Signal detection method and device, and storage medium

Abstract

The present application discloses a signal detection method, an apparatus, and a storage medium, in which the signal detection method includes the steps of: obtaining at least one set of historical signal samples, including a historical channel estimate, a historical received signal, and a historical modulated signal; and, when the current channel estimate and the received signal to be measured are obtained, inputting the current channel estimate and the received signal to be measured into a pre-trained preset neural network to perform signal detection and obtain a signal estimate of the received signal to be measured, where the pre-trained preset neural network is obtained by training the preset neural network through at least one set of historical signal samples.

Description

[Technical field]

[0001] This application is filed based on and claims priority to a Chinese patent application bearing application number 202210831551.7 and filing date July 15, 2022, the entire contents of which are hereby incorporated by reference into this application.

[0002] The present application relates to the field of communication technology, and in particular to a signal detection method, a signal detection device, a computer storage medium and a computer program product. [Background technology]

[0003] Multiple-in Multiple-out (MIMO) space division multiplexing technology and Orthogonal Frequency Division Multiplexing (OFDM) technology are very competitive technologies in wireless communication systems. MIMO space division multiplexing technology allows the system to easily obtain spatial diversity gain and channel capacity gain when the system bandwidth and transmission bandwidth do not change, and OFDM technology transmits data in parallel using multiple orthogonal subcarriers, which greatly reduces the data rate of each path, and has strong resistance to multipath interference and frequency selective fading because a time guard interval is added.

[0004] Signal detection is one of the important technologies in OFDM systems and MIMO systems. The signal detection methods of related technologies mainly include three aspects: linear detection, nonlinear detection, and optimal detection. All of them can realize the signal detection process, but the detection performance is low and cannot meet the ever-increasing signal detection needs. For example, optimal detection algorithms such as maximum likelihood detection algorithms need to globally search all possible transmission symbol regions for the received signal, so they are highly complex. Linear detection algorithms include zero forcing algorithms and minimum mean square error detection algorithms, which have low computational complexity but low detection accuracy. Nonlinear detection algorithms include spherical decoding algorithms and successive interference cancellation algorithms, which increase the detection complexity to improve the detection accuracy. Summary of the Invention [Problem to be solved by the invention]

[0005] SUMMARY OF THE DISCLOSURE The embodiments of the present application provide a signal detection method, a signal detection apparatus, a computer storage medium and a computer program product to improve signal detection performance. [Means for solving the problem]

[0006] According to a first aspect, an embodiment of the present application includes a method for obtaining at least one set of historical signal samples, the at least one set including a historical channel estimate, a historical received signal, and a historical modulated signal; When a current channel estimate and a measurement target received signal are obtained, the current channel estimate and the measurement target received signal are input to a pre-trained preset neural network to perform signal detection, thereby obtaining a signal estimate of the measurement target received signal; The pre-trained preset neural network provides a signal detection method obtained by training the preset neural network through at least one set of the historical signal samples.

[0007] According to a second aspect, an embodiment of the present application further provides a signal detection device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, the computer program realizing the aforementioned signal detection method when the processor executes the computer program.

[0008] According to a third aspect, embodiments of the present application further provide a computer-readable storage medium storing computer-executable instructions for performing the aforementioned signal detection method.

[0009] According to a fourth aspect, an embodiment of the present application further provides a computer program product, wherein a computer program or the computer instructions are stored in a computer readable storage medium, and a processor of a computing device reads the computer program or the computer instructions from the computer readable storage medium, and the processor executes the computer program or the computer instructions to cause the computing device to perform the aforementioned signal detection method.

[0010] In this embodiment, based on the acquired historical signal samples including the historical channel estimation value, the historical received signal, and the historical modulation signal, a preset neural network is trained to obtain a neural network model that meets the requirements, so that instead of the detection model in the related art, a pre-trained preset neural network model is used to perform signal detection, and the channel estimation value and the received signal are simultaneously input into the preset neural network for training or detection. This adopts a data model dual driving method, which allows the preset neural network to more easily learn channel-related features, not only improves the signal detection performance, but also significantly reduces the amount of information required for training or detection, has good environmental adaptability and generalization, and can meet the ever-increasing needs of signal detection, and fills the technical gap in related methods. [Brief description of the drawings]

[0011] [Figure 1]2 is a flowchart of a signal detection method according to an embodiment of the present application. [Diagram 2] 2 is a flow chart of obtaining at least one set of historical signal samples in a signal detection method according to an embodiment of the present application; [Diagram 3] 4 is a flow chart of a signal detection method according to an embodiment of the present application, in which a current channel estimation value and a received signal to be measured are input to a pre-trained preset neural network before signal detection is performed. [Figure 4] 4 is a flowchart of obtaining a pre-trained preset neural network by training a preset neural network in a signal detection method according to an embodiment of the present application; [Diagram 5] 4 is a flowchart of obtaining a first real sequence, which is a training sample, in a signal detection method according to an embodiment of the present application. [Figure 6] 4 is a flowchart of obtaining a second real sequence, which is a label of a training sample, in a signal detection method according to an embodiment of the present application. [Figure 7] 4 is a flowchart of a signal detection method according to an embodiment of the present application after a signal estimate of a historical received signal is obtained; [Figure 8] 4 is a flowchart of obtaining a signal estimate of a received signal to be measured in a signal detection method according to an embodiment of the present application; [Figure 9] 11 is a flowchart illustrating a process for obtaining a third real number sequence in a signal detection method according to another embodiment of the present application. [Figure 10] 4 is a flowchart after a real signal estimation result of a received signal to be measured is obtained in the signal detection method according to an embodiment of the present application; [Figure 11] 4 is a flowchart illustrating a process for obtaining a complex signal estimation result of a received signal to be measured in a signal detection method according to an embodiment of the present application. [Figure 12] 2 is a flowchart illustrating an implementation of a signal detection method according to an embodiment of the present application. [Figure 13]FIG. 1 is a principle schematic diagram of a MIMO system based on a preset neural network according to an embodiment of the present application. [Figure 14] 11 is a flow chart of a signal detection method according to another embodiment of the present disclosure, in which a current channel estimation value and a received signal to be measured are input to a pre-trained preset neural network before signal detection is performed. [Figure 15] 4 is a flowchart of obtaining a signal estimate of a received signal to be measured in a signal detection method according to another embodiment of the present application; [Figure 16] FIG. 1 is a principle schematic diagram of an OFDM system based on a preset neural network according to an embodiment of the present application; [Figure 17] 1 is a schematic diagram of a signal detection device according to an embodiment of the present application; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described in more detail below with reference to the drawings and examples. It should be understood that the specific examples described in this specification are only for the purpose of interpreting the present application, and are not intended to limit the present application.

[0013] It should be noted that although a logical order is depicted in the flowcharts, in some cases the steps shown or described may be performed in an order different from that depicted in the flowcharts. The terms "first," "second," and the like in the specification, claims, and drawings are intended to distinguish between similar objects and are not necessarily intended to describe a particular order or chronology.

[0014] In recent years, artificial intelligence (AI) technology, especially deep learning, has achieved great success in many fields such as computer vision, natural language processing, and voice recognition. Intelligent wireless communication realized by AI technology is considered to be one of the mainstream directions of future 6G development, and its basic idea is to greatly improve the performance of wireless communication systems through the organic integration of wireless communication technology and AI technology. Physical layer AI design includes two mainstream methods: end-to-end communication link design based on AI technology and communication module algorithm design based on AI technology. At present, the idea of ​​communication module algorithm design based on AI technology is mainly based on data-driven, that is, regarding some or some communication modules as unknown black boxes, and instead uses deep learning networks, which are mainly obtained by relying on large amounts of data training, which has a heavy training burden, takes time and effort, and has a relatively single application scenario and is not very applicable, and cannot meet the current signal detection needs.

[0015] Based on this, the present application provides a signal detection method, a signal detection device, a computer storage medium and a computer program product. The signal detection method of an embodiment includes the steps of: obtaining at least one set of historical signal samples, including a historical channel estimate, a historical received signal and a historical modulated signal; when the current channel estimate and the measured received signal are obtained, inputting the current channel estimate and the measured received signal into a pre-trained preset neural network to perform signal detection and obtain a signal estimate of the measured received signal, where the pre-trained preset neural network is obtained by training the preset neural network through at least one set of historical signal samples. In this embodiment, based on the acquired historical signal samples including the historical channel estimation value, the historical received signal, and the historical modulation signal, a preset neural network is trained to obtain a neural network model that meets the requirements, so that instead of the detection model in the related art, a pre-trained preset neural network model is used to perform signal detection, and the channel estimation value and the received signal are simultaneously input into the preset neural network for training or detection. This data model dual driving method allows the preset neural network to more easily learn channel-related features, not only improving the signal detection performance, but also significantly reducing the amount of information required for training or detection, and having good environmental adaptability and generalization, which can meet the ever-increasing needs of signal detection, and fill the technical gap in related methods.

[0016] Hereinafter, the embodiments of the present application will be further described with reference to the drawings.

[0017] As shown in FIG. 1, FIG. 1 is a flow chart of a signal detection method according to an embodiment of the present application, the signal detection method including: obtaining at least one set of historical signal samples including a historical channel estimate, a historical received signal, and a historical modulated signal; Step S120 includes, but is not limited to, when the current channel estimation value and the received signal to be measured are obtained, inputting the current channel estimation value and the received signal to be measured into a pre-trained preset neural network to perform signal detection and obtain a signal estimation value of the received signal to be measured, where the pre-trained preset neural network is obtained by training the preset neural network through at least one set of historical signal samples.

[0018] In this step, based on the acquired historical signal samples including the historical channel estimation value, the historical received signal, and the historical modulation signal, a preset neural network is trained to obtain a neural network model that meets the requirements, so that instead of the detection model in the related art, a pre-trained preset neural network model is used to perform signal detection, and the channel estimation value and the received signal are simultaneously input into the preset neural network for training or detection. This data model dual driving method allows the preset neural network to more easily learn channel-related features, not only improving the signal detection performance, but also significantly reducing the amount of information required for training or detection, and having good environmental adaptability and generalization, which can meet the ever-increasing needs of signal detection, and fill the technical gap in related methods.

[0019] It can be seen from this that, compared with the data-driven deep learning network of the related technology, the embodiments of the present application are based on data model dual-driving, that is, based on the conventional technology of wireless communication systems, without changing the structure of the wireless communication system, a deep learning network is used instead of the related detection module to improve the performance of the model; compared with the data-driven deep learning network that mainly relies on large amounts of data training, the data model dual-driving deep learning network in the embodiments of the present application only relies on the communication model or algorithm model, and is based on the existing model of the physical layer, which can significantly reduce the amount of information required for training, upgrading or detection, has good environmental adaptability and generalization, and has broad development prospects.

[0020] In one embodiment, the embodiment of the present application can be applied to OFDM technology or MIMO technology, but is not limited thereto, for example, applied to an OFDM receiver based on MIMO technology, that is, the preset neural network in the embodiment of the present application can be integrated into an OFDM receiver, and in the OFDM receiver, the preset neural network can be trained and, after receiving relevant information, a signal can be detected based on the pre-trained preset neural network, the basic principle of which is consistent with the embodiment of the present application, and detailed description is omitted, and those skilled in the art can also select a manner of applying the signal detection method of the embodiment of the present application according to a specific application scenario, and are not limited thereto, for example, with the further development of network systems, it is understood that multiple application scenarios that can be applied to the signal detection method of the embodiment of the present application are not excluded.

[0021] In one embodiment, the preset neural network includes: Deep Neural Networks, Convolutional Neural Networks, Residual convolutional neural network, and residual convolutional neural network with attention mechanism.

[0022] The examples of preset neural networks shown above may be those that undergo offline training and online deployment, may be those that undergo online deployment after completing unified offline training, or may be those that complete unified online training and deployment, and are not limited thereto. In addition to the examples of preset neural networks shown above, those skilled in the art may also consider setting an appropriate preset neural network according to factors such as the characteristics and needs of a specific application scenario, but detailed descriptions will not be omitted here.

[0023] In one embodiment, the historical channel estimate and the historical received signal are distinguished from the current channel estimate and the measured received signal, that is, the role of the historical channel estimate and the historical received signal is used to train the preset neural network, and the current channel estimate and the measured received signal are signal parameters detected by the preset neural network, but as the training process continues, the current channel estimate and the measured received signal may also become new historical channel estimates and historical received signals, that is, the training of the preset neural network can be dynamic, and has a certain transferability in time, so that more signal samples can be used to train the preset neural network to achieve better training effect, and the training process of the preset neural network will be gradually described in detail in the following embodiments.

[0024] In one embodiment, after training the preset neural network through at least one set of historical signal samples to obtain a pre-trained preset neural network, the historical channel estimation value and the historical received signal are input as a set of detection data into the preset neural network to perform re-detection, and the actual training effect of the preset neural network is judged according to the detection result, which is advantageous for further judging whether the entire training process of the preset neural network may cause an error.

[0025] In one embodiment, the historical channel estimate, the historical received signal, the current channel estimate and the measured received signal may be obtained by sequentially transmitting a set of signals, or may be obtained by individually transmitting each corresponding signal, and is not limited thereto.

[0026] In one embodiment, the specific number of the history signal samples can be set according to a specific application scenario, that is, the training samples of the preset neural network can be set according to a specific application scenario, and is not limited herein.

[0027] As shown in FIG. 2, in one embodiment of the present application, when there are multiple sets of historical signal samples, and each wireless fading channel corresponds to multiple sets of historical signal samples, step S110 includes: Step S111 of acquiring a historical modulation signal transmitted by a transmitting side; Step S112 of performing predetermined signal processing on the historical modulated signal in each wireless fading channel to obtain a received pilot signal, a local pilot signal, and a historical received signal at the receiving side; and step S113 of multiplying the received pilot signal with the conjugate of the local pilot signal to obtain a historical channel estimate.

[0028] In this step, there are multiple sets of historical signal samples, and each wireless fading channel corresponds to multiple sets of historical signal samples, so by obtaining the historical modulation signal sent by the transmitting side, in each wireless fading channel, perform predetermined signal processing on the historical modulation signal respectively, to obtain a received pilot signal, a local pilot signal and a historical received signal at the receiving side, and then multiply the received pilot signal by the conjugate of the local pilot signal to obtain a historical channel estimation value, that is, due to the change of the channel, the amplitude of the received signal will change randomly, and signal fading will occur, so that each wireless fading channel can correspondingly obtain at least one complete set of different historical signal samples, in this way, a plurality of historical signal samples can be generated in total for training the preset neural network, and the training effect can be greatly guaranteed.

[0029] In one embodiment, the sender and the receiver are relative concepts as a set, i.e., the sender corresponds to the initial source of the signal, and the receiver corresponds to the back-end result of the signal, and the specific representation form of the sender and the receiver is not limited here, but is merely for clarifying the operating principle of the embodiment of the present application, and specific senders and receivers can be set in specific application scenarios, and are not limited here.

[0030] In one embodiment, there may be multiple ways to obtain the history adjustment signal sent by the transmitting side, but are limited here. For example, by obtaining at least one set of transmission bit stream signals sent by the transmitting side, the at least one set of transmission bit stream signals can be further coded and modulated to obtain a history modulation signal, that is, the history modulation signal can be obtained by pre-processing the initial transmission bit stream signal, or the signal to be processed can be pre-adjusted by the transmitting side to a history modulation signal, so that the history modulation signal can be directly obtained from the transmitting side, etc.

[0031] In one embodiment, the specific method of multiplying the received pilot signal by the conjugate of the local pilot signal to obtain the historical channel estimation value may be multiple, but is not limited here. For example, the least squares LS channel estimation algorithm can be used to multiply the received pilot signal by the conjugate of the local pilot signal to obtain the historical channel estimation value, but is not limited here, or other conventional communication algorithms commonly used in the art can be used, but are not limited here, and the least squares method is a mathematical model applied in data processing fields such as error estimation, uncertainty, system identification and prediction and forecast, and is well known to those skilled in the art, so detailed description thereof is omitted here.

[0032] In one embodiment, there may be multiple methods for performing predetermined signal processing on the historical modulation signal, but not limited thereto. For example, the fast Fourier transform can be performed on the historical modulation signal, and a first output signal can be obtained after adding a cyclic prefix, and then the first output signal can be passed through different wireless fading channels to obtain different second output signals, and the second output signal can be subjected to cyclic prefix removal and inverse Fourier transform to realize signal restoration, and finally obtain a received pilot signal, a local pilot signal, and a historical reception signal; for example, the historical modulation signal can be input into a predetermined network model, and the historical modulation signal can be subjected to demodulation and equalization processing through the predetermined network model, and finally obtain a received pilot signal, a local pilot signal, and a historical reception signal.

[0033] As shown in FIG. 3, in one embodiment of the present application, if there are multiple sets of historical signal samples, before step S120, The method further includes, but is not limited to, a step S130 of obtaining a pre-trained preset neural network by training the preset neural network using the historical channel estimate value and the historical received signal as training samples and the historical modulated signal as the label of the training sample for each set of historical signal samples.

[0034] In this step, the historical channel estimation value and the historical received signal are used as training samples, so that the preset neural network is easy to learn the characteristics of different wireless fading channels when it is trained, thereby obtaining a better training effect, and realizing a complete training flow without using other training samples, thereby reducing the parameters and time required for training, and the historical modulation signal associated with the historical channel estimation value and the historical received signal are used as the label of the training sample, so as to improve the stability of the preset neural network training.

[0035] The principles of operation of the above-mentioned embodiments will now be described with reference to specific examples.

[0036] Example 1 First, we need to determine the preset neural network. In a MIMO system with N receive antennas, assume that the received signal is represented as Y.

number

number

number

number

number

number

number

[0037] Next, the LS channel estimate

number

number

number

[0038] As shown in FIG. 4, in one embodiment of the present application, step S130 is further described, which includes: Step S131: combine the obtained historical channel estimate with the real and imaginary parts of the historical received signal to obtain a first real array, which is a training sample; Step S132 of combining the real and imaginary parts of the acquired hysteretic modulation signal to obtain a second real array, which is a label of the training sample; and S133, inputting the first real sequence and the second real sequence into a preset neural network for training to obtain a signal estimate of the historical received signal.

[0039] In this step, the historical channel estimation value is combined with the real part and the imaginary part of the historical received signal to obtain a first real sequence, which is the training sample, and the real part and the imaginary part of the acquired historical modulation signal are combined to obtain a second real sequence, which is the label of the training sample, that is, the input forms of the historical channel estimation value, the historical received signal and the historical modulation signal are unified and set to a form that meets the input requirements of the preset neural network, so that the preset neural network can easily train based on the real sequence to obtain a signal estimation value of the historical received signal; and in certain application scenarios, the values ​​of the corresponding real part and imaginary part can respectively represent the values ​​of the antenna port and the receiving antenna, that is, the form of the real sequence can be better adapted to the operation of the neural network.

[0040] In one embodiment, the preset neural network may be of other types in addition to the above-mentioned real neural network, and accordingly, the input to the preset neural network may vary according to the type of the preset neural network; that is, steps S131 to S133 in the above embodiment do not limit the contents input to the preset neural network, and furthermore, appropriate training input contents can be selected according to a specific scenario, but are not limited here.

[0041] As shown in FIG. 5, in one embodiment of the present application, step S131 is further described, which includes: a step S1311 of obtaining a first real part and a first imaginary part of the historical channel estimate and obtaining a second real part and a second imaginary part of the historical received signal; and step S1312 of arranging the first real part, the first imaginary part, the second real part, and the second imaginary part in a preset order to obtain a first real number sequence, which is a training sample, but is not limited to this.

[0042] In this step, a first real part and a first imaginary part of the historical channel estimation value are obtained, and a second real part and a second imaginary part of the historical received signal are obtained, so that a real number set corresponding to the historical channel estimation value and the historical received signal can be generated, and a first real number array including the historical channel estimation value and the historical received signal can be obtained by arranging the corresponding first real part, first imaginary part, second real part and second imaginary part in each real number set in a preset order.

[0043] In one embodiment, the pre-set arrangement order is not limited, and may be arranged based on the real part and the imaginary part corresponding to the historical channel estimate and the historical received signal, respectively, that is, including, but not limited to, arranging the real part and the imaginary part of the historical channel estimate in adjacent positions, arranging the real part and the imaginary part of the historical received signal in adjacent positions, or arranging in order according to a specific application scenario, etc.

[0044] As shown in FIG. 6, in one embodiment of the present application, when the historical modulation signal includes a modulation signal stream including multiple hierarchical modulation signals, step S132 is further described, and step S132 includes: Step S1321 of acquiring real and imaginary parts of each hierarchical modulation signal; and step S1322 of arranging the real and imaginary parts of each hierarchical modulation signal in a preset order to obtain a second real number array, which is a label of the training sample.

[0045] In this step, when the hysteretic modulation signal includes a modulated signal stream including multiple hierarchical modulation signals, the hysteretic modulation signal at this time corresponds to one signal layer, and obtaining the real part and imaginary part of each hierarchical modulation signal in the hysteretic modulation signal corresponds to obtaining the real part and imaginary part of the entire hysteretic modulation signal. Next, the real part and imaginary part of each hierarchical modulation signal are arranged in a predetermined order to obtain a second real number array, which is the label of the training sample.

[0046] In one embodiment, the pre-set arrangement order is not limited, and can be arranged based on the real and imaginary parts of each hierarchical modulation signal, i.e., arranging the real and imaginary parts of each hierarchical modulation signal in adjacent positions, and arranging the real and imaginary parts of adjacent hierarchical modulation signals in adjacent positions, etc., or arranging in order according to a specific application scenario, etc., but is not limited to this.

[0047] As shown in FIG. 7, in one embodiment of the present application, after step S133, A step S134 of obtaining a current value of a preset loss function corresponding to the preset neural network based on the signal estimate value of the historical received signal and the second real number array; The method further includes, but is not limited to, a step S135 of determining a pre-trained preset neural network when the current value of the pre-set loss function meets a pre-set convergence condition, and steps S134 to S135 are used to determine the pre-trained preset neural network.

[0048] In this step, a loss function is calculated from the signal estimation value obtained by training and the predetermined second real number array, and it is judged according to the calculated value whether it meets the preset convergence condition. If it is judged that the calculated value does not meet the preset convergence condition, the preset neural network is further trained and optimized to obtain a new signal estimation value. By analogy, a new loss function value is obtained by calculating a new signal estimation value, and judgment is made again, so that the preset neural network can be continuously optimized, and the final trained signal estimation value is continuously close to the transmission signal, and can flexibly support data bit sources of different lengths and different modulation methods, so that the loss function can be more reliably applied to the correlation scenario in which it is calculated, and a preset neural network that meets the requirements can finally be obtained by training.

[0049] In one embodiment, the preset loss function is: Mean Square Error (MSE) loss function, Root Mean Squared Error (RMSE) loss function, Mean Absolute Error (MAE) loss function, Euclidean distance loss function, Cosine distance loss function, and at least one of a linear weighting function of an MSE loss function, an RMSE loss function, an MAE loss function, a Euclidean distance loss function, and a cosine distance loss function.

[0050] The above loss function is well known to those skilled in the art, so a detailed description thereof is omitted here. Alternatively, those skilled in the art can set corresponding pre-set loss functions for the pre-set neural network according to specific application scenarios, but are not limited here.

[0051] The principles of operation of the above-mentioned embodiments will now be described with reference to specific examples.

[0052] Example 2 First, a training input of a preset neural network is constructed based on a MIMO system, and the preset neural network is a real-valued neural network, so that:

number

number

number

number

number

[0053] Then, the label of the training sample of the preset neural network is further generated, and model training is performed. Referring to FIG. 12, after the transmission bit stream sent by the transmitting side is encoded and modulated, the transmission modulated signal including the multi-layer MIMO signal is obtained.

number

number

number

number

number

number

number

[0054] Where:

number

number

number

number

number

[0055] As shown in FIG. 8, in one embodiment of the present application, step S120 is further described, which includes: Step S121 of combining the obtained current channel estimate with the real and imaginary parts of the received signal to be measured to obtain a third real array; and S122, inputting the third real sequence into a pre-trained neural network to perform signal detection and obtain a real signal estimation result of the received signal to be measured, but not limited thereto.

[0056] In this step, the current channel estimation value and the real part and the imaginary part of the received signal of the object to be measured are combined to obtain a third real sequence as a detection input, so as to provide corresponding input parameters that meet the requirements for the preset neural network; that is, the input forms of the current channel estimation value and the received signal of the object to be measured are unified and set to a form that meets the input requirements of the preset neural network, so that the preset neural network can detect and obtain a real signal estimation result of the received signal of the object to be measured based on the real sequence.

[0057] In one embodiment, the preset neural network may output other types of real signal estimation results in addition to the above output, and accordingly, the input to the preset neural network may vary depending on the output type of the preset neural network, and is not limited here.

[0058] As shown in FIG. 9, in one embodiment of the present application, step S121 is further described, which includes: a step S1211 of obtaining a third real part and a third imaginary part of the current channel estimate and obtaining a fourth real part and a fourth imaginary part of the received signal to be measured; and step S1212 of arranging the third real part, the third imaginary part, the fourth real part, and the fourth imaginary part in a preset order to obtain a third real number array, but this is not limited to this.

[0059] In this step, the third real part and the third imaginary part of the current channel estimation value are obtained, and the fourth real part and the fourth imaginary part of the received signal to be measured are obtained, so that a real number set corresponding to the current channel estimation value and the received signal to be measured can be generated, and further, the corresponding third real part, third imaginary part, fourth real part and fourth imaginary part in each real number set are arranged in a preset order to obtain a third real number array including the current channel estimation value and the received signal to be measured.

[0060] In one embodiment, the pre-set arrangement order is not limited, and may be arranged based on the real part and the imaginary part corresponding to the current channel estimate and the received signal to be measured, respectively, i.e., including, but not limited to, arranging the real part and the imaginary part of the current channel estimate in adjacent positions, arranging the real part and the imaginary part of the received signal to be measured in adjacent positions, or arranging in order according to a specific application scenario, etc.

[0061] As shown in FIG. 10, in one embodiment of the present application, the step embodiment after step S122 is further described, and after step S122, Step S123 of performing a complex transformation process on the real signal estimation result of the received signal under measurement to obtain a complex signal estimation result of the received signal under measurement; and step S124 of demodulating and decoding the complex signal estimation result to obtain an output signal, but this is not limited to this.

[0062] In this step, a complex signal estimation result of the received signal under measurement is obtained from the real signal estimation result of the received signal under measurement, and the complex signal estimation result can be further demodulated and decoded to obtain a final output signal.

[0063] As shown in FIG. 11, in an embodiment of the present application, when the real signal estimation result includes a real signal estimation array including real parts and imaginary parts of multiple hierarchical signals arranged in order, step S123 is further described, and step S123 includes the following steps: Step S1231 of complex-transforming the real and imaginary parts of each hierarchical signal in the real signal estimation array to obtain complex-transformed data corresponding to each hierarchical signal; and step S1232 of arranging each complex transform data in a preset order to obtain a complex signal estimation result of the received signal to be measured, but this is not limited to this.

[0064] In this step, the real signal estimation result includes a real signal estimation array including real and imaginary parts of a plurality of hierarchical signals arranged in a preset order, so that the real and imaginary parts of each hierarchical signal in the real signal estimation array can be complex transformed to obtain complex transformed data corresponding to each hierarchical signal, and by further arranging each complex transformed data, a complex signal estimation result of the signal to be measured can be obtained accurately and reliably.

[0065] The principles of operation of the above-mentioned embodiments will now be described with reference to specific examples.

[0066] Example 3 As shown in Figure 12, a preset neural network can be used to perform online signal detection, and after the training of the preset neural network is completed, it can be placed at the receiving end of the MIMO system to perform online signal detection. During online signal detection, the input parameters of the preset neural network can be similarly obtained by referring to the methods of Example 1 and Example 2, and the preset neural network performs online detection, and the output of the preset neural network, i.e., the estimated value of the transmission signal, can be obtained, and the conversion from the real number sequence to the complex value data after equalization can be completed as follows:

number

[0067] Here, complex(·) represents the operation that creates one complex value, and the complex value obtained by the conversion

number

[0068] A complete process example is given below to explain the working principle of each of the above embodiments.

[0069] Example 4 As shown in Figure 13, in an OFDM system based on MIMO technology, a MIMO signal detector and an LS channel estimation module based on neural network (i.e., preset neural network in the embodiment of the present application) are provided, where L=2 data layers, P=2 antenna ports, T=2 physical transmitting antennas, and R=2 physical receiving antennas. The receiving side frequency domain signal Y can be expressed by Equation (2).

number

[0070] where X is the transmitted signal modulated by the two data layers, with dimension L×1, i.e., 2×1, H is the wireless channel matrix between the transmitter and receiver, with dimension T×R, i.e., 2×2, and N is the additive white Gaussian noise, with dimension R×1, i.e., 2×1. Frequency-domain received signal

number

number

number

[0071] The neural network-based MIMO signal detection method proposed in this example focuses on how to detect the transmitted signal X from the frequency domain signal Y on the receiving side, and the specific steps are as follows.

[0072] Step 1: Determine the structure of a MIMO signal detector based on a neural network.

number

number

number

number

number

number

[0073] Step 2: Construct the training inputs for the convolutional neural network-based MIMO signal detector. In a similar manner to Example 1 above,

number

number

number

number

[0074] Step 3: After generating the labels of the training samples for the MIMO signal detector, model training is performed.

number

number

number

number

[0075] In addition, the modulated signal

number

number

[0076] Where:

number

number

number

number

number

number

[0077] Here, n is the number of samples in the batch size for each training. When training the model, the Adam optimizer is used to train the CNN based on the MSE loss function, and the network training is determined to be complete after the loss function has converged to a certain extent. At this time, the model parameters are saved and used for subsequent online signal detection.

[0078] Step 4: Perform online MIMO signal detection. After completing CNN training, it can be deployed at the receiving side of the OFDM system to perform online signal detection. When detecting online signals, refer to the processing process of step 1 and step 2, similarly obtain the input parameters of CNN, and then perform online detection, and obtain the output of CNN, that is, the estimated value of the transmitted signal, and the estimated value of the transmitted signal includes the corresponding real array, and then complete the conversion from the real array to the equalized complex value data as follows:

number

[0079] Here, complex(·) represents the operation that creates one complex value, and the complex value obtained by the conversion

number

[0080] As can be seen from this, the signal detection method of the embodiments of the present application outputs data after equalization and layer demapping, and by using the modulated signal from the transmitting side as the training label and MSE as the loss function, the estimated value continuously approaches the transmitted signal, and can flexibly support data bit streams of different lengths and different modulation methods, thereby providing excellent versatility.

[0081] As shown in FIG. 14, in one embodiment of the present application, when there are multiple sets of history signal samples and the preset neural network includes an enhanced channel estimation sub-network and a channel equalization sub-network, before step S120: Step S140 of inputting the historical channel estimates for each set of historical signal samples into an enhanced channel estimation sub-network to perform channel enhanced estimation training to obtain a first estimation result; The method further includes, but is not limited to, a step S150 of inputting the first estimation result and the historical received signal into a channel equalization sub-network to perform channel equalization training to obtain a pre-trained preset neural network.

[0082] In this step, the preset neural network includes two sub-networks, namely, an enhanced channel estimation sub-network and a channel equalization sub-network, and each sub-network has its own independent performance. Therefore, for each set of historical signal samples, the historical channel estimation value is input to the enhanced channel estimation sub-network to perform channel enhanced estimation training to obtain a first estimation result, and then the first prediction result and the historical received signal are input to the channel equalization sub-network to perform channel equalization training to obtain a pre-trained preset neural network, which can achieve a similar effective and reliable training effect compared with the preset neural network with integrated functions.

[0083] In one embodiment, the enhanced channel estimation sub-network and the channel equalization sub-network may be trained together or separately, or a person skilled in the art may select a corresponding training method according to a specific application scenario to train the enhanced channel estimation sub-network, the channel equalization sub-network, etc., but is not limited thereto.

[0084] As shown in FIG. 15, in one embodiment of the present application, step S120 includes: Step S125: inputting the current channel estimate to a pre-trained enhanced channel estimation sub-network to perform enhanced channel estimation and obtain a second estimation result; and S126, inputting the second estimation result and the received signal to be measured into a pre-trained channel equalization sub-network to perform channel equalization and obtain a signal estimate of the received signal to be measured, but this is not limited to this.

[0085] In this step, when signal detection is required, the current channel estimation value is input to a pre-trained enhanced channel estimation sub-network to perform channel enhanced estimation, thereby directly obtaining a second estimation result; and the second estimation result and the received signal to be measured are input to a pre-trained channel equalization sub-network to perform channel equalization, thereby obtaining a signal estimation value of the received signal to be measured. The channel estimation result and the final signal estimation value are respectively tested and calculated, which is advantageous to improving the signal detection accuracy of the preset neural network.

[0086] In addition, the operating principles of steps S140-S150 and steps S125-S126 are similar to those of the related steps in the above-mentioned embodiments. The only difference is that the preset neural network in the above-mentioned embodiments has an integrated structure, while the preset neural network in the present application is divided into two sub-networks, and the function of one preset neural network can be fully realized through the cooperation of the two sub-networks. Since the detailed principles and processes of the preset neural network have been described in detail in the above-mentioned embodiments, the step embodiments of this embodiment may refer to the detailed principles and processes of the preset neural network in the above-mentioned embodiments, and will not be described in detail here to avoid redundancy.

[0087] Another complete process example is given below to illustrate the working principles of each of the above embodiments.

[0088] Example 5 As shown in Figure 16, in an OFDM system based on MIMO technology, a MIMO signal detector and an LS channel estimation module based on neural network (i.e., preset neural network in the embodiment of the present application) are provided, where L=4 data layers, P=4 antenna ports, T=4 physical transmitting antennas, and R=4 physical receiving antennas. The receiving side frequency domain signal Y can be expressed by Equation (4).

number

[0089] where X is the transmitted signal modulated by these four data layers, with dimension L×1, i.e., 4×1, H is the wireless channel matrix between the transmitter and the receiver, with dimension T×R, i.e., 4×4, and N is the additive white Gaussian noise, with dimension R×1, i.e., 4×1. Frequency-domain received signal

number

number

number

[0090] The neural network-based MIMO signal detection method proposed in this example focuses on how to detect the transmitted signal X from the frequency domain signal Y on the receiving side, and the specific steps are as follows.

[0091] Step 1: Determine the structure of a MIMO signal detector based on a neural network.

number

number

number

number

number

number

[0092] Step 2: Construct the training input for the enhanced channel estimation sub-network, in a similar manner to Example 1 above.

number

number

number

number

number

number

[0093] Then, construct the training input for the channel equalization sub-network;

number

number

number

[0094] Where:

number

number

[0095] Step 3: After generating the labels of the training samples for the MIMO signal detector, model training is performed.

number

number

number

number

[0096] In addition, the modulated signal

number

number

[0097] Where:

number

number

number

number

number

number

[0098] Here, n is the number of samples in the batch size for each training. When training the model, the Adam optimizer is used to train the residual neural network based on the Euclidean distance loss function, and the network training is determined to be complete after the loss function has converged to a certain extent. At this time, the model parameters are saved and used for subsequent online signal detection.

[0099] Step 4: Perform online MIMO signal detection, and after completing residual neural network training, it can be deployed at the receiving side of OFDM system to perform online signal detection. When detecting online signal, also refer to the processing process of step 1 and step 2, similarly obtain the input parameters of residual neural network, and then perform online detection, and obtain the output of residual neural network, that is, the estimated value of transmission signal, and the estimated value of transmission signal contains the corresponding real number array, and the conversion from real number array to equalized complex value data is completed as follows:

number

[0100] Here, complex(·) represents the operation that creates one complex value, and the complex value obtained by the conversion

number

[0101] As can be seen from this, the signal detection method of the embodiments of the present application outputs data after equalization and layer demapping, and by using the modulated signal from the transmitting side as the training label and Euclidean distance as the loss function, the estimated value continuously approaches the transmitted signal, and can flexibly support data bit streams of different lengths and different modulation methods, providing excellent versatility.

[0102] Moreover, as shown in FIG. 17, one embodiment of the present application further discloses a signal detection device 100 including at least one processor 110 and at least one memory 120 for storing at least one program, and which, when the at least one program is executed by the at least one processor 110, realizes the signal detection method in any of the above-mentioned embodiments.

[0103] Moreover, an embodiment of the present application further discloses a computer-readable storage medium storing computer-executable instructions for performing the signal detection method in any of the above embodiments.

[0104] Furthermore, one embodiment of the present application further discloses a computer program product including a computer program or computer instructions stored in a computer-readable storage medium, wherein a processor of a computing device reads the computer program or computer instructions from the computer-readable storage medium, and the processor executes the computer program or computer instructions, thereby causing the computing device to perform the signal detection method in the previous embodiment.

[0105] Those skilled in the art can understand that all or some steps in the methods and systems disclosed above can be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As is well known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (e.g., computer readable instructions, data structures, program modules, or other data). Computer storage media may include, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk memory, magnetic cartridges, magnetic tape, magnetic disk memory or other magnetic storage devices, or any other medium that can be used to store the desired information and accessible by a computer.Those skilled in the art will also recognize that communication media typically include computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier or other transport mechanism, and may include any information delivery media.

Claims

1. 1. A method for detecting a signal, comprising: obtaining at least one set of historical signal samples including a historical channel estimate, a historical received signal, and a historical modulated signal; When a current channel estimate and a measurement target received signal are obtained, the current channel estimate and the measurement target received signal are input to a pre-trained preset neural network to perform signal detection, thereby obtaining a signal estimate of the measurement target received signal; The pre-trained preset neural network is obtained by training the preset neural network through at least one set of the historical signal samples. Signal detection methods.

2. There are a plurality of sets of historical signal samples, each wireless fading channel corresponding to a plurality of sets of the historical signal samples; The step of obtaining at least one set of historical signal samples comprises: obtaining a historical modulated signal transmitted by a transmitter; performing predetermined signal processing on the historical modulated signal in each of the wireless fading channels to obtain a received pilot signal, a local pilot signal, and a historical received signal at a receiving side; multiplying the received pilot signal by a conjugate of the local pilot signal to obtain a historical channel estimate; Includes The signal detection method according to claim 1 .

3. Obtaining a pre-trained preset neural network by training the preset neural network with at least one set of the historical signal samples based on the following steps: For each set of the historical signal samples, the historical channel estimate and the historical received signal are taken as training samples, and the historical modulated signal is taken as a label of the training samples to train the preset neural network, thereby obtaining the pre-trained preset neural network. The signal detection method according to claim 2 .

4. When the preset neural network includes an enhanced channel estimation sub-network and a channel equalization sub-network, obtain a pre-trained preset neural network by training the preset neural network with at least one set of the historical signal samples according to the following steps: For each set of the historical signal samples, input the historical channel estimates into the enhanced channel estimation sub-network to perform channel enhanced estimation training to obtain a first estimation result; The first estimation result and the historical received signal are input to the channel equalization sub-network to perform channel equalization training, thereby obtaining the pre-trained preset neural network. The signal detection method according to claim 2 .

5. the step of inputting the current channel estimate and the received signal to be measured into a pre-trained neural network to perform signal detection and obtain a signal estimate of the received signal to be measured, performing enhanced channel estimation by inputting the current channel estimate to the pre-trained enhanced channel estimation sub-network to obtain a second estimation result; inputting the second estimation result and the received signal to be measured into the pre-trained channel equalization sub-network to perform channel equalization and obtain a signal estimate of the received signal to be measured; Includes The signal detection method according to claim 4.

6. the step of training the preset neural network using the historical channel estimates and the historical received signal as training samples and the historical modulated signal as labels of the training samples, combining the obtained historical channel estimate with a real part and an imaginary part of the historical received signal to obtain a first real array, which is a training sample; combining the real and imaginary parts of the acquired historical modulation signal to obtain a second real array, which is a label of the training sample; inputting the first real number sequence and the second real number sequence into the preset neural network for training to obtain a signal estimate of the historical received signal; Includes The signal detection method according to claim 3 .

7. The step of combining the obtained historical channel estimate with a real part and an imaginary part of the historical received signal to obtain a first real number array, which is a training sample, includes: obtaining a first real part and a first imaginary part of the historical channel estimate and obtaining a second real part and a second imaginary part of the historical received signal; A step of arranging the first real part, the first imaginary part, the second real part, and the second imaginary part in a preset order to obtain a first real number array which is a training sample; Includes The signal detection method according to claim 6.

8. the historical modulation signal includes a modulation signal layer including a plurality of hierarchical modulation signals; The step of combining the real part and the imaginary part of the acquired historical modulation signal to obtain a second real number array which is a label of the training sample, obtaining a real part and an imaginary part of each of the hierarchical modulation signals; arranging the real and imaginary parts of each of the hierarchical modulation signals in a predetermined order to obtain a second real number array which is a label of the training sample; Includes The signal detection method according to claim 6.

9. After obtaining a signal estimate of the historical received signal, obtaining a current value of a preset loss function corresponding to the preset neural network based on the signal estimate of the historical received signal and the second real number array; determining the preset neural network that has been pre-trained when the current value of the preset loss function meets a preset convergence condition; Determine the pre-trained preset neural network based on The signal detection method according to claim 6.

10. The preset loss function is Mean Squared Error MSE loss function, Root Mean Square Error RMSE loss function, Mean Absolute Error (MAE) loss function, Euclidean distance loss function, Cosine distance loss function, and at least one of the following linear weighting functions: MSE loss function, RMSE loss function, MAE loss function, Euclidean distance loss function, and Cosine distance loss function. The signal detection method according to claim 9.

11. the step of inputting the current channel estimate and the received signal to be measured into a pre-trained neural network to perform signal detection and obtain a signal estimate of the received signal to be measured, combining the obtained current channel estimate with a real part and an imaginary part of the received signal to be measured to obtain a third real number array; inputting the third real sequence into a pre-trained neural network to perform signal detection, and obtaining a real signal estimation result of the received signal to be measured; Includes The signal detection method according to claim 1 .

12. The step of combining the obtained current channel estimate with a real part and an imaginary part of the received signal to be measured to obtain a third real number array includes: obtaining a third real part and a third imaginary part of the current channel estimate and obtaining a fourth real part and a fourth imaginary part of the received signal to be measured; arranging the third real part, the third imaginary part, the fourth real part, and the fourth imaginary part in a preset order to obtain a third real number array; Includes The signal detection method according to claim 11.

13. After obtaining a real signal estimate of the received signal to be measured, performing a complex transformation process on the real signal estimation result of the received signal under measurement to obtain a complex signal estimation result of the received signal under measurement; demodulating and decoding the complex signal estimation result to obtain an output signal; Also includes The signal detection method according to claim 11.

14. the real signal estimation result includes a real signal estimation array including real parts and imaginary parts of a plurality of hierarchical signals arranged in order; the step of performing a complex transform process on the real signal estimation result of the received signal to be measured to obtain a complex signal estimation result of the received signal to be measured, performing complex transformation on the real and imaginary parts of each of the hierarchical signals in the real signal estimation array to obtain complex transformation data corresponding to each of the hierarchical signals; arranging each of the complex transform data in a predetermined order to obtain a complex signal estimation result of the received signal to be measured; Includes The signal detection method according to claim 13.

15. said step of multiplying said received pilot signal by a conjugate of said local pilot signal to obtain a historical channel estimate comprises: using a least squares LS channel estimation algorithm to multiply the received pilot signal by a conjugate of the local pilot signal to obtain a historical channel estimate. The signal detection method according to claim 2 .

16. The step of obtaining a historical modulated signal transmitted by the transmitting side comprises: obtaining at least one set of transmitted bitstream signals transmitted by a transmitter; encoding and modulating at least one set of said transmission bit stream signals to obtain a hysteretic modulated signal; Includes The signal detection method according to claim 2 .

17. The preset neural network includes: Deep Neural Networks, Convolutional Neural Networks, Residual convolutional neural network, and a residual convolutional neural network with attention mechanism. The signal detection method according to claim 1 .

18. At least one processor; at least one memory for storing at least one program; Including, At least one of the programs, when executed by at least one of the processors, implements the signal detection method according to any one of claims 1 to 17. Signal detection device.

19. A program executable by the processor is stored; The processor-executable program, when executed by the processor, realizes the signal detection method according to any one of claims 1 to 17. A computer-readable storage medium.

20. A computer program or computer instructions stored on a computer readable storage medium, A processor of a computer device reads the computer program or the computer instructions from the computer-readable storage medium, and the processor executes the computer program or the computer instructions, thereby causing the computer device to perform the signal detection method according to any one of claims 1 to 17. Computer program products.

Images