A low-power mixed signal processing method based on multi-carrier receiver

Abstract

The present application discloses a mixed signal low-power processing method based on a multi-carrier receiver, which relates to the field of communication technology, including: mapping a pre-processed information bit sequence to a filter bank multi-carrier modulation symbol set FBMC; using Logistic mapping to generate a chaotic sequence and construct an original information carrying matrix; dividing the original information carrying matrix into M sub-matrices according to the number of sub-carriers M of a differential chaotic shift keying DCSK system, and performing matrix decomposition processing; adaptively optimizing the decomposed information carrying sub-matrix; reconstructing the decomposed information carrying sub-matrix; inserting pilot symbols and control information into the obtained information carrying reconstruction matrix; pre-coding the information carrying reconstruction matrix into which the pilot symbols and control information are inserted; transmitting the generated time domain transmission symbol sequence to a wireless channel after digital-to-analog conversion and radio frequency front-end processing. In view of the high power consumption of multi-carrier receivers in the prior art, the present application reduces the power consumption.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a mixed signal low-power consumption processing method based on a multi-carrier receiver. Background Art

[0002] With the rapid development of wireless communication technology, mobile communication systems are evolving towards high speed, large capacity and low latency. In order to meet the growing user needs and diversified business applications, multi-carrier technologies such as orthogonal frequency division multiplexing (OFDM) and filter bank multi-carrier (FBMC) are widely used in modern communication systems. Multi-carrier technology effectively improves spectrum utilization efficiency and transmission reliability by dividing high-speed data streams into multiple low-speed sub-data streams and transmitting them in parallel on different sub-carriers.

[0003] However, receiver design for multi-carrier systems faces many challenges, among which power consumption is particularly prominent. Traditional multi-carrier receivers usually adopt a fully digital processing architecture, which requires independent channel estimation, equalization, demodulation and other operations for each subcarrier, resulting in a significant increase in the receiver's computational complexity and power consumption. High power consumption not only shortens the battery life of mobile terminals, but also limits the application of multi-carrier technology in low-power IoT, wireless sensor networks and other fields. Therefore, how to reduce the power consumption of multi-carrier receivers while ensuring communication performance has become a key issue that needs to be solved urgently.

[0004] In the related art, for example, Chinese patent document CN116319212B provides a multi-carrier DCSK signal reconstruction method and device. The multi-carrier DCSK signal reconstruction method provided by the present application obtains the original information carrying matrix by cross-multiplying the information bit sequence to be transmitted after modulation with the chaotic signal, and then reconstructs the original information carrying matrix through a preset reconstruction matrix, and generates a transmission signal for transmission to the receiving end by combining the obtained information carrying reconstruction matrix with the reference signal matrix, so that the receiving end demodulates the received transmission symbol according to the same reconstruction matrix, thereby restoring the information bits to be transmitted. However, in this scheme, the generation and application of the reconstruction matrix requires a large number of matrix operations, such as matrix multiplication, matrix inversion, etc. These operations usually have high computational complexity, requiring the receiver to consume a large amount of computing resources and energy, resulting in an increase in power consumption. Summary of the invention

[0005] In view of the problem of high power consumption of multi-carrier receivers in the prior art, the present application provides a low-power mixed signal processing method based on a multi-carrier receiver, which adopts Logistic chaotic mapping to construct an information carrying matrix, utilizes Hadamard matrix decomposition to realize information carrying matrix division, etc., thereby reducing the power consumption of the multi-carrier receiver.

[0006] Technical solution, the purpose of this application is achieved through the following technical solution.

[0007] The present specification provides a low-power processing method for mixed signals based on a multi-carrier receiver, including: information bit sequence preprocessing (step 1): performing serial-to-parallel conversion on the information bit sequence to be transmitted, and converting the serial bit stream into a parallel bit stream. Polar code is used to perform channel coding on the parallel bit stream to improve the reliability of information transmission. Multi-carrier modulation mapping (step 2): mapping the channel-coded bit sequence to a filter bank multi-carrier (FBMC) modulation symbol set to obtain a modulation symbol sequence. Constructing the original information carrying matrix (step 3): using Logistic mapping to generate a chaotic sequence. Logistic mapping is a simple and widely used chaotic system with the characteristics of being highly sensitive to initial values ​​and parameters and having good randomness of the generated sequence. Performing a cross product operation on the modulation symbol sequence and the chaotic sequence to construct the original information carrying matrix. The cross product operation can increase the randomness and security of the information. Matrix decomposition processing (step 4): according to the number of subcarriers M of the differential chaotic shift keying (DCSK) system, the original information carrying matrix is ​​divided into M sub-matrices. Performing matrix decomposition processing on each sub-matrix to obtain the decomposed information carrying sub-matrix. Matrix decomposition can transform the original matrix into multiple low-rank sub-matrices, which is convenient for subsequent parallel processing and optimization.

[0008] Adaptive optimization (step 5): The reconstruction matrix parameters and power allocation strategy of the decomposed information-bearing submatrix are optimized. The reconstruction matrix parameter optimization aims to find the optimal reconstruction matrix to minimize the reconstruction error. It can be solved iteratively by using optimization algorithms such as gradient descent. The power allocation strategy optimization aims to optimally allocate the total power to each subcarrier according to the channel state to maximize the system throughput. It can be solved by using water filling algorithm and the like. Information reconstruction (step 6): The decomposed information-bearing submatrix is ​​reconstructed using the optimized reconstruction matrix parameters and power allocation strategy to obtain the information-bearing reconstruction submatrix. The reconstruction process can restore the original information while reducing the influence of noise and interference. Weighted summation (step 7): The obtained information-bearing reconstruction submatrices are weighted summed to obtain the final information-bearing reconstruction matrix. The weighted summation can comprehensively analyze the reconstruction effects of each submatrix and improve the overall reconstruction performance. The weighting coefficient can be adaptively adjusted according to the reconstruction quality of the submatrix.

[0009] Insert pilot and control information (step 8): insert pilot symbols and control information into the information bearing reconstruction matrix for channel estimation and transmission control. Set a reasonable frame structure and specify the arrangement of pilot, control information, synchronization sequence and data symbols to achieve effective use and management of the channel. Precoding processing (step 9): perform precoding processing on the information bearing reconstruction matrix after inserting pilot and control information, and perform adaptive optimization according to channel state information (CSI). Precoding can pre-distortion compensate the signal, offset the frequency selective fading and phase distortion of the channel, and improve the signal quality at the receiving end. According to the set frame structure, convert the precoded matrix into a time domain transmission symbol sequence. Specifically, according to the set frame structure, convert the precoded information bearing reconstruction matrix into a time domain transmission symbol sequence. Usually, the information bearing reconstruction matrix in the frequency domain is converted to the time domain by using technologies such as inverse Fourier transform (IFFT) or inverse filter bank transform (IFBT) to generate a transmission symbol sequence in the time domain. Digital-to-analog conversion and radio frequency processing (step 10): map the generated time domain transmission symbol sequence into a continuous time signal through digital-to-analog conversion (DAC). The continuous time signal is processed by the RF front end, including up-conversion, filtering and power amplification, and finally the signal is transmitted into the wireless channel. The power amplifier of the RF front end can increase the transmission power of the signal and expand the coverage. Specifically, the digital-to-analog conversion converts the time domain transmission symbol sequence in the digital domain into a continuous time signal in the analog domain so that it can be transmitted on the physical channel. After the precoded information carrying reconstruction matrix is ​​converted in the time domain, its corresponding time domain transmission symbol sequence has a discrete time form suitable for digital-to-analog conversion. The analog signal after digital-to-analog conversion needs to be processed by the RF front end before it can be transmitted into the wireless channel. The RF front end includes components such as power amplifiers, filters, and antennas to amplify, filter, and transmit the analog signal. After the digital-to-analog conversion, the generated time domain transmission symbol sequence becomes the input of the RF front end processing and is finally transmitted through the antenna. The precoding process is to adaptively optimize the information carrying reconstruction matrix based on the channel state information (CSI). The purpose of precoding is to preprocess the signal at the transmitting end to offset the influence of the channel and improve the detection performance of the signal at the receiving end. After the information-carrying reconstruction matrix optimized by precoding is processed by time domain conversion, digital-to-analog conversion and RF front-end, it can better adapt to channel conditions and reduce distortion and interference caused by the channel. The set frame structure determines the arrangement of pilot, control information, synchronization sequence and data symbols in the time domain transmission symbol sequence. The design of the frame structure will affect the processing of the RF front-end and the transmission of the wireless channel. Reasonable frame structure design can improve the synchronization performance of the signal, reduce pilot overhead, improve the quality of channel estimation, etc., thereby optimizing the entire transmission and transmission process.

[0010] Furthermore, in step 3, constructing the original information bearing matrix includes: reordering the modulation symbol sequence: reordering the modulation symbol sequence obtained in step 2 according to the number of subcarriers N of the FBMC modulation symbol set. If the original modulation symbol sequence is [s1, s2, ..., s M}, where M is the total number of modulation symbols and N is the number of subcarriers, then the rearranged modulation symbol sequence is: The rearranged modulation symbol sequence is arranged in the order of subcarriers, and each row corresponds to the modulation symbol of a subcarrier. Initialize the information bearing matrix: Initialize an information bearing matrix A with the number of rows equal to the number of subcarriers N in the FBMC modulation symbol set and the number of columns equal to the product of the chaotic sequence length L and the symbol period T of the FBMC modulation symbol set. The dimension of matrix A is N rows and LT columns, expressed as: Initially, all elements of matrix A are 0. Loading modulation symbols: Write the rearranged modulation symbol sequence row by row into the initialized information-carrying matrix A. The elements a(i, 1) to a(i, T) in the i-th row are sequentially written with the T modulation symbols corresponding to the i-th subcarrier, and the remaining elements remain 0. The written matrix A is called the information-carrying matrix for loading modulation symbols, which is expressed as: Generate chaotic sequence: Use Logistic mapping to generate a chaotic sequence C with the same number of columns as the information-carrying matrix A. The iterative formula of Logistic mapping is: x(n+1)=μ×x(n)×(1-x(n)), where μ is the control parameter and x(n) is the value of the nth iteration. Select the appropriate initial value x(0) and control parameter μ, iterate LT times, and obtain a chaotic sequence C=[c1,c2,…,C L×T ].

[0011] Cross product operation: The information carrying matrix A loaded with modulation symbols s Each row element in is cross-producted with the corresponding element of chaotic sequence C to obtain the original information carrying matrix A c A c The cross product operation is to multiply the corresponding elements. The matrix A c The element a in the i-th row and j-th column c The calculation formula for (i, j) is: c (i, j) = a s (i, j)×c(j). After the cross product operation, the dimension of the original information-bearing matrix A_c is still N rows and LT columns, expressed as:

[0012]

[0013] Furthermore, in step 4, the decomposed information carrying matrix is ​​obtained, including: dividing the sub-matrix: the original information carrying matrix A obtained in step 3 is cAccording to the number of subcarriers M in the DCSK system, we can obtain M submatrices. c The dimension of is N rows LT columns, then the dimension of each sub-matrix is ​​(N / M) rows LT columns. The sub-matrices after division are represented as A c1 , A c2 ,......,A cM . Get the transformation matrix order: For each submatrix A ci (i=1, 2, ..., M), obtain the transformation matrix order n of its Hadamard matrix decomposition i . Transformation matrix order n i Equal to the corresponding submatrix A ci The number of rows, that is, n_i = N / M. Construct the Hadamard transformation matrix: According to the obtained transformation matrix order n i , construct the Hadamard transformation matrix H_i of the corresponding order. Hadamard transformation matrix H i is a i H_i is a square matrix whose elements consist of 1 and -1 and satisfy the following conditions: the first row and first column elements of H_i are both 1; i The remaining elements of H satisfy the following recursive relation: i =[H i-1 , H i-1 ;H i-1 , -H i-1 ], where H0 = [1], H i With its transposed matrix The product of is equal to n i Multiply by the identity matrix In i ,Right now The construction of the Hadamard transformation matrix can be achieved by using a recursive algorithm or a table lookup method. Hadamard matrix decomposition: Using the constructed Hadamard transformation matrix H i , for each submatrix A ci Perform Hadamard matrix decomposition. The process of Hadamard matrix decomposition is as follows: Sub-matrix A ci Right multiply the Hadamard transformation matrix H i , get the intermediate matrix B i , that is, B i =A ci ×H i , the intermediate matrix B i Left multiply the Hadamard transformation matrix H i The transposed matrix of Get the decomposed information carrying submatrix D i ,Right now The decomposed information-carrying submatrix Di The dimension of the atomic matrix A ci The same, (N / M) rows and L*T columns. Through Hadamard matrix decomposition, the original information-carrying submatrix A ci Transformed into the decomposed information-carrying submatrix D i . Combine the decomposed sub-matrices: Combine the decomposed information-carrying sub-matrices D1, D2, ..., D M Combine them in the original order to get the decomposed information carrying matrix D. The dimension of the decomposed information carrying matrix D is the same as the original information carrying matrix A. c Same, N rows and L*T columns.

[0014] Furthermore, the constructed Hadamard transformation matrix is ​​used to perform Hadamard matrix decomposition on each sub-matrix, including: partitioning the Hadamard transformation matrix: partitioning the constructed Hadamard transformation matrix H_i into multiple sub-blocks in an alternating manner of rows and columns. The size of the sub-blocks can be selected according to actual needs, and common partitioning methods include 2x2, 4x4, etc. Assume that H i Divided into 4x4 sub-blocks, H i It can be expressed as: Among them, H ijk Indicates H i Recursive one-dimensional Hadamard transform: For each sub-block H ijk Perform a recursive one-dimensional Hadamard transform to obtain the decomposed information-carrying submatrix D i The process of one-dimensional Hadamard transform is as follows: Sub-block H ijk Expand by row (or column) into a one-dimensional vector h ijk ; For a one-dimensional vector h ijk Perform a one-dimensional Hadamard transform to obtain the transformed one-dimensional vector d ijk ; The transformed one-dimensional vector d ijk Rearrange to H ijk Sub-blocks of the same size D ijk. The one-dimensional Hadamard transform can be implemented using the fast Hadamard transform (FHT) algorithm, and its computational complexity is O(nlogn), where n is the transform length. The basic idea of ​​the FHT algorithm is to decompose the one-dimensional Hadamard transform into multiple smaller transforms, and achieve fast calculation through recursion and butterfly operations. The pseudo code of the FHT algorithm is as follows: function FHT(x): n = length(x), if n == 1: return x else: x_even = FHT(x[0:2:n-1]), x_odd = FHT(x[1:2:n]), x_combined = concat(x_even+x_odd, x_even-x_odd), return x_combined. By performing a ijk Recursively applying the FHT algorithm, we can get the corresponding transformed sub-block D ijk . Combine the transformed sub-blocks: Combine the transformed sub-blocks D ijk According to H ijk The same position combination is used to obtain the decomposed information-carrying submatrix D i . D i The dimension of the original information-carrying submatrix A ci Same, (N / M) rows and L*T columns.

[0015] Among them, compared with directly performing Hadamard matrix decomposition on the entire sub-matrix, the method of using sub-block partitioning and recursive one-dimensional transformation has the following advantages: Reduce computational complexity: By decomposing a large matrix into multiple small sub-blocks, the number of matrix multiplications can be reduced, reducing the overall computational complexity. Improve parallelism: Parallel one-dimensional Hadamard transformations can be performed on different sub-blocks, making full use of the parallel computing capabilities of modern processors to accelerate the calculation process. Save storage space: When performing a one-dimensional Hadamard transformation, only the data of the current sub-block needs to be stored, and there is no need to store the entire sub-matrix at the same time, reducing the demand for storage space. Facilitate hardware implementation: The one-dimensional Hadamard transform can be implemented through simple addition, subtraction and shift operations, which is very suitable for hardware implementation. A special Hadamard transform circuit can be designed to further improve the computing efficiency.

[0016] Specifically, the constructed Hadamard transform matrix is ​​divided into multiple Hadamard transform sub-blocks alternately according to rows and columns, the number of rows and the number of columns of the Hadamard transform sub-block are equal, and the number of rows and the number of columns are both non-negative integer powers of 2; a Hadamard transform sub-block obtained by the division is selected, and the row vector of the Hadamard transform sub-block and the column vector of the corresponding sub-matrix obtained by the division are subjected to a one-dimensional Hadamard transform to obtain a one-dimensional transform result; the obtained one-dimensional transform result is transposed to obtain a transposed one-dimensional transform result; the row vector of the transposed one-dimensional transform result and the column vector of the selected Hadamard transform sub-block are subjected to a one-dimensional Hadamard transform to obtain a two-dimensional Hadamard transform result; all the Hadamard transform sub-blocks obtained by the division and the one-dimensional Hadamard transform results of the corresponding sub-matrix are recursively transformed until all sub-blocks are transformed to obtain a decomposed information-carrying sub-matrix.

[0017] Further, in step 5, the reconstruction matrix parameters are optimized, including: initializing the reconstruction matrix parameters: initializing the parameters of the reconstruction matrix R according to the decomposed information carrying matrix D obtained in step 4. The dimension of the reconstruction matrix R is LT rows and N columns, where LT is the number of rows of the decomposed information carrying matrix D, and N is the original information carrying matrix A c The number of rows. Initialize the value range of the elements of the reconstruction matrix R. Usually, the elements can be initialized to random values ​​between -1 and 1. Construct the reconstruction matrix and calculate the product: Use the initialized reconstruction matrix parameters to construct the reconstruction matrix R. Perform a product operation on the decomposed information-carrying matrix D and the reconstruction matrix R to obtain the reconstructed information-carrying matrix A. r , that is, A r =D×R. A r The dimension of the original information-carrying matrix A c Same, N rows and L*T columns. Calculate the mean square error: Calculate the reconstructed information carrying matrix A r The original information carrying matrix A obtained in step 3 c The mean square error (MSE) between the two. The calculation formula of the mean square error is: Compare the mean square error with the threshold: Compare the calculated mean square error MSE with the preset threshold threshold. If the MSE is greater than the threshold, it is considered that the reconstruction error is large and the element values ​​of the reconstruction matrix R need to be adjusted. Use the gradient descent algorithm to optimize the element values ​​of R. The specific steps are as follows: Calculate the partial derivative of the mean square error with respect to each element in R to obtain the gradient matrix G; Update the element values ​​of R according to the gradient matrix G and the learning rate lr, that is, R′=R-lr×G; Use the updated R as the new reconstruction matrix, reconstruct the reconstruction matrix and calculate the product. If the MSE is less than or equal to the threshold, it is considered that the reconstruction error is small enough, and the current reconstruction matrix parameters are used as the optimal parameters. Construct the optimal reconstruction matrix and calculate the product: Use the obtained optimal reconstruction matrix parameters to construct the optimal reconstruction matrix R opt . The decomposed information carrying matrix D and the optimal reconstruction matrix R opt Perform product operation to obtain the optimized reconstructed information carrying matrix A ropt , that is, A ropt =D×R opt . A ropt The dimension of the original information-carrying matrix A c The same is N rows and L*T columns. The optimization of the reconstruction matrix parameters aims to reduce the error between the reconstructed information-carrying matrix and the original information-carrying matrix and improve the accuracy of reconstruction. By introducing the reconstruction matrix, the decomposed information-carrying matrix can be mapped back to the original information space to achieve information reconstruction and recovery. In the optimization process, the mean square error is used as an indicator to measure the reconstruction error. The element values ​​of the reconstruction matrix are continuously adjusted through the gradient descent algorithm to gradually reduce the mean square error until the preset threshold requirement is reached.

[0018] Furthermore, in step 5, the power allocation strategy is optimized, including: calculating the channel capacity: calculating the channel capacity of each subchannel according to the current channel state and noise power. For the i-th subchannel, its channel capacity C i The calculation formula is: C i =B×log2(1+SINR i ), where B is the bandwidth of the subchannel, SINR i is the signal-to-noise ratio of the ith subchannel. Determine the power allocation coefficient: According to the calculated channel capacity of each subchannel, the total power is optimally allocated among the subchannels so that the signal-to-noise ratios of the subchannels are equal, and the power allocation coefficient of each subchannel is determined. Let the total power be P total , the power allocation coefficient of the i-th subchannel is α i , then: P i =α i ×P totalTo make the signal-to-noise ratio of each subchannel equal, the following conditions need to be met: SINR1 = SINR2 = ... = SINR N , where N is the number of subchannels. Combined with the calculation formula of signal-to-noise ratio, we can get: At the same time, the sum of all power allocation coefficients is 1, that is: α1+α2+......+α N =1. According to the above conditions, the optimal power allocation coefficient of each subchannel can be obtained by solving the linear equations. Perform power allocation: According to the determined power allocation coefficient, perform power allocation on each subchannel of the optimized reconstructed information bearing matrix obtained in step 4. Let the optimized reconstructed information bearing matrix be A ropt , whose ith row corresponds to the data of the ith subchannel. For the ith subchannel, the corresponding row vector A ropt (i,:) multiplied by Get the row vector A after power allocation rpower (i,:). to A ropt Power is allocated to each row of the matrix A, and the reconstructed information carrying matrix A after power allocation optimization is obtained. rpower .

[0019] The channel capacity is calculated based on the channel state and noise power, the power allocation coefficient of each subchannel is determined through optimal power allocation, and the power allocation is performed on the optimized reconstructed information bearing matrix to obtain the reconstructed information bearing matrix after power allocation optimization. The optimization of the power allocation strategy aims to dynamically adjust the power allocation of each subchannel according to the channel conditions to achieve the purpose of improving system performance. By making the signal-to-noise ratio of each subchannel equal, the balanced utilization of power resources can be achieved, avoiding the waste of power due to excessive signal strength of some subchannels, and also avoiding the transmission quality affected by too low signal strength of some subchannels. When determining the power allocation coefficient, it is necessary to comprehensively consider factors such as channel gain and noise power, and obtain the optimal power allocation scheme by solving a group of linear equations. This process can use numerical optimization algorithms, such as Lagrange multiplier method, convex optimization algorithm, etc., to efficiently solve the optimal power allocation coefficient.

[0020] Furthermore, in step 7, the final information carrying reconstruction matrix is ​​obtained, including: obtaining channel state information CSI: obtaining the current channel state information CSI through channel estimation technology, including channel gain, phase, delay and other parameters. Channel state information can be obtained through pilot sequences, channel detection and other methods. Common channel estimation algorithms include least squares estimation (LS), minimum mean square error estimation (MMSE) and the like. Calculating the signal-to-interference-plus-noise ratio SINR: calculating the signal-to-interference-plus-noise ratio SINR corresponding to each information carrying reconstruction submatrix based on the obtained channel state information CSI. Constructing a signal-to-interference-plus-noise ratio SINR maximization objective function: constructing a signal-to-interference-plus-noise ratio SINR maximization objective function based on the signal-to-interference-plus-noise ratio SINR of each information carrying reconstruction submatrix. The expression of the objective function is:

[0021] Where M is the total number of information-carrying reconstruction submatrices, w i is the weight coefficient of the ith submatrix. The physical meaning of the objective function is to maximize the weighted sum of the channel capacity of all information-carrying reconstruction submatrices, that is, to maximize the total transmission rate of the system.

[0022] The convex optimization algorithm is used to solve the maximization objective function: Since the objective function is about the weight coefficient w i convex function, so convex optimization algorithms can be used to solve the maximization problem. Commonly used convex optimization algorithms include gradient descent, Newton method, interior point method, etc. These algorithms can efficiently solve the optimal weighting coefficient. The following constraints need to be met during the solution process: weighting coefficient w i The value range is between 0 and 1, that is, 0≤w i ≤1; the sum of all weighted coefficients is 1, that is The optimal weight coefficient w of each information-carrying reconstruction submatrix is ​​obtained by solving the convex optimization algorithm. opt . Calculate the final information-carrying reconstruction matrix: Use the optimal weighting coefficient w opt , reconstruct the submatrix A for each information carrier obtained in step 6 poweri Perform weighted summation to obtain the final information-carrying reconstruction matrix A final The final information-carrying reconstruction matrix A final The calculation formula is: where w opti is the optimal weight coefficient of the i-th submatrix, A poweriis the matrix after power allocation optimization of the ith submatrix. Through the above steps, the calculation of the final information-carrying reconstruction matrix is ​​realized. By obtaining the channel state information, calculating the signal-to-interference-noise ratio, constructing the signal-to-interference-noise ratio maximization objective function, using the convex optimization algorithm to solve the optimal weighting coefficient, and performing weighted summation on each information-carrying reconstruction submatrix, the final information-carrying reconstruction matrix is ​​obtained. The final information-carrying reconstruction matrix integrates the influence of factors such as channel state, interference and noise, and optimizes the transmission performance of the system by maximizing the weighted sum of channel capacity. This process makes full use of the channel state information and convex optimization theory to achieve adaptive optimization of the information-carrying reconstruction matrix.

[0023] Specifically, the decomposed information bearing submatrix is ​​adaptively optimized according to the estimated channel state information CSI, wherein the CSI acquisition scheme is as follows: the transmitting end estimates the channel state information through the pilot symbol before data transmission. Specifically, the transmitting end inserts a known pilot symbol sequence into the data symbol sequence, and the receiving end uses the received pilot symbol sequence and the locally known pilot symbol sequence, and adopts an estimation algorithm such as the least squares method (LS) or the minimum mean square error method (MMSE) to estimate the channel frequency response between the transmitting end and the receiving end, and obtains the estimated channel state information CSI.

[0024] Preferably, in order to further improve the accuracy of channel estimation and reduce pilot overhead, the following improved technical solutions can be adopted: Pilot symbol optimization: In the construction of pilot symbols, a better pilot sequence design method is adopted, such as Zadoff-Chu sequence, Golay sequence, etc. These sequences have good cross-correlation performance and peak-to-average power ratio characteristics, which are conducive to improving the quality of channel estimation. At the same time, the insertion density and power allocation of pilot symbols are optimized to reduce pilot overhead while ensuring estimation performance. Compressed sensing estimation: Using the sparse characteristics of the channel, the channel estimation problem is modeled as a compressed sensing reconstruction problem. The high-dimensional channel vector is mapped to a low-dimensional observation vector through random projection, and then the sparse channel vector is restored using a compressed sensing reconstruction algorithm (such as OMP, CoSaMP, etc.) to achieve high-precision channel estimation. Compared with traditional estimation methods, compressed sensing estimation can significantly reduce pilot overhead and improve spectrum utilization efficiency. Feedback-based adaptive estimation: On the basis of the initial channel estimation, a limited feedback mechanism is introduced to feed back the quantized CSI to the transmitter. The transmitter adjusts the pilot symbol transmission strategy (such as power, density, etc.) according to the feedback CSI to generate a better pilot sequence. The receiver uses the better pilot sequence to further improve the channel estimation accuracy. Through iterative interaction between the transmitter and the receiver, the channel estimation performance is gradually optimized. Channel estimation based on deep learning: Utilize the powerful learning and modeling capabilities of deep neural networks to build an end-to-end channel estimation model. Through offline training of a large amount of channel sample data, the statistical characteristics and mapping relationship of the channel are learned to achieve pilot-free blind estimation or semi-blind estimation of a small number of pilots. Compared with traditional estimation algorithms, deep learning methods have stronger nonlinear fitting capabilities and robustness, and can adapt to complex and changing channel environments.

[0025] Furthermore, the signal-to-interference-to-noise ratio (SINR) is an important indicator for measuring the performance of wireless communication systems, which represents the ratio of the signal power to the interference and noise power of each information-carrying reconstruction submatrix. For the i-th information-carrying reconstruction submatrix, its signal-to-interference-to-noise ratio (SINR) is i The calculation formula is: The meaning of each parameter is as follows: i G represents the transmission power of the i-th information-carrying reconstruction submatrix, that is, the signal power of the i-th submatrix. Transmission power is an important parameter of the communication system, which determines the transmission distance and quality of the signal. j G represents the channel gain of the i-th information-carrying reconstruction submatrix, reflecting the degree of amplification or attenuation of the channel on the i-th submatrix signal. The channel gain is affected by factors such as signal frequency, transmission distance, and environmental obstacles. j represents the channel gain of the jth information-bearing reconstruction submatrix, where j is the number of the interference submatrix, ranging from 1 to M, and is not equal to i. jiIt represents the channel transmission coefficient of the jth information-carrying reconstruction submatrix to the ith information-carrying reconstruction submatrix, reflecting the interference degree of the jth submatrix signal on the ith submatrix after being transmitted through the channel. The channel transmission coefficient depends on the fading characteristics and multipath effects of the channel. i Indicates the noise power of the i-th information-carrying reconstruction submatrix. The noise comes from electronic components, thermal noise, etc., which will affect the detection and recovery of the signal. M: Indicates the total number of information-carrying reconstruction submatrices. When calculating SINR, the numerator P i ×G i represents the received signal power of the ith submatrix, which is the product of the transmit power and the channel gain. The denominator includes two terms: interference power and noise power. Interference power It represents the interference caused by the signals of other sub-matrices to the i-th sub-matrix after being transmitted through the channel. It is the sum of the products of the transmission power of the interference sub-matrix, the channel gain and the channel transmission coefficient. i Indicates the noise impact on the ith submatrix, which is a constant. Divide the received signal power by the sum of the interference and noise power to obtain the SINR of the ith submatrix. i . SINR i The higher it is, the better the signal quality of the i-th sub-matrix is ​​and the less affected by interference and noise.

[0026] Furthermore, in step 8, the Zadoff-Chu sequence is a commonly used pilot sequence with good correlation and low peak-to-average ratio characteristics, which can effectively realize channel estimation and synchronization functions. In step 8, in order to further improve the performance of pilot symbols, an extended Zadoff-Chu sequence is used to construct pilot symbols. The generation process of the extended Zadoff-Chu sequence is as follows: select multiple different root sequence numbers, each root sequence number corresponds to a Zadoff-Chu sequence. The selection of the root sequence number should meet the following conditions: the root sequence number and the sequence length are coprime, that is, their greatest common divisor is 1; different root sequence numbers are different from each other to ensure low correlation between the generated Zadoff-Chu sequences. For each selected root sequence number, a corresponding Zadoff-Chu sequence is generated. The generation formula of the Zadoff-Chu sequence is: When the sequence length N is an odd number: k=0,1,........,N-1,when the sequence length N is an even number: k=0,1,........,N-1,where u is the root sequence number, N is the sequence length, and j is the imaginary unit. The generated multiple Zadoff-Chu sequences are combined in a multiplexing manner to form an extended Zadoff-Chu sequence. The multiplexing manner may be time division multiplexing, frequency division multiplexing, or code division multiplexing. Time division multiplexing: Zadoff-Chu sequences with different root sequence numbers are arranged in sequence in the time domain to form an extended sequence; frequency division multiplexing: Zadoff-Chu sequences with different root sequence numbers are interleaved in the frequency domain, and each subcarrier carries symbols of different sequences; code division multiplexing: Zadoff-Chu sequences with different root sequence numbers are multiplied with different orthogonal codes to generate orthogonal extended sequences. The generated extended Zadoff-Chu sequence is used as a pilot symbol and inserted into a specified position in the information carrying reconstruction matrix for functions such as channel estimation and synchronization. Compared with a single Zadoff-Chu sequence, the extended Zadoff-Chu sequence has the following advantages: Higher metric resolution: By combining multiple Zadoff-Chu sequences with different root numbers, the length of the extended sequence can be significantly increased, thereby improving the metric resolution of the pilot symbol, which is conducive to more accurate channel estimation and synchronization. Stronger interference suppression capability: Zadoff-Chu sequences with different root numbers have very low cross-correlation, and the interference between different components in the extended sequence is small, which can effectively suppress multiple access interference and inter-symbol interference. More flexible resource allocation: By selecting different multiplexing modes and parameter settings, the resources of pilot symbols can be flexibly allocated according to system requirements and channel conditions to optimize system performance.

[0027] Further, in step 8, the synchronization sequence is constructed using an extended Barker code, and the extended Barker code is composed of an original Barker code and a cyclic shift sequence. Specifically, the Barker code is a commonly used synchronization sequence with good autocorrelation characteristics, and can achieve accurate frame synchronization and symbol synchronization. In step 8, in order to further improve the performance of the synchronization sequence, the extended Barker code is used to construct the synchronization sequence. The generation process of the extended Barker code is as follows: select an original Barker code as a basic sequence. The Barker code is a binary sequence whose autocorrelation function has a low sidelobe value at a non-zero offset. Commonly used Barker code lengths are 2, 3, 4, 5, 7, 11 and 13. Generate a cyclic shift sequence. The cyclic shift sequence is a set of sequences obtained by performing a cyclic shift operation on the original Barker code. Assuming the length of the original Barker code is L, L-1 cyclic shift sequences can be generated. The method for generating the i-th cyclic shift sequence is: cyclically shift the original Barker code to the right by i positions, and the value of i ranges from 1 to L-1. For example, for a Barker code of length 5 [1, 1, 1, -1, 1], its cyclic shift sequence includes: [1, 1, -1, 1, 1], [1, -1, 1, 1, 1], [-1, 1, 1, 1, 1], [1, 1, 1, 1, -1]. The original Barker code is combined with the cyclic shift sequence to form an extended Barker code. The combination can be in the form of series connection, interleaving or parallel connection. Series connection: The original Barker code and the cyclic shift sequence are connected in series in a certain order to form an extended sequence; Interleaving: The original Barker code and the cyclic shift sequence are interleaved according to certain rules, such as odd-even interleaving, symbol interleaving, etc.; Parallel connection: The original Barker code and the cyclic shift sequence are transmitted in parallel, and then combined and processed at the receiving end. The generated extended Barker code is used as a synchronization sequence and inserted into the specified position in the information carrying reconstruction matrix for functions such as frame synchronization, symbol synchronization and channel estimation. Compared with the original Barker code, the extended Barker code has the following advantages: Longer sequence length: By combining the original Barker code and the cyclic shift sequence, the length of the extended Barker code can be significantly increased, providing more synchronization information, which is conducive to more accurate and stable synchronization. Stronger synchronization performance: The extended Barker code retains the excellent autocorrelation characteristics of the original Barker code, and at the same time introduces more synchronization information through the cyclic shift sequence, which can improve the accuracy and robustness of synchronization. More flexible sequence design: By selecting different combination methods and parameter settings, the extended Barker code can be flexibly designed to optimize the synchronization performance according to the system requirements and channel conditions.

[0028] Beneficial effects: Compared with the prior art, the advantages of this application are:

[0029] The use of polarization codes for channel coding fully utilizes the channel polarization effect and channel matching characteristics of polarization codes, achieves better coding gain under low signal-to-noise ratio conditions, and effectively reduces channel coding complexity and power consumption.

[0030] Logistic chaotic mapping is introduced to construct the information carrying matrix, and the pseudo-randomness and robustness of chaotic sequences are utilized to enhance the confidentiality and anti-interference ability of the system and reduce the risk of signal interception and decryption.

[0031] Hadamard matrix decomposition is used to realize information bearing matrix partition. Through recursive one-dimensional Hadamard transformation, the matrix decomposition operation is simplified, and the computational complexity and hardware implementation cost are reduced.

[0032] Based on channel state information and mean square error criterion, the gradient descent algorithm is used to optimize the adaptive reconstruction matrix and dynamically adjust the reconstruction matrix parameters to reduce the signal reconstruction error and improve the signal recovery quality.

[0033] According to the channel capacity and signal-to-noise ratio, the power allocation strategy is optimized through a convex optimization algorithm, and optimal power allocation is performed among various sub-channels, thereby improving the spectrum efficiency and energy utilization of the system.

[0034] A weighted coefficient optimization objective function based on signal-to-interference-noise ratio maximization is constructed, and a convex optimization algorithm is used to solve the optimal weighted coefficient. The final information-carrying reconstruction matrix is ​​obtained through weighted summation, which effectively suppresses interference between sub-channels and improves detection performance.

[0035] The extended Zadoff-Chu sequence is used to construct the pilot symbol. By combining multiple Zadoff-Chu sequences with different root numbers, better cross-correlation performance and more pilot sequence options are obtained, which is beneficial to improving the channel estimation quality.

[0036] The extended Barker code is used to construct the synchronization sequence. By combining the original Barker code and its cyclic shift sequence, a longer sequence length and better correlation characteristics are obtained, which is beneficial to improving the synchronization effect and stability of the system.

[0037] A frame structure including pilot, control information, synchronization sequence and data symbols is designed, and adaptive precoding optimization is performed in combination with channel state information to generate a time domain transmission symbol sequence, thereby reducing channel estimation error and rate loss.

[0038] Through digital-to-analog conversion and RF front-end processing, the optimized signal is transmitted to the wireless channel, and a power amplifier is used to amplify the signal power, thereby improving the signal transmission distance and coverage range. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is an exemplary flow chart of a mixed signal low power consumption processing method based on a multi-carrier receiver according to some embodiments of this specification;

[0040] Figure 2 is an exemplary flow chart of obtaining an original information bearing matrix according to some embodiments of this specification;

[0041] Figure 3 is an exemplary flow chart of obtaining a decomposed information bearing matrix according to some embodiments of this specification;

[0042] Figure 4 This is an exemplary flow chart of obtaining an optimized reconstructed information bearing matrix according to some embodiments of this specification. DETAILED DESCRIPTION

[0043] The method and system provided in the embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0044] like Figure 1As shown, this embodiment provides a mixed signal low-power processing method based on a multi-carrier receiver, including the following steps: Step 1, obtaining an information bit sequence to be transmitted and preprocessing it; the preprocessing includes serial-to-parallel conversion and channel coding; wherein the channel coding uses polar code Polar Code; Step 2, mapping the preprocessed information bit sequence to a filter group multi-carrier modulation symbol set FBMC to obtain a modulation symbol sequence; Step 3, using Logistic mapping to generate a chaotic sequence, cross-product the obtained modulation symbol sequence with the generated chaotic sequence, and constructing an original information carrying matrix; Step 4, according to the number M of subcarriers of a differential chaotic shift keying DCSK system, the obtained original information carrying matrix is ​​divided into M sub-matrices, and each sub-matrix is ​​subjected to matrix decomposition processing to obtain a decomposed information carrying sub-matrix; Step 5, adaptively optimizing the obtained decomposed information carrying sub-matrix; the adaptive optimization includes reconstruction matrix parameter optimization and power allocation strategy optimization; Step 6, reconstructing the obtained decomposed information carrying sub-matrix using the reconstruction matrix parameters and the power allocation strategy, and obtaining to the information carrying reconstruction sub-matrix; step 7, weighted summing the obtained information carrying reconstruction sub-matrices to obtain the final information carrying reconstruction matrix; step 8, inserting pilot symbols and control information into the obtained information carrying reconstruction matrix; setting a frame structure, the frame structure includes the arrangement of pilot, control information, synchronization sequence and data symbols; step 9, precoding the information carrying reconstruction matrix inserted with pilot symbols and control information to obtain a precoded information carrying reconstruction matrix; the precoding processing includes adaptive optimization according to channel state information CSI; converting the precoded information carrying reconstruction matrix into a time domain transmission symbol sequence according to the set frame structure; step 10, transmitting the generated time domain transmission symbol sequence to the wireless channel after digital-to-analog conversion and RF front-end processing; wherein the RF front-end includes a power amplifier.

[0045] Step 1: Get the information bit sequence {b1, b2, ..., bK} to be transmitted and preprocess it. Preprocessing includes serial-to-parallel conversion and channel coding. Serial-to-parallel conversion converts the serial bit sequence into a parallel bit sequence. Channel coding uses polarization code to encode the parallel bit sequence into a polarization code bit sequence {c1, c2, ..., cN}, where K is the information bit length and N is the encoded codeword length. Specifically, assume that the information bit sequence to be transmitted is {b1, b2, ..., bK}, where K is the length of the information bit. First, serial-to-parallel conversion is used to convert the serial bit sequence into a parallel bit sequence. Specifically, the parallelism is set to P, and the K serial bits are evenly divided into P subsequences, each subsequence containing L = K / P bits. Then, the P subsequences are input in parallel into the channel coding module in the next step. Channel coding uses polarization code for encoding. Polar code is a linear block code based on channel polarization phenomenon. Through recursive channel combination and splitting, polarization subchannels with a capacity close to the channel capacity are constructed, and information bits are sent on these subchannels to achieve reliable communication. Let the parallel bit sequence be {d1, d2, ......., dL}. The polar code encoding process is as follows:

[0046] Channel polarization: Perform Kronecker matrix product operation on the parallel bit sequence {d1, d2, ......., dL} to obtain the polarized bit sequence {u1, u2, ......., uN}. The Kronecker matrix is ​​an N-order Hadamard matrix, where N = 2^n and n is a positive integer. The Kronecker matrix product operation is expressed as: [u1, u2, ......., uN]^T = G_N×[d1, d2, ......., dL, 0, ......., 0]^T, where G_N is an N-order Kronecker matrix, T represents the transposition operation, and 0 represents the zero-padding bit, so that the input bit sequence length is extended to N. Information bit selection: According to the channel polarization effect, the K bit positions with the highest reliability are selected as information bits, and the remaining NK bit positions are frozen bits and fixed to 0. The selection of information bit positions can be pre-calculated by Monte Carlo simulation or Gaussian approximation method. Bit insertion: insert the parallel bit sequence {d1, d2, ...., dL} into the selected K information bit positions, and fill the remaining NK frozen bit positions with zeros to obtain the polarization code bit sequence {c1, c2, ...., cN}. After the above polarization code encoding process, the parallel bit sequence {d1, d2, ...., dL} is encoded into the polarization code bit sequence {c1, c2, ...., cN}, where the coding rate R = K / N. By reasonably selecting the information bit position and coding rate, the polarization code can achieve reliable communication at a low signal-to-noise ratio. The pre-calculated information bit positions can be stored in a lookup table (LUT) to reduce the real-time computing overhead. In addition, the Kronecker matrix product operation can be recursively implemented by butterfly operation to reduce the computational complexity.

[0047] Step 2: Map the polar code bit sequence {c1, c2, ..., cN} to the FBMC modulation symbol set to obtain the modulation symbol sequence {s1, s2, ..., sM}, where M is the number of modulation symbols. The modulation symbol set adopts the OQAM-FBMC modulation format, with orthogonal amplitude modulation between symbols and frequency shift keying modulation within symbols.

[0048] Specifically, assume that the polar code bit sequence is {c1, c2, ..., cN}, where N is the codeword length. FBMC modulation adopts the OQAM-FBMC modulation format, with quadrature amplitude modulation (QAM) between symbols and frequency shift keying (FSK) modulation within symbols. The modulation process is as follows: Symbol mapping: The polar code bit sequence is symbol mapped according to the preset QAM constellation diagram to obtain a complex modulation symbol sequence {d1, d2, ..., dM}, where M is the number of modulation symbols. Taking 16QAM as an example, every 4 bits are mapped to a complex modulation symbol, where the real part and the imaginary part are respectively {-3, -1, 1, 3}. Then M = N / 4, and the modulation symbol sequence length is M. OQAM preprocessing: The complex modulation symbol sequence {d1, d2, ..., dM} is preprocessed with quadrature amplitude modulation (OQAM). Specifically, the imaginary part of the even-numbered modulation symbol is multiplied by j, and the real part of the odd-numbered modulation symbol is multiplied by j to obtain the preprocessed modulation symbol sequence {d'1, d'2,......., d'M}. Wherein, j is the imaginary unit, j^2=-1. Frequency shift keying modulation: The preprocessed modulation symbol sequence {d'1, d'2,......., d'M} is subjected to frequency shift keying (FSK) modulation. Assuming that the frequency interval of FSK modulation is Δf, and each modulation symbol contains K subcarriers, the K subcarrier frequencies corresponding to the mth modulation symbol are: f_m, k=f_c+(m-1)*Δf+(k-1-K / 2)*Δf / K, k=1, 2,......., K, where f_c is the center frequency, m=1, 2,......., M, k=1, 2,......., K. The pre-processed modulation symbol d'm is mapped to the corresponding K subcarriers to obtain the FSK modulated symbol s_m, k. FBMC synthesis: The FSK modulated symbol {s_m, k} is synthesized by FBMC to obtain the final FBMC modulation symbol sequence {s1, s2, ..., sM}. Specifically, the frequency domain {s_m, k} is converted to the time domain by IFFT transformation, and then pulse shaping is performed by the prototype filter group to obtain the FBMC modulation symbol sequence in the time domain. The prototype filter group usually adopts Hermite function or PHYDYAS filter to achieve good time-frequency locality and subcarrier orthogonality. The OQAM-FBMC modulation format is used to map the polar code bit sequence to the modulation symbol sequence. Among them, QAM modulation is used between symbols to ensure spectrum utilization; FSK modulation is used within the symbol to improve the anti-interference ability. Through OQAM preprocessing and prototype filtering, the orthogonality between subcarriers is achieved and the inter-symbol interference is suppressed. At the same time, FBMC synthesis uses IFFT transformation to reduce the implementation complexity of the transmitter.Compared with traditional OFDM modulation, OQAM-FBMC has lower out-of-band leakage and peak-to-average power ratio while maintaining spectrum utilization, and is more suitable for low-power and high-reliability communication scenarios.

[0049] Figure 2 This is an exemplary flow chart for obtaining the original information carrying matrix according to some embodiments of this specification, step 3: Use Logistic mapping to generate a chaotic sequence {x1, x2,......., xL}, where L is the length of the chaotic sequence. Perform element-by-element cross product operation on the modulation symbol sequence {s1, s2,......., sM} and the chaotic sequence {x1, x2,......., xL} to construct the original information carrying matrix S. The dimension of the original information carrying matrix S is M×L, where the matrix element Si, j = si*xj (i = 1, 2,......., M; j = 1, 2,......., L). Specifically, chaotic sequence generation: Use Logistic mapping to generate a chaotic sequence {x1, x2,......., xL}, where L is the length of the chaotic sequence. Logistic mapping is a common one-dimensional chaotic mapping, and its iteration formula is: x_n+1=μ*x_n*(1-x_n), n=0, 1, 2, ..., where x_n is the chaotic value of the nth iteration, and μ is the control parameter with a value range of (0, 4]. When μ=4, the Logistic mapping is in a completely chaotic state, and the generated sequence has good randomness and unpredictability. Assume that the initial value x_0∈(0, 1), and the chaotic sequence {x1, x2, ..., xL} is calculated by the iteration formula. In order to improve the randomness of the chaotic sequence, the generated sequence can be binarized to obtain a binary chaotic sequence {b1, b2, ..., bL}. The binarization process can use the threshold comparison method, for example: b_i=0, if x_i<0.5, b_i=1, if x_i≥0.5, Where i = 1, 2, ......., L. After binarization, the resulting chaotic sequence {b1, b2, ......., bL} consists of 0 and 1, which has stronger randomness and security. Element-by-element cross product operation: Perform element-by-element cross product operation on the modulation symbol sequence {s1, s2, ......., sM} and the chaotic sequence {x1, x2, ......., xL} to construct the original information carrying matrix S. The dimension of the original information carrying matrix S is M×L, where the calculation formula of the matrix element S_i, j is: S_i, j = s_i*x_j, i = 1, 2, ......., M; j = 1, 2, ......., L, where s_i is the i-th modulation symbol and x_j is the j-th element of the chaotic sequence. Through the element-by-element cross product operation, the modulation symbol sequence is combined with the chaotic sequence to generate the original information carrying matrix S.

[0050] Figure 3 This is an exemplary flow chart for obtaining a decomposed information-carrying matrix according to some embodiments of the present specification. Step 4: According to the number of subcarriers M of the DCSK system, the original information-carrying matrix S is divided into M sub-matrices {S1, S2, ...., SM}. Perform Hadamard matrix decomposition on each submatrix Si (i = 1, 2, ...., M) to obtain M decomposed information-carrying sub-matrices {Hi, Bi} (i = 1, 2, ...., M), where Hi is an orthogonal matrix and Bi is a diagonal matrix. Specifically, matrix partitioning: According to the number of subcarriers M of the DCSK system, the original information-carrying matrix S is divided into M sub-matrices {S1, S2, ...., SM}. Assume that the dimension of the original information-carrying matrix S is N×L, where N is the number of modulation symbols and L is the length of the chaotic sequence. The dimension of each submatrix is ​​(N / M)×L, that is: S1=S[1:(N / M), :], S2=S[(N / M+1):2(N / M), :], ...., SM=S[(M-1)(N / M)+1:N, :]. Among them, S[a:b, :] means extracting all elements from the ath row to the bth row of the matrix S. Through matrix partitioning, the original information-bearing matrix S is divided into M submatrices, each of which corresponds to a DCSK subcarrier. Hadamard matrix decomposition: Perform Hadamard matrix decomposition on each submatrix Si (i=1,2,.......,M). The Hadamard matrix is ​​an orthogonal matrix whose elements only contain +1 and -1, and any two rows or columns are orthogonal. Let the dimension of the i-th submatrix Si be P×L, where P=N / M. Perform Hadamard matrix decomposition on Si and obtain: Si = Hi × Bi, where Hi is a P × P Hadamard matrix and Bi is a P × L diagonal matrix. The Hadamard matrix Hi satisfies: Hi × Hi^T = P × I_P, where Hi^T is the transposed matrix of Hi and I_P is the P × P unit matrix. The main diagonal elements of the diagonal matrix Bi are the singular values ​​of Si, and the other elements are 0. Through Hadamard matrix decomposition, each submatrix Si is decomposed into the product of the orthogonal matrix Hi and the diagonal matrix Bi, and M decomposed information-bearing submatrices {Hi, Bi} (i = 1, 2, ...., M) are obtained.

[0051] The submatrix Si is recursively decomposed to better capture the local structure and correlation of the signal through multi-level decomposition, thereby improving the information carrying capacity and transmission efficiency. The dimension of the i-th submatrix Si is P×L, where P=N / M. Si is subjected to K-level Hadamard matrix decomposition, and the recursive process is as follows: Initialization: The submatrix Si is regarded as the input matrix of the 0th level decomposition, denoted as Si^((0))=Si. K-th level decomposition (k=1, 2,......., K): a. Divide the matrix Si^((k-1)) obtained by the k-1th level decomposition into Qk sub-blocks, and the dimension of each sub-block is (P / Qk)×(L / Qk). Qk is the number of sub-blocks of the k-th level decomposition, which can be adaptively selected according to the statistical characteristics and correlation of the signal. b. Perform Hadamard matrix decomposition on each sub-block Si,j^((k-1))(j=1,2,.......,Qk) to obtain: Si,j^((k-1))=Hi,j^((k))×Bi,j^((k)), where Hi,j^((k)) is a Hadamard matrix of (P / Qk)×(P / Qk), and Bi,j^((k)) is a diagonal matrix of (P / Qk)×(L / Qk). c. According to the statistical characteristics and correlation of the signal, select the optimal Hadamard matrix Hi,j^((k)) for each sub-block for decomposition. Adaptive selection criteria can be used, such as minimum mean square error (MMSE), maximum mutual information (MI), etc. d. Reorganize the sub-blocks obtained from the k-th level decomposition into the matrix Si^((k)) as the input for the next level of decomposition. Termination condition: After completing the K-level decomposition, the final decomposition result is obtained: Si = Hi^((1))×Hi^((2))×...×Hi^((K))×Bi^((K)), where Hi^((k)) is the Hadamard matrix of the k-th level decomposition, and Bi^((K)) is the diagonal matrix of the K-th level decomposition. Through multi-level Hadamard matrix decomposition, the submatrix Si is recursively decomposed into the product of multiple local Hadamard matrices and diagonal matrices. Each level of decomposition adaptively selects the optimal Hadamard matrix based on the statistical characteristics and correlation of the signal to better match the local signal structure.

[0052] Figure 4This is an exemplary flow chart for obtaining an optimized reconstructed information-bearing matrix according to some embodiments of this specification, step 5: adaptively optimize the decomposed information-bearing submatrix {Hi, Bi} (i = 1, 2, ...., M) according to the channel state information CSI estimated by the pilot symbol. Adaptive optimization includes reconstruction matrix parameter optimization and power allocation strategy optimization. The reconstruction matrix parameter optimization uses a gradient descent algorithm to minimize the mean square error between the information-bearing reconstruction matrix and the original information-bearing matrix to obtain the optimal reconstruction matrix parameters. The power allocation strategy optimization optimally allocates total power among the subchannels according to the channel capacity and the signal-to-noise ratio to obtain the reconstructed information-bearing submatrix {Hoi, Boi} (i = 1, 2, ...., M) after power allocation optimization.

[0053] Specifically, a. Define the optimization objective: minimize the mean square error (MSE) between the information-carrying reconstruction matrix and the original information-carrying matrix. Suppose the original information-carrying submatrix is ​​Si, and the reconstructed information-carrying submatrix is ​​Sri, then the optimization objective can be expressed as: minMSE=E[||Si-Sri||^2], where E[x] represents the mathematical expectation, and ||x|| represents the Frobenius norm of the matrix. b. Parameterize the reconstruction matrix: express the reconstructed matrix in a parameterized form, such as Sri=f(Hi, Bi, θ), where θ is the parameter vector to be optimized. The parameterized form can use linear combinations, nonlinear transformations and other methods to enhance the expressive power of the reconstructed matrix. c. Gradient descent optimization: Use the gradient descent algorithm to optimize the parameter θ. According to the chain rule, calculate the gradient of MSE with respect to θ: Then, the parameter θ is iteratively updated to move in the direction of the negative gradient until it converges to the optimal value: Where t is the number of iterations and η is the learning rate. d. Obtain the optimal reconstruction matrix: Substitute the optimized parameter θ into the parameterized form of the reconstruction matrix to obtain the optimal reconstruction matrices Hoi and Boi, so that the MSE is minimized. Power allocation strategy optimization: Define the optimization goal: maximize the channel capacity of the system while satisfying the total power constraint. Let the signal-to-noise ratio of the i-th subchannel be SNRi, the power allocated to the i-th subchannel be Pi, and the total power be Ptotal. The optimization goal can be expressed as: maxC = ∑log2(1+SNRi·Pi), st∑Pi≤Ptotal, Pi≥0, where ∑ represents the sum of all subchannels. Water filling algorithm: The water filling algorithm is used to optimize the power allocation. The basic idea of ​​the water filling algorithm is to regard the total power as a certain amount of water and each subchannel as a container of different depths. The water will automatically flow to the container with a larger depth until the water surface height of all containers is equal. Calculate the effective signal-to-noise ratio of each subchannel: SNRi_eff = SNRi / (Hoi^2*Boi^2), sort the subchannels from large to small according to the effective signal-to-noise ratio, initialize the water level Level to 0, and allocate power Pi to 0. Starting from the subchannel with the largest effective signal-to-noise ratio, allocate power one by one: Pi = max(0, Level-1 / SNRi_eff), update the remaining total power: Ptotal = Ptotal-Pi, update the water level: Level = (Ptotal+∑1 / SNRi_eff) / (number of remaining subchannels), and repeat the previous step until all subchannels have allocated power. Obtain the reconstructed information-bearing submatrix after power allocation optimization: Substitute the optimal power allocation result Pi into the reconstruction matrix to obtain the reconstructed information-bearing submatrices Hoi and Boi after power allocation optimization.

[0054] Step 6: Reconstruct the decomposed information-carrying submatrices {Hi, Bi} (i = 1, 2, ...., M) using the optimized reconstruction matrix parameters and power allocation strategy to obtain M information-carrying reconstruction submatrices {Hri, Bri} (i = 1, 2, ...., M). Specifically, for the i-th information-carrying submatrix {Hi, Bi}, use the reconstruction matrix parameters θi optimized in step 5 to construct the reconstruction matrices Hri and Bri. The method of constructing the reconstruction matrix depends on the choice of parameterization form. For example, if the parameterization form of linear combination is adopted: Hri = ∑θi, j*Hj, Bri = ∑θi, k*Bk, where Hj and Bk are predefined basis matrices, and θi, j and θi, k are corresponding combination coefficients. Substitute the optimized parameters θi into the above equations to calculate the reconstruction matrices Hri and Bri. For the i-th information-carrying submatrix, power scale the reconstruction matrices Hri and Bri according to the power allocation result Pi optimized in step 5. The power scaling factor root Pi scales the elements of the reconstruction matrix in equal proportion: Hri = root Pi*Hri, Bri = root Pi*Bri. The power-scaled reconstruction matrices Hri and Bri satisfy the allocated power budget while maintaining the original matrix structure and characteristics. a. Combine the power-scaled reconstruction matrices Hri and Bri to obtain the i-th information-carrying reconstruction submatrix: {Hri, Bri}. b. Reconstruct all M information-carrying submatrices to obtain M information-carrying reconstruction submatrices: {Hr1, Br1}, {Hr2, Br2},......., {HrM, BrM}.

[0055] Step 7: According to the signal to interference plus noise ratio SINR, construct the SINR maximization objective function, and use the convex optimization algorithm to solve the optimal weighting coefficients {w1, w2, ..., wM}. Perform weighted summation on the M information carrying reconstruction sub-matrices {Hri, Bri} (i = 1, 2, ..., M) to obtain the final information carrying reconstruction matrix Sr. Specifically, for the i-th information carrying reconstruction sub-matrix, its SINR expression is: SINRi = Pi*Gi / [∑(j = 1toM, j≠i)Pj*Gj*|Hji|^2+Ni], where Pi represents the transmit power of the i-th sub-matrix, Gi represents the channel gain of the i-th sub-matrix, Gj represents the channel gain of the j-th sub-matrix, Hji represents the channel transmission coefficient of the j-th sub-matrix to the i-th sub-matrix, and Ni represents the noise power received by the i-th sub-matrix. The physical meaning of the SINR expression is the ratio of the signal power of the i-th sub-matrix to the interference and noise power it receives. The numerator represents the signal power of the i-th submatrix, and the denominator represents the sum of the interference power and noise power from other submatrices. The channel gains Gi and Gj represent the attenuation and amplification of the signal during transmission, which are related to the channel environment and distance. The channel transmission coefficient Hji represents the change of the signal of the j-th submatrix when it is transmitted to the i-th submatrix, including attenuation, phase shift, etc.

[0056] Define the weighting coefficient wi to represent the weight of the i-th information-carrying reconstruction submatrix, ranging from 0 to 1. The weighting coefficient reflects the contribution of each submatrix to the final reconstruction matrix. Construct the SINR maximization objective function: maximize∑(i=1toM)[log2(1+SINRi)]*wi, where M represents the total number of information-carrying reconstruction submatrices, and i represents the number of the submatrix, ranging from 1 to M. The physical meaning of the objective function is to maximize the sum of the weighted SINRs of all submatrices. Take the logarithm of the SINR and convert it into the form of channel capacity, so that the objective function has a clearer communication meaning. The introduction of the weighting coefficient wi allows the SINR contribution of different submatrices to be flexibly adjusted. The SINR maximization objective function is transformed into a convex optimization problem: maximize∑(i=1toM)[log2(1+SINRi)]*wi, subjectto∑(i=1toM)wi=1, 0≤wi≤1, where the objective function is the sum of weighted SINRs, and the constraints are that the sum of weighted coefficients is 1, and each weighted coefficient is between 0 and 1. The above optimization problem is solved by a convex optimization algorithm such as the interior point method to obtain the optimal weighted coefficients {w1, w2,......., wM}. The interior point method constructs a barrier function to transform the constraints into part of the objective function, and gradually approaches the optimal solution by iteratively updating the variables. The information-bearing reconstruction submatrix is ​​weighted and summed using the optimal weighted coefficients: Sr=∑(i=1toM)wi*{Hri, Bri}, where Sr represents the final information-bearing reconstruction matrix, and {Hri, Bri} represents the i-th information-bearing reconstruction submatrix. The weighted summation result Sr integrates the information of each submatrix and takes into account their SINR contributions. Sr, as the final reconstruction matrix, will be used for subsequent signal detection and decoding.

[0057] Step 8: Insert pilot symbols and control information into the information bearing reconstruction matrix Sr. The pilot symbols are constructed using an extended Zadoff-Chu sequence and are generated by combining multiple Zadoff-Chu sequences with different root numbers. The control information includes transmission parameters such as data rate, modulation order, and coding type. Set the frame structure, which defines the number and arrangement of pilot symbols, control information, synchronization sequences, and data symbols. The synchronization sequence is constructed using an extended Barker code and is generated by combining the original Barker code and its cyclic shift sequence. Specifically, the construction of pilot symbols: The pilot symbols are constructed using an extended Zadoff-Chu sequence. The Zadoff-Chu sequence is a commonly used pilot sequence with good correlation properties and low peak-to-average ratio characteristics. The pilot symbols are generated by combining multiple Zadoff-Chu sequences with different root numbers. By setting different root numbers, multiple orthogonal pilot sequences can be generated to distinguish different information bearers. The length and number of pilot symbols are determined according to the requirements of channel estimation and the design of the frame structure. Pilot symbols are usually inserted into data symbols periodically to facilitate channel estimation and synchronization tracking by the receiver. Insertion of control information: Control information includes transmission parameters such as data rate, modulation order, and coding type, which are used to guide the demodulation and decoding operations of the receiver. The control information is encoded and inserted into the frame structure, usually placed at a fixed position at the head or tail of the frame so that the receiver can quickly obtain it. The coding of control information uses error correction coding technology, such as convolutional coding and LDPC coding, to improve the reliability of control information. Design of frame structure: Set the frame structure to define the number and arrangement of pilot symbols, control information, synchronization sequences, and data symbols. The design of the frame structure needs to consider factors such as the transmission requirements, channel conditions, and delay requirements of the system. Reasonable frame structure design can improve transmission efficiency, reduce overhead, and enhance robustness. Common frame structures include fixed frame structure and flexible frame structure. The length and position of each part of the fixed frame structure are fixed, which is suitable for stable transmission environments. The flexible frame structure can dynamically adjust the length and position of each part according to transmission requirements to adapt to changing transmission environments. Construction of synchronization sequence: The synchronization sequence is used for frame synchronization and symbol synchronization to help the receiver determine the starting position of the frame and the symbol boundary. The synchronization sequence is constructed using the extended Barker code, which is a binary sequence with good autocorrelation properties. By combining the original Barker code and its cyclic shift sequence, multiple orthogonal synchronization sequences can be generated. The synchronization sequence is usually inserted at the starting position of the frame, and the length is determined according to the synchronization requirements and the design of the frame structure. Insertion and framing: The constructed pilot symbols, control information, synchronization sequences, and data symbols are inserted and framed according to the designed frame structure. The insertion process needs to ensure that the position and length of each part match the definition of the frame structure to avoid conflicts or omissions. The framed signal is used as the output of the transmitter and is transmitted to the channel after modulation, amplification, and other processing.

[0058] Step 9: Precode the information-carrying reconstruction matrix Sr after inserting the pilot and control information. Precoding adaptively optimizes the precoding matrix through CSI and multiplies it with Sr. According to the set frame structure, the precoded matrix is ​​converted into a time domain transmission symbol sequence {t1, t2, ......., tP}, where P is the number of transmission symbols. Specifically, optimization of the precoding matrix: The purpose of precoding is to preprocess the signal at the transmitter to adapt to the channel conditions and improve the transmission quality and spectrum efficiency. The precoding matrix is ​​adaptively optimized through channel state information (CSI). CSI includes parameters such as channel response and signal-to-interference-to-noise ratio, which reflects the transmission characteristics of the channel. Common precoding optimization criteria include maximizing channel capacity, minimizing bit error rate, maximizing signal-to-interference-to-noise ratio, etc. According to different optimization criteria, different precoding algorithms can be used, such as singular value decomposition (SVD) precoding, minimum mean square error (MMSE) precoding, vector perturbation (VP) precoding, etc. The dimension of the precoding matrix matches the dimension of the information-carrying reconstruction matrix Sr, which is usually the product of the number of transmitting antennas and the number of information streams. Precoding processing: The optimized precoding matrix is ​​recorded as W, and its dimension is M×M, where M is the dimension of Sr. Precoding processing is performed on Sr, that is, Sr is multiplied by the precoding matrix W: X=W×Sr, where X is the precoded matrix and has the same dimension as Sr. Precoding processing is equivalent to linear transformation of Sr, mapping it to the optimal transmission subspace to adapt to channel conditions. Time domain conversion: According to the set frame structure, the precoded matrix X is converted into a time domain transmission symbol sequence {t1, t2,......., tP}, where P is the number of transmission symbols. Time domain conversion is usually implemented by inverse Fourier transform (IFFT), which converts the precoding matrix in the frequency domain into a transmission symbol sequence in the time domain. The parameter setting of IFFT is related to the frame structure, such as the number of IFFT points, cyclic prefix length, etc. Appropriate parameter settings can improve the spectrum efficiency and anti-interference ability of the system. The transmission symbol sequence after time domain conversion is arranged according to the definition in the frame structure to form a complete time domain transmission frame. Time domain transmission: The time domain transmission symbol sequence {t1, t2, ..., tP} is sent to the RF part of the transmitter for modulation and amplification. The modulation method corresponds to the modulation order specified in the control information, such as QPSK, 16QAM, etc. The amplifier amplifies the modulated signal according to the channel conditions and power control requirements to ensure the transmission power and coverage of the signal. The modulated and amplified time domain signal is transmitted to the wireless channel through the transmitting antenna and transmitted to the receiving end.

[0059] Step 10: Map the time domain transmission symbol sequence {t1, t2, ..., tP} to a continuous time signal through digital-to-analog conversion (DAC), and then transmit it to the wireless channel after being processed by the RF front end (including up-conversion, filtering, power amplification, etc.). Specifically, digital-to-analog conversion (DAC): Map the time domain transmission symbol sequence {t1, t2, ..., tP} to a continuous time signal through digital-to-analog conversion (DAC). The function of DAC is to convert discrete-time digital signals into continuous-time analog signals to achieve the conversion of digital signals to analog signals. The key parameters of DAC include sampling rate, quantization bit number and conversion speed. The sampling rate determines the time resolution of the analog signal, the quantization bit number determines the amplitude resolution, and the conversion speed determines the processing capacity of the DAC. In order to avoid spectrum aliasing and distortion, the sampling rate of the DAC needs to meet the Nyquist sampling theorem, that is, the sampling rate should be at least twice the signal bandwidth. At the same time, the number of quantization bits should be high enough to ensure conversion accuracy and signal-to-noise ratio. RF front-end processing: The continuous-time analog signal output by the DAC is processed by the RF front-end, including up-conversion, filtering, power amplification and other steps. Up-conversion moves the baseband signal to the required RF carrier frequency to achieve signal spectrum shifting. Up-conversion is usually implemented using a mixer and a local oscillator. Filtering removes the image frequency and out-of-band interference generated by up-conversion to ensure the spectral purity of the signal. Commonly used RF filters include surface acoustic wave (SAW) filters, bulk acoustic wave (BAW) filters, etc. Power amplification amplifies the filtered RF signal to the required transmission power level to meet the requirements of transmission distance and coverage. The key indicators of the power amplifier include gain, efficiency, linearity, etc. Antenna transmission: The RF signal processed by the RF front-end is transmitted into the wireless channel through the antenna. The antenna converts the RF signal from wire transmission to electromagnetic wave transmission in free space to achieve radiation and propagation of wireless signals. The key parameters of the antenna include gain, radiation pattern, polarization mode, bandwidth, etc. Appropriate antenna design can improve transmission efficiency, increase coverage, and reduce interference. For multi-antenna systems, such as MIMO and beamforming, it is also necessary to consider factors such as mutual coupling and phase relationship between antennas to achieve spatial multiplexing and diversity gain. Wireless channel transmission: RF signals are transmitted in wireless channels and are affected by factors such as path loss, fading, multipath, and interference. Path loss is the phenomenon of energy attenuation during signal propagation, which is related to factors such as transmission distance and frequency. Fading is the change in amplitude and phase caused by obstacles, terrain, etc. encountered by the signal during propagation, including large-scale fading and small-scale fading. Multipath is the signal reaching the receiving end through different propagation paths, resulting in delay spread and frequency selective fading. Interference from other users, other systems or environmental noise will reduce the quality and reliability of the signal.

[0060] At the receiving end, the received signal is processed by the RF front end and converted to analog-to-digital (ADC) to obtain a digital baseband signal. The pilot symbol is extracted, and the least squares (LS) algorithm is used to estimate the channel frequency response to obtain the channel state information CSI. The CSI is used for channel equalization and synchronization, and then the control information, synchronization sequence and data symbols are extracted according to the frame structure of the transmitter. The control information is used to guide subsequent demodulation and decoding operations. The synchronization sequence is used for symbol timing synchronization and carrier frequency offset estimation and compensation. The data symbols undergo a series of reverse operations such as precoding matrix inversion, weighted summation, reconstruction, and matrix synthesis to restore the modulation symbol sequence. Then, through FBMC demodulation, polar code decoding and parallel-to-serial conversion, the reconstructed bit sequence is obtained, completing the low-power processing process of the entire mixed signal.

Claims

1. A mixed signal low power consumption processing method based on a multi-carrier receiver, comprising: Step 1, obtaining the information bit sequence to be transmitted and performing preprocessing; Preprocessing includes serial-to-parallel conversion and channel coding. The channel coding uses polar code. Step 2, mapping the preprocessed information bit sequence to a filter bank multi-carrier FBMC modulation symbol set to obtain a modulation symbol sequence; Step 3, using Logistic mapping to generate a chaotic sequence, performing a cross product between the obtained modulation symbol sequence and the generated chaotic sequence, and constructing the original information carrying matrix; Step 4, according to the number M of subcarriers of the differential chaotic shift keying DCSK system, the obtained original information carrying matrix is ​​divided into M submatrices, and each submatrix is ​​subjected to matrix decomposition processing to obtain a decomposed information carrying submatrix; Step 5, adaptively optimizing the decomposed information-carrying submatrix obtained; the adaptive optimization includes reconstructed matrix parameter optimization and power allocation strategy optimization; Step 6, reconstructing the decomposed information carrying sub-matrix using the reconstruction matrix parameters and the power allocation strategy to obtain an information carrying reconstructed sub-matrix; Step 7, performing weighted summation on each of the obtained information carrying reconstruction sub-matrices to obtain a final information carrying reconstruction matrix; Step 8, inserting pilot symbols and control information into the obtained information bearing reconstruction matrix; setting a frame structure, the frame structure including an arrangement of pilot symbols, control information, synchronization sequence and data symbols; Step 9, precoding the information bearing reconstruction matrix into which the pilot symbols and control information are inserted to obtain the precoded information bearing reconstruction matrix; the precoding processing includes adaptive optimization according to the channel state information CSI; converting the precoded information bearing reconstruction matrix into a time domain transmission symbol sequence according to the set frame structure; Step 10, transmitting the generated time domain transmission symbol sequence into a wireless channel after digital-to-analog conversion and radio frequency front-end processing; wherein the radio frequency front-end includes a power amplifier.

2. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 1, characterized in that: In step 3, the original information carrying matrix is ​​constructed, including: Rearrange the modulation symbol sequence obtained in step 2 according to the number of subcarriers in the FBMC modulation symbol set; Initialize an information bearing matrix, wherein the number of rows of the information bearing matrix is ​​equal to the number of subcarriers of the FBMC modulation symbol set; the number of columns of the information bearing matrix is ​​the product of the length of the chaotic sequence and the symbol period of the FBMC modulation symbol set; Writing the rearranged modulation symbol sequence into each row of the initialization information bearing matrix to obtain an information bearing matrix loaded with modulation symbols; Logistic mapping is used to generate a chaotic sequence with the same number of columns as the initial information-carrying matrix; The cross product operation is performed on each row element of the information carrying matrix loaded with the modulation symbol and the corresponding chaotic sequence element to obtain the original information carrying matrix.

3. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 2, characterized in that: In step 4, the decomposed information carrying submatrix is ​​obtained, including: Divide the original information-carrying matrix obtained in step 3 into M sub-matrices; Get the transformation matrix order of the Hadamard matrix decomposition of each submatrix; where the transformation matrix order is equal to the number of rows of the corresponding submatrix; According to the obtained transformation matrix order, a Hadamard transformation matrix of the corresponding order is constructed; wherein the elements of the Hadamard transformation matrix are composed of 1 and -1, and the product of the Hadamard transformation matrix and the corresponding transposed matrix is ​​equal to the transformation matrix order multiplied by the unit matrix; Using the constructed Hadamard transformation matrix, each sub-matrix is ​​decomposed into a Hadamard matrix to obtain the corresponding decomposed information-carrying sub-matrix.

4. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 3, characterized in that: Using the constructed Hadamard transformation matrix, perform Hadamard matrix decomposition on each sub-matrix, including: Divide the Hadamard transform matrix into multiple sub-blocks alternately according to rows and columns; The sub-blocks obtained by the division are recursively subjected to one-dimensional Hadamard transformation to obtain the decomposed information-carrying sub-matrix.

5. The mixed signal low power consumption processing method based on a multi-carrier receiver according to claim 1, characterized in that: In step 7, the final information carrying reconstruction matrix is ​​obtained, including: Obtain the channel state information CSI and calculate the signal-to-interference-noise ratio corresponding to each information-carrying reconstruction submatrix ; Among them, signal to interference and noise ratio It represents the ratio of signal power to interference and noise power; The signal-to-interference-noise ratio of the sub-matrix reconstructed according to each information carrier , construct the signal-to-interference-noise ratio Maximize the objective function. The objective function expression is as follows: ; Wherein, M represents the total number of information-bearing reconstruction submatrices; i represents the number of the information-bearing reconstruction submatrix, ranging from 1 to M; represents the signal-to-interference-noise ratio of the i-th information-carrying reconstruction submatrix; represents the weight coefficient of the i-th information-carrying reconstruction submatrix, ranging from 0 to 1; The convex optimization algorithm is used to solve the maximization objective function and obtain the optimal weighting coefficients of each information-carrying reconstruction submatrix; The obtained optimal weight coefficients are used to perform weighted summation on each information-bearing reconstruction sub-matrix to obtain the final information-bearing reconstruction matrix.

6. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 5, characterized in that: Signal to Interference and Noise Ratio The calculation formula is: ; in, represents the transmit power of the i-th information-bearing reconstruction submatrix; represents the channel gain of the i-th information-carrying reconstruction submatrix; represents the channel gain of the jth information-bearing reconstruction submatrix; represents the channel transmission coefficient of the j-th information carrying reconstruction sub-matrix to the i-th information carrying reconstruction sub-matrix; represents the noise power to which the i-th information-bearing reconstruction submatrix is ​​subjected.

7. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 6, characterized in that: In step 8, the pilot symbol is constructed by using an extended Zadoff-Chu sequence, and the extended Zadoff-Chu sequence is generated by combining multiple Zadoff-Chu sequences with different root sequence numbers in a multiplexing manner.

8. The low-power mixed signal processing method based on a multi-carrier receiver according to claim 7, characterized in that: In step 8, the synchronization sequence is constructed using an extended Barker code, which is a combination of an original Barker code and a cyclic shift sequence.

Images