Signal processing method and device based on generalized Rayleigh quotient and readable storage medium

Abstract

The invention relates to the technical field of communication, and provides a generalized Rayleigh quotient-based signal processing method and device and a computer readable storage medium, and the method comprises the steps: firstly carrying out the high-dimensional mapping of an initial receiving signal of a receiving end, and obtaining a signal feature vector; then determining an optimal weight vector of a corresponding target function in a generalized Rayleigh quotient form based on the initial receiving signal; and finally, based on the optimal weight vector and the signal feature vector, optimizing the distribution of the initial receiving signal, and generating a target receiving signal. According to the method, the distribution of the initial receiving signal is optimized through the optimal weight vector corresponding to the target function of the generalized Rayleigh quotient and the signal feature vector after high-dimensional mapping, so that non-linear damage and other residual damage caused by an optical fiber link in the initial receiving signal are compensated. Therefore, the calculation complexity of the compensation method is reduced, and the compensation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a signal processing method, device and computer-readable storage medium based on generalized Rayleigh quotient. Background Art

[0002] With the continuous development of big data, cloud computing, artificial intelligence and other related fields, the demand for communication services is also growing. In optical communication systems, fiber nonlinear effects, mainly Kerr nonlinear effects, are one of the main causes of signal damage and a key factor limiting the improvement of the system capacity of optical communication systems. Therefore, the research on methods and technologies for compensating and suppressing fiber nonlinear damage has important theoretical significance and application value. However, compensation methods such as digital back propagation (DBP), perturbation method, Wott series nonlinear equalizer, and constellation shaping based on the step-by-step Fourier transform method have high computational complexity, which leads to the problem of low compensation efficiency. Summary of the invention

[0003] The main purpose of the present invention is to provide a signal processing method, device and storage medium based on the generalized Rayleigh quotient, aiming to solve the technical problem of low compensation efficiency of the current optical fiber nonlinear damage compensation method.

[0004] To achieve the above object, the present invention provides a signal processing method based on the generalized Rayleigh quotient, the signal processing method based on the generalized Rayleigh quotient comprising:

[0005] Perform high-dimensional mapping on the initial received signal at the receiving end to obtain a signal feature vector;

[0006] Based on the initial received signal, determining an optimal weight vector of a corresponding objective function in the form of a generalized Rayleigh quotient;

[0007] Based on the optimal weight vector and the signal feature vector, the distribution of the initial received signal is optimized to generate a target received signal.

[0008] In addition, to achieve the above-mentioned purpose, the present invention also provides a signal processing device based on the generalized Rayleigh quotient, wherein the signal processing device based on the generalized Rayleigh quotient comprises a processor, a memory, and a signal processing program based on the generalized Rayleigh quotient stored in the memory and executable by the processor, wherein when the signal processing program based on the generalized Rayleigh quotient is executed by the processor, the steps of the signal processing method based on the generalized Rayleigh quotient as described above are implemented.

[0009] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a signal processing program based on the generalized Rayleigh quotient is stored, wherein when the signal processing program based on the generalized Rayleigh quotient is executed by a processor, the steps of the signal processing method based on the generalized Rayleigh quotient as described above are implemented.

[0010] The present invention provides a signal processing method based on generalized Rayleigh quotient, the method first performs high-dimensional mapping on the initial received signal of the receiving end to obtain a signal feature vector; then, based on the initial received signal, the optimal weight vector of the corresponding objective function in the form of generalized Rayleigh quotient is determined; finally, based on the optimal weight vector and the signal feature vector, the distribution of the initial received signal is optimized to generate a target received signal. In the above manner, the present invention optimizes the distribution of the initial received signal through the optimal weight vector corresponding to the objective function of the generalized Rayleigh quotient and the signal feature vector after high-dimensional mapping, thereby compensating for nonlinear damage and other residual damage caused by the optical fiber link in the initial received signal. Therefore, the present invention not only reduces the computational complexity of the compensation method, but also improves the compensation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Schematic diagram of the hardware structure of a signal processing device based on the generalized Rayleigh quotient involved in the embodiment of the present application;

[0012] Figure 2 This is a flow chart of a first embodiment of a signal processing method based on the generalized Rayleigh quotient of the present application;

[0013] Figure 3 This is a flow chart of a second embodiment of a signal processing method based on the generalized Rayleigh quotient of the present application;

[0014] Figure 4 A schematic diagram of the optimization process of the enhanced space algorithm provided in this application;

[0015] Figure 5 A schematic diagram of the working process of the optical communication system of this application;

[0016] Figure 6 This is a schematic diagram of the setup of the coherent reception wavelength division multiplexing optical communication experimental system of this application;

[0017] Figure 7 This is a schematic diagram of the digital signal processing flow of this application;

[0018] Figure 8 This is a schematic diagram of the structure of sending data frames in this application.

[0019] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0020] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0021] The signal processing method based on the generalized Rayleigh quotient involved in the embodiment of the present invention is mainly applied to the signal processing device based on the generalized Rayleigh quotient, which can be a PC, a portable computer, a mobile terminal and other devices with display and processing functions.

[0022] Reference Figure 1 , Figure 1 Schematic diagram of the hardware structure of the signal processing device based on the generalized Rayleigh quotient involved in the embodiment of the present invention. In the embodiment of the present invention, the signal processing device based on the generalized Rayleigh quotient may include a processor 1001 (such as a CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components; the user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard); the network interface 1004 may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface); the memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory, and the memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0023] Those skilled in the art will understand that Figure 1 The hardware structure shown in the figure does not constitute a limitation on the signal processing device based on the generalized Rayleigh quotient, and may include more or less components than those shown in the figure, or combine certain components, or arrange the components differently.

[0024] Continue to refer to Figure 1 , Figure 1 The memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, and a signal processing program based on the generalized Rayleigh quotient.

[0025] exist Figure 1 In the embodiment, the network communication module is mainly used to connect to the server and perform data communication with the server; and the processor 1001 can call the signal processing program based on the generalized Rayleigh quotient stored in the memory 1005, and execute the signal processing method based on the generalized Rayleigh quotient provided in the embodiment of the present invention.

[0026] An embodiment of the present invention provides a signal processing method based on generalized Rayleigh quotient.

[0027] Reference Figure 2 , Figure 2 4 is a flow chart of a first embodiment of a signal processing method based on the generalized Rayleigh quotient of the present invention.

[0028] In this embodiment, the signal processing method based on the generalized Rayleigh quotient includes the following steps:

[0029] Step S10, performing high-dimensional mapping on the initial received signal of the receiving end to obtain a signal feature vector;

[0030] Currently, compensation methods are mainly divided into two categories: traditional methods and methods based on machine learning and neural networks.

[0031] Traditional methods include digital back propagation (DBP) based on the split-step Fourier transform method, perturbation method, Wott series nonlinear equalizer, constellation shaping and other nonlinear compensation methods. However, the above methods have problems such as high computational complexity, limited compensation performance, and difficulty in implementation and application.

[0032] At present, various neural networks such as artificial neural network (ANN), convolutional neural network (CNN), recurrent neural network (RNN) are used to compensate the receiving end signal in optical communication systems. For example, the digital back propagation method (LDBP) based on neural network combines the traditional method with machine learning and neural network compensation methods. However, the above methods have the problems of high training difficulty, high computational complexity and poor generalization ability.

[0033] In order to solve the above problem, this embodiment optimizes the distribution of the initial received signal through the optimal weight vector corresponding to the objective function of the generalized Rayleigh quotient, thereby compensating for the nonlinear damage and other residual damage caused by the optical fiber link in the initial received signal, thereby reducing the computational complexity and improving the compensation efficiency.

[0034] The signal processing method provided in this embodiment is mainly used in the digital signal processing flow at the receiving end of a long-distance coherent optical communication system. The received signal distribution at the receiving end is enhanced by an enhanced spatial algorithm, including enhanced spatial transformation, divergence matrix calculation, objective function optimization, and enhanced spatial optimization. This effectively suppresses optical fiber nonlinear damage and other residual damage, thereby improving system performance.

[0035] Specifically, a linear model is used to perform high-dimensional mapping on the initial received signal to be processed at the receiving end, and the received signal is enhanced spatially transformed to obtain the corresponding signal feature vector, thereby preliminarily suppressing nonlinear damage between channels and other residual damage.

[0036] Furthermore, the step S10 includes:

[0037] Preprocessing the initial received signal to obtain a received signal in a dual polarization state;

[0038] High-dimensional mapping is performed on the dual-polarization received signal to obtain the signal feature vector.

[0039] In this embodiment, the digital signal received by the receiving end (i.e., the initial received signal) is preprocessed to obtain a QAM signal sequence of dual polarization state sampling (i.e., a received signal of dual polarization state). The received signal of each polarization state is mapped in high dimension through a linear model to obtain a corresponding signal feature vector sequence as the signal feature vector.

[0040] Among them, the specific calculation method of the high-dimensional mapping of the linear model is as follows:

[0041] F n,y =[···,s n+l,y ,···,s n+l,x ,···] T

[0042] F n,x =[···,s n+l,x ,···,s n+l,y ,···] T

[0043] In s n,x / y represents the nth symbol in the received signal sequence of x or y polarization state, F n,x / y The signal eigenvector corresponding to the received signal of the x or y polarization state, F n,x / y It is calculated from the received signal of the x or y polarization state and its adjacent signals, l represents the range of adjacent signal subscripts, l = -L1, ..., 0, ..., L1, L = 2 (2L1 + 1) is the dimension of the signal feature vector.

[0044] Step S20, determining an optimal weight vector of a corresponding objective function in the form of a generalized Rayleigh quotient based on the initial received signal;

[0045] In this embodiment, before optimizing the initial received signal, the optimal weight vector of the objective function in the form of the generalized Rayleigh quotient corresponding to the initial received signal can be pre-calculated, and when optimizing the initial received signal, the optimal weight vector corresponding to the initial received signal can be obtained; or when optimizing the initial received signal, the optimal weight vector corresponding to the initial received signal can be calculated.

[0046] Step S30: Optimizing the distribution of the initial received signal based on the optimal weight vector and the signal feature vector to generate a target received signal.

[0047] In this embodiment, the signal distribution of the initial received signal is optimized using an optimal weight vector to obtain an optimized target received signal. The optimal weight vector is a weight vector that maximizes the objective function in the form of a generalized Rayleigh quotient.

[0048] This embodiment provides a signal processing method based on generalized Rayleigh quotient, which first performs high-dimensional mapping on the initial received signal of the receiving end to obtain a signal feature vector; then, based on the initial received signal, determines the optimal weight vector of the corresponding objective function in the form of generalized Rayleigh quotient; finally, based on the optimal weight vector and the signal feature vector, optimizes the distribution of the initial received signal to generate a target received signal. In the above manner, this embodiment optimizes the distribution of the initial received signal through the optimal weight vector corresponding to the objective function of the generalized Rayleigh quotient and the signal feature vector after high-dimensional mapping, thereby compensating for nonlinear damage and other residual damage caused by the optical fiber link in the initial received signal. Therefore, the present invention not only reduces the computational complexity of the compensation method, but also improves the compensation efficiency.

[0049] Reference Figure 3 , Figure 3 4 is a flow chart of a second embodiment of a signal processing method based on the generalized Rayleigh quotient of the present invention.

[0050] Based on the above Figure 2 In the embodiment shown, in this embodiment, the step S20 specifically includes:

[0051] Step S21, based on the transmission signal category corresponding to the initial received signal and the signal feature vector, respectively calculating the dispersion metric within the symbol category and the dispersion metric between the symbol categories;

[0052] Step S22, defining the objective function according to the initial weight vector, the dispersion measure within the symbol category and the dispersion measure between the symbol categories;

[0053] Step S23, determining the optimal weight vector based on the maximum value of the objective function, wherein the optimal weight vector is the weight vector that achieves the maximum value of the objective function.

[0054] In this embodiment, the calculation method of the optimal weight vector of the objective function in the form of generalized Rayleigh quotient includes a training mode and a decision feedback mode.

[0055] like Figure 4As shown in the training mode in , the training mode is that when the transmission signal category corresponding to the partial received signal is known, a part of the initial received signal in the data frame is fixed as the training sequence, and the dispersion measure within the symbol category and the dispersion measure between the symbol categories are calculated using the part of the training sequence to complete the dispersion matrix calculation, and then the generalized Rayleigh quotient form objective function is defined by using the calculated dispersion measure within the symbol category and the dispersion measure between the symbol categories, and the objective function is maximized to complete the objective function optimization and obtain the optimal weight vector. Finally, the optimal weight vector is used to optimize the initial received signal to complete the enhanced space optimization.

[0056] Furthermore, before step S21, the method further includes:

[0057] A category decision is performed on the initial received signal to obtain a category decision result as a sending signal category corresponding to the initial received signal.

[0058] In this embodiment, when the type of the transmitted signal corresponding to the received signal is unknown, the type of the transmitted signal is obtained through the decision feedback mode. Figure 4 As shown in the decision feedback mode in , the decision feedback mode is to make category decisions on all signals of the initial received signal in the data frame, and use the decision result as the category estimate of the transmitted signal. Then, based on the obtained category decision result, the category of the transmitted signal corresponding to the initial received signal is used, and finally the dispersion measure within the symbol category and the dispersion measure between the symbol categories are calculated. Then, the generalized Rayleigh quotient form objective function is defined using the calculated dispersion measure within the symbol category and the dispersion measure between the symbol categories, and the objective function is maximized to obtain the optimal weight vector. Finally, the optimal weight vector is used to optimize the signal distribution of all initial received signals of the transmitted data.

[0059] It can be understood that the training mode has a lower computational complexity, but requires the known categories of transmitted signals corresponding to some training sequences, and has a lower spectrum efficiency; the decision feedback mode has known categories of transmitted signals corresponding to some training sequences, and has a higher spectrum efficiency, but has a higher computational complexity. Therefore, when the power is low and the bit error rate is low, the optimal weight vector of the initial received signal can be trained through the decision feedback mode, and when the power is high and the bit error rate is high, the optimal weight vector of the initial received signal can be trained through the training mode. In actual applications, the mode can be selected according to specific needs and application scenarios, so that different modes can be used at different power points to improve system performance. This embodiment uses the decision feedback method to perform weight training using the signal category information obtained by the decision, thereby improving spectrum efficiency and system performance.

[0060] Specifically, calculations are performed based on the signal feature vector sequence to obtain the dispersion measure within the symbol category and the dispersion measure between the symbol categories. Then, an objective function in the form of a generalized Rayleigh quotient is defined based on the dispersion measure within the symbol category, the dispersion measure between the symbol categories, and the weight vector to be optimized. Thus, the optimal weight vector for optimizing the signal distribution is obtained by maximizing the objective function in the form of a generalized Rayleigh quotient.

[0061] In a specific embodiment, the solution formula for the optimal weight vector is as follows:

[0062]

[0063]

[0064] in, represents the optimal weight vector for the received signal in x or y polarization state, w x / y The initial weight vector representing the received signal of the x or y polarization state, the generalized Rayleigh quotient J is the weight vector w x / y Function, weight vector w x / y is an arbitrary matrix in 1×L dimensional complex space.

[0065] The optimal weight vector for the received signal in the x or y polarization state is the matrix The transposed complex conjugate of the eigenvector corresponding to the largest eigenvalue, the formula for solving the matrix eigenvalue is:

[0066]

[0067]

[0068] Furthermore, before step S01, the method further includes:

[0069] The original received signal of the receiving end is divided into a plurality of data windows, and the signal corresponding to each data window is respectively obtained as the initial received signal, so as to respectively calculate the optimal weight vector of the signal corresponding to each data window.

[0070] In this embodiment, when the decision feedback mode is used for optimal weight vector training, the original received signal data received by the receiving end is divided into multiple data windows. The signals corresponding to each data window are respectively obtained as the initial received signal. Thus, the optimal weight vector training and signal distribution optimization are respectively performed on the signal corresponding to each data window. Thus, by optimizing the length of the data window corresponding to the signal, the received signal is segmented, and the optimal weight vector corresponding to each segment is calculated, so as to improve the accuracy of the optimal weight vector corresponding to each segmented signal, thereby improving the system performance.

[0071] Furthermore, the step S21 includes:

[0072] Based on the signal feature vector and the category of the transmitted signal corresponding to the initial received signal, calculating the mean of the signal feature vectors of each category and the mean of all the initial received signal feature vectors;

[0073] Based on the mean of the characteristic vectors of the signals of each category and the mean of the characteristic vectors of all the initial received signals, the dispersion measure within the symbol category and the dispersion measure between the symbol categories are respectively calculated and obtained.

[0074] In this embodiment, the mean of the characteristic vectors of signals of different categories is first calculated based on the initial received signal and the corresponding transmitted signal category, and the dispersion measure within the symbol category is further calculated according to the formula based on the mean and category information. At the same time, the mean of all signal characteristic vectors is calculated based on the initial received signal, and the dispersion measure between symbol categories is further calculated according to the formula.

[0075] The calculation formula for the dispersion measure within the symbol category is:

[0076]

[0077]

[0078] The calculation formula for the dispersion measure between symbol categories is:

[0079]

[0080]

[0081] Among them, F n,x / y represents the signal feature vector corresponding to the initial received signal, m represents the type of transmitted signal, v m,x / y represents the mean of the signal eigenvectors corresponding to the received signals of category m for all transmitted signals in the x or y polarization state. The superscript H represents the complex conjugate of the matrix transpose, N represents the number of initial received signals, and μ x / y represents the mean of the signal feature vectors corresponding to all received symbols of the x or y polarization state. C(n,x / y) represents the category of the transmitted signal corresponding to the nth symbol in the received signal sequence of the x or y polarization state.

[0082] Furthermore, the step S30 includes:

[0083] The signal feature vector is dot-producted with the optimal weight vector to obtain the target received signal.

[0084] In this embodiment, after the optimal weight vector is calculated, the signal feature vector is dot-producted with the optimal weight vector to restore the signal feature vector to the received signal as the target received signal, and the received signal distribution is optimized. The specific formula is as follows:

[0085]

[0086]

[0087] Among them, s′ n,x / y The nth symbol in the received signal sequence representing the optimized x or y polarization state, that is, the optimized symbol is calculated by the signal feature vector and the optimal weight vector.

[0088] Furthermore, after step S30, the method further includes:

[0089] The target received signal is judged based on a preset clustering model to optimize clustering of unconventional signals in the target received signal.

[0090] In this embodiment, the specific clustering methods that can be used are K-means clustering and mixed Gaussian model clustering. The signal category is determined for the optimized unconventional signal distribution, and the determination result is obtained and subsequent signal processing is performed. Among them, the unconventional signal is a signal outside each category, that is, the unconventional signal outside each category in the target received signal is clustered and optimized again.

[0091] In this embodiment, the initial received signal is mapped to a high dimension, and the optimal weight vector is obtained by maximizing the generalized Rayleigh quotient, that is, the characteristic basis that satisfies the goal of maximizing inter-class separation is determined by analyzing the intra-class divergence and inter-class divergence of the signal characteristic vector. Then, the signal distribution is optimized according to the optimal weight vector, the symbol distribution spacing is increased, and the clustering algorithm is used to judge the optimized unconventional signal distribution, further suppressing the nonlinear damage of the optical fiber and other types of residual damage, improving the system performance of the coherent optical communication system, and improving the system performance while reducing the computational complexity.

[0092] Furthermore, this embodiment is described in conjunction with a specific coherent optical communication experimental system.

[0093] like Figure 5 As shown in the figure, the basic working process of the optical communication system is as follows:

[0094] 1. Send a binary bit sequence, encode, modulate and perform digital signal processing on the binary bit sequence to obtain an electrical signal to be sent;

[0095] 2. Convert the electrical signal into an optical signal by means of a modulated laser in the transmitter to obtain an optical signal to be transmitted;

[0096] 3. Transmitting the optical signal to a receiver via an optical fiber channel;

[0097] 4. In the receiver, the optical signal is converted into an electrical signal through a photodiode, and the electrical signal is processed to obtain a digital signal to be processed;

[0098] It is understandable that dispersion and nonlinear effects in optical fibers will cause optical signals to be damaged during transmission. Therefore:

[0099] 5. The signal processing method provided by the present application (i.e., enhanced space algorithm) is used to perform distribution optimization (i.e., enhanced space transformation, divergence matrix calculation, objective function optimization, and enhanced space optimization) and clustering judgment on the above-mentioned digital signal to be processed, and the nonlinear damage of the optical fiber and other types of residual damage in the signal are compensated to obtain an optimized digital signal. The optimized digital signal is demodulated and decoded to obtain the final received binary sequence, completing the entire communication process.

[0100] By building a coherent receiving wavelength division multiplexing optical communication experimental system (such as Figure 6 The specific parameters of the experimental system are: the number of wavelength division multiplexing wavelengths is 8, the baud rate is 64GBaud, the modulation format is 16QAM, the transmission distance is 1600km, and the span length is 80km.

[0101] like Figure 7 As shown in the figure, the digital signal processing (DSP) process at the transmitting end is as follows: first, the modulated symbol sequence is upsampled, and then the sampling sequence is waveform-shaped through a square root raised cosine filter; the digital signal processing process at the receiving end is as follows: first, the received signal is resampled to obtain a sampling sequence of twice the symbol rate, and then the sampling sequence of twice the symbol rate is sequentially subjected to dispersion compensation, frequency offset elimination, root raised cosine matched filtering, synchronization, and 2*2MMIMO time domain equalization processing, thereby processing and compensating for signal damage such as dispersion, time offset, and frequency offset, and depolarization multiplexing is completed through time domain equalization. Then, the processed signal is downsampled to obtain a sampling sequence of one symbol rate, and then the sampling sequence of one symbol rate is sequentially subjected to pilot-based phase recovery and blind phase search-based phase recovery.

[0102] Figure 8 The diagram is a structural diagram of the data frame sent in the transmission experiment. The first part of the data frame is a synchronization sequence, which is used for synchronization in the digital signal processing at the receiving end described in the above steps; the second part of the data frame is a training sequence, which is used for time domain equalization in the digital signal processing at the receiving end described in the above steps to complete polarization demultiplexing. The third part of the data frame is the transmission data, in which pilot processing is added, which is used for pilot-based phase recovery in the digital signal processing at the receiving end described in the above steps.

[0103] The signal processing method proposed in this application is verified based on the results of the transmission experiment, that is, the system performance measured by adding the signal processing method proposed in this application to the digital signal processing process at the receiving end is compared. According to the comparison results, the system performance of optimizing the signal processing by adding the signal processing method proposed in this application to the digital signal processing process at the receiving end increases the system Q value gain by about 0.55dB, which proves that the signal processing method proposed in this application effectively suppresses nonlinear damage and other residual damage, and improves the performance of the coherent optical communication system.

[0104] In addition, an embodiment of the present invention further provides a signal processing device based on the generalized Rayleigh quotient.

[0105] In this embodiment, the signal processing device based on the generalized Rayleigh quotient includes:

[0106] A signal feature extraction module is used to perform high-dimensional mapping on the initial received signal at the receiving end to obtain a signal feature vector;

[0107] An optimal vector determination module, used to determine an optimal weight vector of a corresponding objective function in the form of a generalized Rayleigh quotient based on the initial received signal;

[0108] The received signal optimization module is used to optimize the distribution of the initial received signal based on the optimal weight vector and the signal feature vector to generate a target received signal.

[0109] Furthermore, the optimal vector determination module includes:

[0110] A dispersion metric calculation unit, configured to calculate a dispersion metric within a symbol category and a dispersion metric between symbol categories based on the category of the transmitted signal corresponding to the initial received signal and the signal feature vector;

[0111] An objective function definition unit, configured to define the objective function according to an initial weight vector, a dispersion measure within the symbol category, and a dispersion measure between the symbol categories;

[0112] The optimal vector determining unit is used to determine the optimal weight vector based on the maximum value of the objective function, wherein the optimal weight vector is the weight vector that obtains the maximum value of the objective function.

[0113] Furthermore, the optimal vector determination module further includes:

[0114] The signal category estimation unit is used to perform category determination on the initial received signal to obtain a category determination result as the sending signal category corresponding to the initial received signal.

[0115] Furthermore, the optimal vector determination module further includes:

[0116] The data window division unit is used to divide the original received signal of the receiving end into multiple data windows, and obtain the signal corresponding to each data window as the initial received signal, so as to calculate the optimal weight vector of the signal corresponding to each data window.

[0117] Furthermore, the dispersion metric calculation unit is also used for:

[0118] Based on the signal feature vector and the category of the transmitted signal corresponding to the initial received signal, calculating the mean of the signal feature vectors of each category and the mean of all the initial received signal feature vectors;

[0119] Based on the mean of the characteristic vectors of the signals of each category and the mean of the characteristic vectors of all the initial received signals, the dispersion measure within the symbol category and the dispersion measure between the symbol categories are respectively calculated and obtained.

[0120] Furthermore, the signal processing based on the generalized Rayleigh quotient also includes:

[0121] The received signal clustering module is used to judge the target received signal based on a preset clustering model to optimize the clustering of unconventional signals in the target received signal.

[0122] Furthermore, the received signal optimization module includes:

[0123] The vector point set restoration unit is used to perform a dot product between the signal feature vector and the optimal weight vector to obtain the target received signal.

[0124] Furthermore, the signal feature extraction module includes:

[0125] A signal processing unit, used for preprocessing the initial received signal to obtain a received signal in a dual polarization state;

[0126] The signal mapping unit is used to perform high-dimensional mapping on the received signal in the dual polarization states to obtain the signal feature vector.

[0127] Among them, each module in the above-mentioned signal processing device based on generalized Rayleigh quotient corresponds to each step in the above-mentioned signal processing method embodiment based on generalized Rayleigh quotient, and their functions and implementation processes are no longer described one by one here.

[0128] In addition, an embodiment of the present invention further provides a computer-readable storage medium.

[0129] The computer-readable storage medium of the present invention stores a signal processing program based on the generalized Rayleigh quotient, wherein when the signal processing program based on the generalized Rayleigh quotient is executed by a processor, the steps of the signal processing method based on the generalized Rayleigh quotient as described above are implemented.

[0130] Among them, the method implemented when the signal processing program based on the generalized Rayleigh quotient is executed can refer to the various embodiments of the signal processing method based on the generalized Rayleigh quotient of the present invention, and will not be repeated here.

[0131] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0132] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0133] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0134] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0135] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A signal processing method based on generalized Rayleigh quotient, characterized in that: The signal processing method based on the generalized Rayleigh quotient includes: Perform high-dimensional mapping on the initial received signal at the receiving end to obtain a signal feature vector; Based on the initial received signal, determining an optimal weight vector of a corresponding objective function in the form of a generalized Rayleigh quotient; Based on the optimal weight vector and the signal feature vector, the distribution of the initial received signal is optimized to generate a target received signal.

2. The signal processing method according to claim 1, characterized in that: Determining the optimal weight vector of the objective function in the form of a generalized Rayleigh quotient corresponding to the initial received signal comprises: Based on the transmission signal category corresponding to the initial received signal and the signal feature vector, respectively calculating the dispersion metric within the symbol category and the dispersion metric between the symbol categories; The objective function is defined according to the initial weight vector, the dispersion measure within the symbol category, and the dispersion measure between the symbol categories; Based on the maximum value of the objective function, the optimal weight vector is determined, wherein the optimal weight vector is the weight vector that achieves the maximum value of the objective function.

3. The signal processing method according to claim 2, characterized in that: Before respectively calculating the dispersion measure within the symbol category and the dispersion measure between the symbol categories based on the transmitted signal category corresponding to the initial received signal and the signal feature vector, the method further includes: A category decision is performed on the initial received signal to obtain a category decision result as a sending signal category corresponding to the initial received signal.

4. The signal processing method according to claim 2, characterized in that: Before respectively calculating the dispersion measure within the symbol category and the dispersion measure between the symbol categories based on the transmitted signal category corresponding to the initial received signal and the signal feature vector, the method further includes: The original received signal of the receiving end is divided into a plurality of data windows, and the signal corresponding to each data window is respectively obtained as the initial received signal.

5. The signal processing method according to claim 2, characterized in that: The calculating, based on the transmitted signal category corresponding to the initial received signal and the signal feature vector, respectively obtaining the dispersion measure within the symbol category and the dispersion measure between the symbol categories, comprises: Based on the signal feature vector and the category of the transmitted signal corresponding to the initial received signal, calculating the mean of the signal feature vectors of each category and the mean of all the initial received signal feature vectors; Based on the mean of the characteristic vectors of the signals of each category and the mean of the characteristic vectors of all the initial received signals, the dispersion measure within the symbol category and the dispersion measure between the symbol categories are respectively calculated and obtained.

6. The signal processing method according to claim 1, characterized in that: After optimizing the distribution of the initial received signal based on the optimal weight vector and the signal feature vector to generate a target received signal, the method further includes: The target received signal is judged based on a preset clustering model to optimize clustering of unconventional signals in the target received signal.

7. The signal processing method based on generalized Rayleigh quotient according to claim 1, characterized in that: The optimizing the distribution of the initial received signal based on the optimal weight vector and the signal feature vector to generate a target received signal includes: The signal feature vector is dot-producted with the optimal weight vector to obtain the target received signal.

8. The signal processing method based on the generalized Rayleigh quotient according to any one of claims 1 to 7, characterized in that: The high-dimensional mapping of the initial received signal at the receiving end to obtain a signal feature vector includes: Preprocessing the initial received signal to obtain a received signal in a dual polarization state; High-dimensional mapping is performed on the dual-polarization received signal to obtain the signal feature vector.

9. A signal processing device based on generalized Rayleigh quotient, characterized in that: The signal processing device based on the generalized Rayleigh quotient includes a processor, a memory, and a signal processing program based on the generalized Rayleigh quotient stored in the memory and executable by the processor, wherein when the signal processing program based on the generalized Rayleigh quotient is executed by the processor, the steps of the signal processing method based on the generalized Rayleigh quotient as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a signal processing program based on a generalized Rayleigh quotient, wherein when the signal processing program based on a generalized Rayleigh quotient is executed by a processor, the steps of the signal processing method based on a generalized Rayleigh quotient as claimed in any one of claims 1 to 8 are implemented.

Images