symbol and the second symbol, and input the third real number array into the channel estimation sub-model to obtain channel estimation parameters.

[0222] In some embodiments, the processing module 1102 is specifically configured to: determine a third initial real number array according to the first symbol and the second symbol; the elements in the third initial real number array are determined based on the value of the real part of the first symbol, the value of the imaginary part of the first symbol, the value of the real part of the second symbol, and the value of the imaginary part of the second symbol. The size of the second subband, the number of the second subbands, and the number of elements in the third initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain the third real number array.

[0223] In some embodiments, the second subband size is determined based on a processing granularity of the channel estimation submodel.

[0224] In some embodiments, the processing module 1102 is specifically configured to: pre-process the first communication signal and the channel estimation parameter to obtain a fourth real number array, and input the fourth real number array into the channel equalization sub-model to obtain a second communication signal.

[0225] In some embodiments, the processing module 1102 is specifically used to: determine a fourth initial real number array according to the first communication signal and the channel estimation parameter, wherein the elements in the fourth initial real number array are determined based on the value of the real part of the first communication signal, the value of the imaginary part of the first communication signal, the value of the real part of the channel estimation parameter, and the value of the imaginary part of the channel estimation parameter. The size of the third subband, the number of the third subband, and the number of elements of the fourth initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first communication signal is rearranged based on the dimensions to obtain a fourth real number array.

[0226] In some embodiments, the third subband size is determined based on a processing granularity of the channel equalization sub-model.

[0227] In some embodiments, the pre-demodulation sub-model is trained based on the first sample set. The first sample set includes multiple first samples and labels corresponding to the multiple first samples, one first sample is a first symbol, and the label corresponding to the first sample is a third symbol corresponding to the first symbol.

[0228] In some embodiments, the channel estimation submodel is trained based on the second sample set. The second sample set includes multiple second samples and labels corresponding to the multiple second samples, a second sample is a first symbol and a third symbol corresponding to the first symbol, and the label corresponding to the second sample is a measured channel parameter.

[0229] In some embodiments, when the measured channel parameters include channel parameters in multiple antenna dimensions, the channel parameters in multiple antenna dimensions are arranged based on a preset order of the channel estimation submodel, and the channel estimation parameters output by the channel estimation submodel are also arranged based on a preset order.

[0230] In some embodiments, the channel estimation submodel is trained based on a third sample set; the third sample set includes multiple third samples and labels corresponding to the multiple third samples, one third sample is a first communication signal and a channel estimation parameter output by the channel estimation submodel, and the label corresponding to the third sample is a third communication signal corresponding to the first communication signal.

[0231] In some embodiments, the signal analysis process includes complex processing, demodulation processing, and decoding processing.

[0232] For a more detailed description of the above-mentioned receiving module 1101 and processing module 1102, a more detailed description of each technical feature therein, and a description of the beneficial effects, etc., please refer to the above-mentioned corresponding method embodiment part, which will not be repeated here.

[0233] It should be noted that Fig.11 The modules in the can also be called units, for example, the processing module can be called a processing unit. Fig.11In the illustrated embodiment, the names of the modules may not be the names shown in the figure. For example, the receiving module may also be called a communication module.

[0234] Fig.11 If each unit in the embodiment is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present disclosure is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present disclosure. The storage medium for storing computer software products includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0235] In the case of implementing the functions of the above-mentioned integrated modules in the form of hardware, the embodiment of the present disclosure provides a structural schematic diagram of a communication device. Fig.12 As shown, the communication device 120 includes: a processor 1202 , a communication interface 1203 , and a bus 1204 . Optionally, the communication device 120 may also include a memory 1201 .

[0236] The processor 1202 may be a processor that implements or executes various exemplary logic blocks, modules, and circuits described in conjunction with the present disclosure. The processor 1202 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the present disclosure. The processor 1202 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0237] The communication interface 1203 is used to connect with other devices via a communication network, such as Ethernet, wireless access network, wireless local area network (WLAN), etc.

[0238] The memory 1201 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a disk storage medium or other magnetic storage device, or 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.

[0239] As a possible implementation, the memory 1201 may exist independently of the processor 1202, and the memory 1201 may be connected to the processor 1202 via a bus 1204 to store instructions or program codes. When the processor 1202 calls and executes the instructions or program codes stored in the memory 1201, the method provided in the embodiment of the present disclosure can be implemented.

[0240] In another possible implementation, the memory 1201 may also be integrated with the processor 1202 .

[0241] The bus 1204 may be an extended industry standard architecture (EISA) bus, etc. The bus 1204 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.12 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0242] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment or device is divided into different functional modules to complete all or part of the functions described above.

[0243] The embodiment of the present disclosure also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be completed by computer instructions to instruct the relevant hardware, and the program can be stored in the above computer-readable storage medium. When the program is executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be the memory or memory of any of the above embodiments. The above computer-readable storage medium can also be an external storage device of the above device or apparatus, such as a plug-in hard disk, a smart memory card (smart media card, SMC), a secure digital (secure digital, SD) card, a flash card (flash card), etc. equipped on the above device or apparatus. Further, the above computer-readable storage medium can also include both the internal storage unit of the above device or apparatus and an external storage device. The above computer-readable storage medium is used to store the above computer program and other programs and data required by the above device or apparatus. The above computer-readable storage medium can also be used to temporarily store data that has been output or is to be output.

[0244] The embodiments of the present disclosure also provide a computer program product, which includes a computer program. When the computer program product is run on a computer, the computer is enabled to execute any one of the methods provided in the above embodiments.

[0245] Although the present disclosure is described herein in conjunction with various embodiments, in the process of implementing the present disclosure claimed for protection, those skilled in the art may understand and implement other variations of the disclosed embodiments by reviewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps.

[0246] "A" or "an" does not exclude a plurality of cases. A single processor or other unit may implement several functions listed in the claims. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

[0247] Although the present disclosure has been described in conjunction with specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present disclosure. Accordingly, this specification and the drawings are merely exemplary illustrations of the present disclosure as defined by the appended claims, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present disclosure. Obviously, those skilled in the art may make various modifications and variations to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.

[0248] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A communication method, characterized in that: Applied to the receiving end, the method includes: receiving a first communication signal without a pilot symbol; Processing the first communication signal through a first model to obtain a second communication signal; Perform signal analysis processing on the second communication signal to obtain communication data.

2. The method according to claim 1, characterized in that The first model includes a pre-demodulation sub-model, a channel estimation sub-model and a channel equalization sub-model.

3. The method according to claim 1, characterized in that The sub-model of the first model is constructed based on a convolutional neural network, a residual neural network, or a residual neural network combined with an attention mechanism.

4. The method according to claim 2, characterized in that: The processing of the first communication signal by using the first model includes: Processing a first symbol in the first communication signal based on the pre-demodulation sub-model to obtain a second symbol; Processing the first symbol and the second symbol based on the channel estimation submodel to obtain a channel estimation parameter; The first communication signal and the channel estimation parameter are processed based on the channel equalization sub-model to obtain the second communication signal.

5. The method according to claim 4, characterized in that The first symbol is used to carry data and perform channel estimation.

6. The method according to claim 4, characterized in that The processing of the first symbol in the first communication signal based on the pre-demodulation sub-model includes: Based on the first symbol, determining a first real number array; Inputting the first real number array into the pre-demodulation sub-model to obtain a second real number array; Modulation processing is performed on the second real number array to obtain the second symbol.

7. The method according to claim 6, characterized in that The step of determining a first real number array based on the first symbol comprises: Determine a first initial real number array according to the first symbol; the elements in the first initial real number array are determined based on the value of the real part of the first symbol and the value of the imaginary part of the first symbol; The size of the first subband, the number of the first subbands, and the number of elements of the first initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain the first real number array.

8. The method according to claim 7, characterized in that The size of the first sub-band is determined based on the size of the processing sub-band of the pre-demodulation sub-model.

9. The method according to claim 4, characterized in that The processing of the first symbol and the second symbol based on the channel estimation submodel includes: Determine a third array of real numbers based on the first symbol and the second symbol; The third real number array is input into the channel estimation sub-model to obtain the channel estimation parameters.

10. The method according to claim 9, characterized in that The step of determining a third real number array based on the first symbol and the second symbol comprises: Determine a third initial real number array according to the first symbol and the second symbol; the elements in the third initial real number array are determined based on the value of the real part of the first symbol, the value of the imaginary part of the first symbol, the value of the real part of the second symbol, and the value of the imaginary part of the second symbol; The size of the second subband, the number of second subbands and the number of elements of the third initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first symbol is rearranged based on the dimensions to obtain the third real number array.

11. The method according to claim 10, characterized in that The size of the second subband is determined based on the size of the processing subband of the channel estimation submodel.

12. The method according to claim 4, characterized in that The processing of the first communication signal and the channel estimation parameter based on the channel equalization submodel includes: Preprocessing the first communication signal and the channel estimation parameter to obtain a fourth real number array; The fourth real number array is input into the channel equalization sub-model to obtain the second communication signal.

13. The method according to claim 12, characterized in that The preprocessing of the first communication signal and the channel estimation parameter comprises: Determine a fourth initial real number array according to the first communication signal and the channel estimation parameter; the elements in the fourth initial real number array are determined based on the value of the real part of the first communication signal, the value of the imaginary part of the first communication signal, the value of the real part of the channel estimation parameter, and the value of the imaginary part of the channel estimation parameter; The size of the third subband, the number of third subbands and the number of elements of the fourth initial real number array are respectively used as the dimensions of the three dimensions of the real number array, and the data carried in the first communication signal is rearranged based on the dimensions to obtain the fourth real number array.

14. The method according to claim 13, characterized in that The size of the third sub-band is determined based on the size of the processing sub-band of the channel equalization sub-model.

15. The method according to claim 2, characterized in that The pre-demodulation sub-model is trained based on a first sample set; the first sample set includes multiple first samples and labels corresponding to the multiple first samples, one first sample is a first symbol, and the label corresponding to the first sample is a third symbol corresponding to the first symbol.

16. The method according to claim 2, characterized in that The channel estimation submodel is trained based on the second sample set; the second sample set includes multiple second samples and labels corresponding to the multiple second samples, one second sample is a first symbol and a third symbol corresponding to the first symbol, and the label corresponding to the second sample is a measured channel parameter.

17. The method according to claim 16, characterized in that In the case where the measured channel parameters include channel parameters in multiple antenna dimensions, the channel parameters in multiple antenna dimensions are arranged based on a preset order of the channel estimation sub-model, and the channel estimation parameters output by the channel estimation sub-model are arranged based on the preset order.

18. The method according to claim 2, characterized in that The channel estimation submodel is trained based on a third sample set; the third sample set includes multiple third samples and labels corresponding to each of the multiple third samples, one of the third samples is the first communication signal and the channel estimation parameter output by the channel estimation submodel, and the label corresponding to the third sample is a third communication signal corresponding to the first communication signal.

19. The method according to claim 1, characterized in that The signal analysis process includes complex processing, demodulation processing and decoding processing.

20. A communication device, characterized in that: include: Memory and processor; Memory and processor coupling; The memory is used to store instructions executable by the processor; When the processor executes the instructions, the method according to any one of claims 1 to 19 is performed.

21. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a communication device, the communication device is caused to execute the method according to any one of claims 1 to 19.

Images