MIMO System Signal Detection System and Method Based on Parallel Cholesky Matrix Decomposition

Abstract

The present invention discloses a signal detection system and method for a MIMO system based on parallel Cholesky matrix decomposition, which mainly solves the problems of high complexity in signal detection of existing MIMO systems and doubling of detection delay when performing signal detection in the real number domain. The implementation steps are as follows: the channel matrix conversion module converts the channel complex matrix into a channel real matrix and generates an iterative matrix; the channel matrix decomposition module parallelly decomposes the iterative matrix by using the Cholesky matrix decomposition method; the K-Best parallel search module parallelly searches for paths between adjacent two layers by using the K-Best parallel search method; the detection result decision module completes the detection of signals of all antennas and determines the detection value of the transmitted signal. The parallel design of the signal detection process in the present invention effectively reduces the detection complexity, reduces the detection delay, and can realize signal detection with fewer resource quantities.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and more particularly to a signal detection system and method for a multiple-input multiple-output (MIMO) system based on parallel Cholesky matrix decomposition in the field of multi-antenna technologies. The present invention can complete signal detection for an MIMO system based on parallel Cholesky matrix decomposition with relatively low complexity and low latency. Background Art

[0002] MIMO systems utilize spatial multiplexing technology to parallelize the originally serial data for transmission by multiple antennas. Without increasing the antenna transmission power and consuming more spectrum resources, the system throughput can be doubled, and it is applied to various wireless communication scenarios. In recent years, the research on MIMO signal detection methods mainly focuses on the K-Best signal detection method. This method can make a choice between performance and hardware resource consumption by setting the number of reserved paths, and has extremely high flexibility. However, there are problems such as high implementation complexity in the matrix decomposition process and a large number of expanded child nodes in the signal search process, resulting in a large amount of time and computing resources being consumed to complete the signal detection task, which is not in line with the development trend of low latency in the new generation of communication systems.

[0003] Espressif Systems (Shanghai) Co., Ltd. disclosed a K-Best based wireless signal detection method and system in its patent document "MIMO-OFDM wireless signal detection method and system with preprocessing of channel matrix in detection center" (Application No.: 201910152084.3, Publication No.: CN111628952 A). After receiving the first OFDM symbol, the method disclosed in this patent document preprocesses the channel matrix of each subcarrier to generate a global dynamic K value table, and determines the number of paths to be retained when performing signal detection on each subcarrier. At the same time, in order to further reduce the overall complexity of the method, the method performs signal detection in the real number domain, avoiding complex complex number operations. Although this method effectively reduces the number of child nodes expanded during the signal detection process and the complexity of the algorithm, the disadvantage of this method is that when performing signal detection in the real number domain, the channel matrix needs to be made real, which will double the size of the channel matrix, resulting in too high computational complexity during the signal detection process. In addition, the doubled channel matrix will bring a problem of doubling the processing delay when performing channel matrix decomposition and K-Best search. The system disclosed in this patent document consists of a channel matrix preprocessing sub-module, a storage sub-module, and a K-best search sub-module. First, the channel matrix preprocessing sub-module performs signal matrix preprocessing on the signal matrix of each subcarrier of the first OFDM symbol to generate a global dynamic K value table; then, the storage sub-module stores the global dynamic K value table and the search parameters of each subcarrier; finally, the K-best search sub-module performs MIMO detection on each subsequent OFDM symbol in the MIMO-OFDM data packet. In order to reduce the system complexity, the system uses the real number domain to replace the complex number domain calculation. However, the disadvantage of this system is that when performing signal detection in the real number domain, the matrix needs to be made real, doubling the size of the signal matrix, increasing the computational complexity during signal detection of the system, and at the same time, the K-bset search will also double the system processing delay. Summary of the Invention

[0004] The purpose of the present invention is to propose a MIMO system signal detection system and method based on parallel Cholesky matrix decomposition for the deficiencies of the above-mentioned existing technologies, so as to solve the problems of high complexity during signal detection in existing MIMO systems and doubling of detection delay when performing signal detection in the real number domain.

[0005] To achieve the above object, the idea of the present invention is that the method of the present invention performs parallel matrix decomposition on the channel matrix after real number conversion using the Cholesky matrix model. The redesigned channel matrix decomposition process can achieve parallel decomposition of two adjacent rows of elements at the cost of calculating one row of elements, reducing the number of calculations required for decomposing the channel matrix after real number conversion, effectively improving the decomposition speed of the channel matrix. Compared with the traditional channel matrix decomposition method, it overcomes the defects in the existing matrix decomposition technology, such as the overly high complexity of channel matrix decomposition and insufficient decomposition doubling time caused by doubling the size of the channel matrix. Moreover, for the upper triangular matrix obtained by decomposition, the first element on the right side of the diagonal element of its odd rows is fixed to 0, making full use of the characteristics of the decomposed matrix. The system of the present invention uses the K-best algorithm to expand the child nodes of two adjacent layers simultaneously during the search process. When expanding the child nodes, the retained paths are divided according to the sorting of the cumulative Euclidean distance. Each type of path expands different numbers of child nodes in the order of the SE enumeration rule. After combining and sorting the child nodes expanded for two adjacent layers, the selection of the K best paths is finally completed. Compared with the traditional search system, the improved system of the present invention only expands some nodes, effectively reducing the number of expanded child nodes and overcoming the deficiency of excessive expansion of child nodes in the existing method. To further reduce the time delay, when the system of the present invention performs signal search, after parallel expansion of the child nodes of two adjacent layers, by performing secondary expansion on the child nodes, the search process of two adjacent layers is combined into one time, and the search for all signals can be completed with only half the number of signal search times of the traditional method, effectively solving the problem of doubling the signal search time caused by doubling the size of the channel matrix.

[0006] The steps of the technical solution for achieving the object of the present invention are as follows:

[0007] The system of the present invention includes a channel matrix conversion module, a detection result decision module, a channel matrix decomposition module, and a K-Best parallel search module; where:

[0008] The channel matrix conversion module is used to convert the channel complex matrix of the MIMO system into a channel real matrix and generate an iterative matrix through the channel real matrix;

[0009] The channel matrix decomposition module is used to parallelly decompose the iteration matrix by using the Cholesky matrix decomposition method, select the first two rows of elements from the current iteration matrix as the elements to be decomposed, obtain the elements of the (2o - 1)-th row and the 2o-th row in the upper triangular matrix R through the elements to be decomposed, and at the same time remove the elements to be decomposed from the current iteration matrix to obtain the matrix after removing the decomposed elements, where the value of o is equal to the number of iterations; transpose the row vector composed of the elements of the (2o - 1)-th row, the (2o + 1)-th column to the 2N-th column in the upper triangular matrix R and multiply it by its own row vector to form an intermediate matrix; where the value of o is equal to the number of iterations, and the value of N is equal to the number of transmit antennas in the MIMO system; update the matrix after removing the decomposed elements by using the relevant elements of the intermediate matrix to obtain the updated iteration matrix; determine whether there are still unremoved elements in the current iteration matrix, if so, select the first two rows of elements from the current iteration matrix as the elements to be decomposed, otherwise, generate an equivalent unitary matrix Q through the upper triangular matrix R and the channel real matrix;

[0010] The K-Best parallel search module is used to parallelly search for the paths of adjacent two layers by using the K-Best method, select an undetected transmit antenna from all the transmit antennas of the MIMO system; divide the reserved search paths into three types of paths according to the sorting of the cumulative Euclidean distance, expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule, and the number of expanded sub-nodes is determined by the path type and the value of K; add the expanded lower-layer nodes to the end of the corresponding reserved path to obtain the expanded path, and then copy the expanded upper-layer nodes and add them to the end of the corresponding expanded path to perform a second expansion on the path to obtain the second-expanded path; calculate the cumulative Euclidean distance of each second-expanded path through the upper triangular matrix R and the equivalent unitary matrix Q, and select the K paths with the smallest cumulative Euclidean distance among them as the reserved paths;

[0011] The detection result decision module is used to determine whether all the transmit antennas of the MIMO system have been selected. If so, determine the detection value of the transmitted signal. Otherwise, parallelly search for the paths of adjacent two layers by using the K-Best method, select an undetected transmit antenna from all the transmit antennas of the MIMO system, and perform parallel search;

[0012] The detection result decision module is used to determine the detection value of the transmitted signal.

[0013] Select the path with the smallest cumulative Euclidean distance among the reserved paths. Starting from the root node, select two values each time, and use the two selected values as the real part and the imaginary part respectively to form a detection value of the complex value of the transmitted modulation signal until reaching the leaf node. The set of complex numbers composed of the detection values of the complex values of all the transmitted modulation signals is the signal detection value of all the transmit antennas.

[0014] The steps of the signal detection method of the present invention are as follows:

[0015] Step 1: The channel matrix conversion module converts the complex channel matrix of the MIMO system into a real channel matrix, and generates an iteration matrix through the real channel matrix;

[0016] Step 2: The channel matrix decomposition module parallelly decomposes the iteration matrix by using the Cholesky matrix decomposition method:

[0017] Step 2.1: Select the first two rows of elements from the current iteration matrix as the elements to be decomposed, obtain the elements of the (2o - 1)-th row and the 2o-th row in the upper triangular matrix R through the elements to be decomposed, and at the same time remove the elements to be decomposed from the current iteration matrix to obtain the matrix after removing the decomposed elements, where the value of o is equal to the number of iterations;

[0018] Step 2.2: Transpose the row vector composed of the elements of the (2o - 1)-th row, the (2o + 1)-th column to the 2N-th column in the upper triangular matrix R and multiply it with its own row vector to form an intermediate matrix; where the value of o is equal to the number of iterations, and the value of N is equal to the number of transmit antennas in the MIMO system;

[0019] Step 2.3: Update the matrix after removing the decomposed elements by using the relevant elements of the intermediate matrix to obtain the updated iteration matrix;

[0020] Step 2.4: Determine whether there are still undeleted elements in the current iteration matrix. If so, execute Step 2.1; otherwise, execute Step 2.5;

[0021] Step 2.5: Generate an equivalent unitary matrix Q through the upper triangular matrix R and the real channel matrix;

[0022] Step 3: The K-Best parallel search module parallelly searches for paths between adjacent two layers by using the K-Best method:

[0023] Step 3.1: Select an undetected transmit antenna from all the transmit antennas of the MIMO system;

[0024] Step 3.2: Divide the reserved search paths into three categories of paths according to the sorting of the cumulative Euclidean distance, and expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule. The number of expanded sub-nodes is determined by the path category and the value of K;

[0025] Step 3.3: Add the expanded lower-layer nodes to the end of the corresponding reserved path to obtain the expanded path, then copy the expanded upper-layer nodes and add them to the end of the corresponding expanded path to perform a second expansion on the path to obtain the path after the second expansion;

[0026] Step 3.4, calculate the cumulative Euclidean distance of each quadratically extended path through the upper triangular matrix R and the equivalent unitary matrix Q, and select the K paths with the smallest cumulative Euclidean distance as the retained paths;

[0027] Step 4, the detection result decision module determines whether all the transmit antennas of the MIMO system have been selected. If so, execute Step 5; otherwise, execute Step 3.

[0028] Step 5, the detection result decision module determines the detection value of the transmitted signal:

[0029] Select the path with the smallest cumulative Euclidean distance among the retained paths. Starting from the root node, select two values each time, and use the two selected values as the real part and the imaginary part respectively to form a detection value of the complex value of the transmitted modulation signal until reaching the leaf node. The set of complex numbers composed of the detection values of the complex values of all the transmitted modulation signals is the signal detection value of all the transmit antennas.

[0030] The present invention has the following advantages compared with the prior art:

[0031] First, the method of the present invention performs parallel decomposition iteration on the channel matrix after real number conversion by using the Cholesky matrix model. Compared with other real-domain signal detection methods, it realizes the parallel decomposition of adjacent two rows of elements at the cost of calculating one row of elements, overcomes the defect of the excessively high complexity of channel matrix decomposition caused by doubling the size of the channel matrix when performing real-domain signal detection in the prior art, and effectively improves the decomposition speed of the channel matrix. The present invention can complete the same channel matrix decomposition task with half of the computational complexity of the traditional real-domain channel matrix decomposition method, and has the advantages of low complexity and high decomposition speed.

[0032] Second, when the system of the present invention expands the child nodes in the K-best module, it divides the retained paths according to the sorting of the cumulative Euclidean distance. Each type of path expands different numbers of child nodes in the order of the SE enumeration rule. After combining and sorting the child nodes expanded in adjacent two layers, the selection of the K best paths is finally completed. By expanding some of the child nodes, the number of expanded child nodes is effectively reduced, and the computational complexity when expanding the child nodes is reduced. It overcomes the deficiency of excessive number of expanded child nodes in the prior system, further reduces the complexity of signal detection of the present invention, and reduces the number of resources used in the MIMO system.

[0033] Third, in order to further reduce the time delay, when the system of the present invention performs signal search, after completing the parallel expansion of adjacent child nodes, it performs secondary expansion on the child nodes, combines the search processes of adjacent two layers into one time, and effectively solves the problem of doubling the signal search time caused by doubling the size of the channel matrix, so that the present invention can complete the search task faster. Description of the Drawings

[0034] Figure 1 is the system block diagram of the present invention;

[0035] Figure 2 is the flowchart of the algorithm implementation steps of the present invention;

[0036] Figure 3 is the flowchart of the K - Best signal search of the present invention;

[0037] Figure 4 is the schematic diagram of SE enumeration of the present invention;

[0038] Figure 5 is the schematic diagram of the secondary expansion scheme of the present invention. Detailed Description of the Invention

[0039] The present invention will be further described in detail below in conjunction with the drawings and embodiments.

[0040] Refer to Figure 1 for a further detailed description of the system of the present invention.

[0041] The system of the present invention includes a channel matrix conversion module, a detection result decision module, a channel matrix decomposition module, and a K - Best parallel search module; where:

[0042] The channel matrix conversion module is used to convert the channel complex matrix of the MIMO system into a channel real matrix and generate an iterative matrix through the channel real matrix.

[0043] The channel matrix decomposition module is used to parallel - decompose the iterative matrix by using the Cholesky matrix decomposition method, select the first two - row elements from the current iterative matrix as the elements to be decomposed, obtain the elements of the (2o - 1) - th row and the 2o - th row in the upper triangular matrix R through the elements to be decomposed, and at the same time remove the elements to be decomposed from the current iterative matrix to obtain the matrix after removing the decomposed elements, where the value of o is equal to the number of iterations; transpose the row vector composed of the elements of the (2o - 1) - th row, the (2o + 1) - th column to the 2N - th column in the upper triangular matrix R and multiply it with its own row vector to form an intermediate matrix; where the value of o is equal to the number of iterations, and the value of N is equal to the number of transmit antennas in the MIMO system; update the matrix after removing the decomposed elements by using the relevant elements of the intermediate matrix to obtain the updated iterative matrix; determine whether there are still elements in the current iterative matrix that have not been removed. If so, select the first two - row elements from the current iterative matrix as the elements to be decomposed, otherwise, generate an equivalent unitary matrix Q through the upper triangular matrix R and the channel real matrix.

[0044] The K-Best parallel search module is used to parallelly search for paths between adjacent two layers by using the K-Best method, select an undetected transmitting antenna from all the transmitting antennas of the MIMO system; divide the reserved search paths into three types of paths according to the sorting of the cumulative Euclidean distance, expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule, and the number of expanded sub-nodes is determined by the path type and the K value; add the expanded lower-layer nodes to the end of the corresponding reserved paths to obtain the expanded paths, then copy the expanded upper-layer nodes and add them to the end of the corresponding expanded paths to perform a second expansion on the paths to obtain the doubly-expanded paths; calculate the cumulative Euclidean distance of each doubly-expanded path through the upper triangular matrix R and the equivalent unitary matrix Q, and select the K paths with the smallest cumulative Euclidean distance among them as the reserved paths.

[0045] The detection result decision module is used to determine whether all the transmitting antennas of the MIMO system have been selected. If so, determine the detection value of the transmitted signal. Otherwise, parallelly search for paths between adjacent two layers by using the K-Best method, select an undetected transmitting antenna from all the transmitting antennas of the MIMO system, and perform parallel search.

[0046] The detection result decision module is used to determine the detection value of the transmitted signal.

[0047] Select the path with the smallest cumulative Euclidean distance among the reserved paths. Starting from the root node, select two values each time, and use the two selected values as the real part and the imaginary part respectively to form a detection value of the transmitted modulation signal complex value until reaching the leaf node. The complex number set composed of the detection values of all the transmitted modulation signal complex values is the signal detection value of all the transmitting antennas.

[0048] Refer to Figure 2 , and further describe the implementation steps of the method of the present invention in detail.

[0049] Step 1, the channel matrix conversion module converts the channel complex matrix of the MIMO system into a channel real matrix, and generates an iterative matrix through the channel real matrix.

[0050] The conversion of the channel complex matrix of the MIMO system into a channel real matrix is realized by the following formula:

[0051]

[0052] where, H′ represents the channel real matrix, Re(·) represents the real part symbol, H1 represents the first column vector of the channel complex matrix, H N represents the Nth column vector of the channel complex matrix, the value of N is equal to the total number of transmitting antennas in the MIMO system, and Im(·) represents the imaginary part symbol.

[0053] Generating an iterative matrix using a channel real matrix refers to obtaining values of elements in odd rows of the iterative matrix using the channel real matrix, selecting elements in odd columns among the elements in the iterative matrix, grouping the element and the elements to its right, below, and below-right, and obtaining values of elements in even rows one by one within the group in such a way that the values of elements in the main diagonal direction are the same and the values of elements in the sub-diagonal direction are opposite to each other. The values of the elements in odd rows of the iterative matrix are obtained by the following formula:

[0054]

[0055] Among them, a i,j Represents the element in the i-th row and j-th column of the iteration matrix, i is an odd value selected in the interval [1,2N], the value of N is equal to the total number of transmitting antennas in the MIMO system, j is a real value selected in the interval [i,2N], H i ′ represents the vector consisting of the elements in column i of the channel real matrix, (·) T Represents the transpose operation on a matrix.

[0056] In the embodiment of the present invention, the total number of rows of the channel real matrix is 4, and the total number of columns is 4. The iterative matrix A obtained by the channel real matrix is in the following form:

[0057]

[0058] Step 2: The channel matrix decomposition module uses the Cholesky matrix decomposition method to decompose the iterative matrix in parallel:

[0059] Step 2.1: Select the first two rows of elements from the current iterative matrix as the elements to be decomposed, and obtain the elements in the 2o-1th and 2oth rows of the upper triangular matrix R through the elements to be decomposed. At the same time, remove the elements to be decomposed from the current iterative matrix to obtain the matrix after removing the decomposition elements, where the value of o is equal to the number of iterations.

[0060] The obtaining of the 2o-1th row and 2oth row elements in the upper triangular matrix R by the elements to be decomposed means performing a square root operation on the elements in the 1st row and the 1st column of the elements to be decomposed to obtain the values of the 2o-1th row and the 2o-1th column elements in the upper triangular matrix R. After obtaining the values of the remaining elements in the 2o-1th row of the upper triangular matrix R by the elements to be decomposed, the elements in the 2o-1th row in odd column positions are made to have values equal to the opposite of the values of the upper right elements, and the elements in odd column positions are made to have values equal to the values of the upper left elements. The values of the remaining elements in the 2o-1th row of the upper triangular matrix R are obtained by the following formula:

[0061]

[0062] Among them, r2o-1,2o+l Denotes the element in the \((2o - 1)\)-th row and \((2o + 1)\)-th column of the upper triangular matrix \(R\). The value of \(o\) is equal to the number of iterations, \(1\leq o\leq N\), where \(N\) is equal to the total number of transmit antennas in the MIMO system, and \(l\) is a real value selected in the interval \([2, 2N - 2o]\), \(a\) 1,l Denotes the element in the 1st row and \(l\)-th column of the element to be decomposed, \(r\) 2o-1,2o-1 Denotes the element in the \((2o - 1)\)-th row and \((2o - 1)\)-th column of the upper triangular matrix \(R\).

[0063] Step 2.2: Transpose the row vector composed of the elements from the \((2o + 1)\)-th column to the \(2N\)-th column in the \((2o - 1)\)-th row of the upper triangular matrix \(R\), and multiply it by itself to form an intermediate matrix; where the value of \(o\) is equal to the number of iterations, and \(N\) is equal to the number of transmit antennas in the MIMO system.

[0064] Step 2.3: Update the matrix after removing the decomposed element using the relevant elements of the intermediate matrix to obtain the updated iterative matrix.

[0065] The update of the matrix after removing the decomposed element using the relevant elements of the intermediate matrix is implemented by the following formula:

[0066] a″ 2x-1,2y-1 = a2′ x-1,2y-1 -(t 2x-1,2y-1 + t 2x,2y )

[0067] a″ 2x-1,2y = a′ 2x-1,2y -(t 2x-1,2y - t 2x,2y-1 )

[0068] where \(a″\) 2x-1,2y-1 Denotes the element in the \((2x - 1)\)-th row and \((2y - 1)\)-th column of the updated iterative matrix. \(x\) is a real value selected in the interval \([1, N - o]\), \(N\) is equal to the total number of transmit antennas in the MIMO system, \(y\) is a real value selected in the interval \([x, N - o]\), \(a′\) 2x-1,2y-1 Denotes the element in the \((2x - 1)\)-th row and \((2y - 1)\)-th column of the matrix after removing the decomposed element, \(t\) 2x-1,2y-1 Denotes the element in the \((2x - 1)\)-th row and \((2y - 1)\)-th column of the intermediate matrix, \(t\) 2x,2y Denotes the element in the \(2x\)-th row and \(2y\)-th column of the intermediate matrix.

[0069] Step 2.4: Determine whether there are still elements in the current iterative matrix that have not been removed. If so, execute Step 2.1; otherwise, execute Step 2.5.

[0070] Step 2.5: Generate the equivalent unitary matrix \(Q\) through the upper triangular matrix \(R\) and the channel real matrix.

[0071] The generation of the equivalent unitary matrix Q from the upper triangular matrix R and the channel real matrix is achieved by the following formula:

[0072] Q = (R -1 ) T H′ T

[0073] where Q represents the equivalent unitary matrix, R -1 represents the inverse matrix of the upper triangular matrix R, (·) T represents the transpose operation on a matrix or vector, and H′ represents the channel real matrix.

[0074] Step 3: The K-Best parallel search module uses the K-Best method to parallel search for paths between adjacent two layers.

[0075] Refer to Figure 3 for a further detailed description of the steps of using the K-Best method in the present invention to parallel search for paths between adjacent two layers.

[0076] Step 3.1: Select an undetected transmit antenna from all the transmit antennas of the MIMO system.

[0077] Step 3.2: Divide the reserved search paths into three categories according to the sorting of the cumulative Euclidean distance, and expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule. The number of expanded sub-nodes is determined by the path category and the value of K.

[0078] In the embodiment of the present invention, for the K-Best search algorithm with a signal modulation order of 64QAM and 8 reserved paths, set the number of high-confidence paths to 3, with each path expanding 4 sub-nodes, the number of medium-confidence paths to 3, with each path expanding 2 sub-nodes, and the number of low-confidence paths to 2, with each expanding 1 sub-node.

[0079] The SE enumeration rule is as Figure 4 shown, Figure 4 in which, -5, -3, -1, 1, 3, 5 represent the real part values after constellation mapping, represents the signal value calculated under the current path. During enumeration, the Euclidean distances from the constellation mapping values to are sorted in ascending order, and the sub-nodes to be expanded are selected from the codebook according to the sorting result.

[0080] The calculation of the signal estimation value is achieved by the following formula:

[0081]

[0082] where represents the signal estimation value of the z-th node in the current path, Qz denotes the vector composed of the elements in the \(z\)-th row of the equivalent unitary matrix, denotes the vector of the real-valued received signal. \(\sum\cdot\) represents the summation operation. The value of \(M\) is equal to the total number of receiving antennas in the MIMO system. \(k\) is a real value sequentially selected in the interval \([z + 1, 2M]\), and \(r\) z,k represents the element in the \(z\)-th row and \(k\)-th column of the upper triangular matrix \(R\).

[0083] Step 3.3: Add the expanded lower-layer nodes to the end of the corresponding reserved path to obtain the expanded path. Then, after copying the expanded upper-layer nodes, add them to the end of the corresponding expanded path to perform a second expansion on the path to obtain the doubly-expanded path.

[0084] In the embodiments of the present invention, after parallelly expanding adjacent two layers of child nodes, the way of performing a second expansion on the child nodes is as Figure 5 shown. First, perform an expansion operation on the same parent node of the \((2i + 1)\)-th layer of antenna \(i\) to obtain \(n\) expanded child nodes of the \(2i\)-th layer and \(n\) expanded child nodes of the \((2i - 1)\)-th layer. Then, perform a second expansion on the \(n\) child nodes of the \(2i\)-th layer and the \((2i - 1)\)-th layer, and merge the child nodes of the adjacent two layers after expansion to obtain an equivalent path.

[0085] Step 3.4: Calculate the cumulative Euclidean distance of each doubly-expanded path through the upper triangular matrix \(R\) and the equivalent unitary matrix \(Q\), and select the \(K\) paths with the smallest cumulative Euclidean distance among them as the reserved paths.

[0086] The calculation of the cumulative Euclidean distance of each doubly-expanded path is implemented by the following formula:

[0087]

[0088] where \(d\) m-2 represents the cumulative Euclidean distance of the doubly-expanded path with the number of nodes to be searched being \(m - 2\). The value of \(m\) is equal to the number of nodes to be searched, \(1\leq m\leq 2M\). \(Q\) m denotes the vector composed of the elements in the \(m\)-th row of the equivalent unitary matrix, denotes the vector of the real-valued received signal, denotes the signal estimation value corresponding to the \(m\)-th node of the current path, and its value range is determined by the signal modulation method used in the current MIMO system. \(r\) m,m represents the element in the \(m\)-th row and \(m\)-th column of the upper triangular matrix \(R\), and \(|\cdot|\) represents the absolute value operation.

[0089] Step 4: The detection result decision module determines whether all the transmitting antennas of the MIMO system have been selected. If so, execute Step 5; otherwise, execute Step 3.

[0090] Step 5: The detection result decision module determines the detection value of the transmitted signal:

[0091] Select the path with the smallest cumulative Euclidean distance in the reserved path. Starting from the root node, select two values each time. Use the two selected values as the real part and the imaginary part respectively to form a detected value of the complex value of the transmitted modulation signal until reaching the leaf node. The set of complex numbers composed of the detected values of the complex values of all transmitted modulation signals is the signal detected value of all transmitting antennas.

Claims

1. A signal detection system for a MIMO system based on parallel Cholesky matrix decomposition, comprising a channel matrix conversion module and a detection result decision module; characterized in that, It further includes a channel matrix decomposition module and a K-Best parallel search module; where: The channel matrix conversion module is used to convert the channel complex matrix of the MIMO system into a channel real matrix, and generate an iterative matrix through the channel real matrix; The channel matrix decomposition module is used to parallelly decompose the iterative matrix by using the Cholesky matrix decomposition method, select the first two rows of elements from the current iterative matrix as the elements to be decomposed, obtain the elements of the (2o - 1)-th row and the 2o-th row in the upper triangular matrix R through the elements to be decomposed, and at the same time remove the elements to be decomposed from the current iterative matrix to obtain the matrix after removing the decomposed elements, where the value of o is equal to the number of iterations; transpose the row vector composed of the elements of the (2o - 1)-th row, the (2o + 1)-th column to the 2N-th column in the upper triangular matrix R and multiply it by its own row vector to form an intermediate matrix; where the value of o is equal to the number of iterations, and the value of N is equal to the number of transmit antennas in the MIMO system; update the matrix after removing the decomposed elements by using the relevant elements of the intermediate matrix to obtain the updated iterative matrix; determine whether there are still unremoved elements in the current iterative matrix, if so, select the first two rows of elements from the current iterative matrix as the elements to be decomposed, otherwise, generate an equivalent unitary matrix Q through the upper triangular matrix R and the channel real matrix; The K-Best parallel search module is used to parallelly search the paths of adjacent two layers by using the K-Best method, select an undetected transmit antenna from all the transmit antennas of the MIMO system; divide the reserved search paths into three types of paths according to the sorting of the cumulative Euclidean distance, expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule, and the number of expanded sub-nodes is determined by the path category and the value of K; add the expanded lower-layer nodes to the end of the corresponding reserved path to obtain the expanded path, and then copy the expanded upper-layer nodes and add them to the end of the corresponding expanded path to perform a second expansion on the path to obtain the second-expanded path; calculate the cumulative Euclidean distance of each second-expanded path through the upper triangular matrix R and the equivalent unitary matrix Q, and select the K paths with the smallest cumulative Euclidean distance as the reserved paths; The detection result decision module is used to determine whether all the transmit antennas of the MIMO system have been selected. If so, determine the detection value of the transmitted signal. Otherwise, parallelly search the paths of adjacent two layers by using the K-Best method, select an undetected transmit antenna from all the transmit antennas of the MIMO system, and perform parallel search; The detection result decision module is used to determine the detection value of the transmitted signal; Select the path with the smallest cumulative Euclidean distance among the reserved paths. Starting from the root node, select two values each time, and use the selected two values as the real part and the imaginary part respectively to form a detection value of the complex value of the transmitted modulation signal until reaching the leaf node. The set of complex numbers composed of the detection values of the complex values of all the transmitted modulation signals is the signal detection value of all the transmit antennas.

2. A signal detection method for a MIMO system based on parallel Cholesky matrix decomposition of the detection system according to claim 1, characterized in that, The iterative matrix is parallelly decomposed by using the Cholesky matrix decomposition method, and the paths of adjacent two layers are parallelly searched by using the K-Best method. The steps of this method are as follows: Step 1, the channel matrix conversion module converts the channel complex matrix of the MIMO system into a channel real matrix, and generates an iterative matrix through the channel real matrix; Step 2, the channel matrix decomposition module parallelly decomposes the iterative matrix by using the Cholesky matrix decomposition method: Step 2.1, select the first two rows of elements from the current iterative matrix as the elements to be decomposed, obtain the elements of the (2o - 1)-th row and the 2o-th row in the upper triangular matrix R through the elements to be decomposed, and at the same time remove the elements to be decomposed from the current iterative matrix to obtain the matrix after removing the decomposed elements, where the value of o is equal to the number of iterations; Step 2.2, transpose the row vector composed of the elements of the (2o - 1)-th row, the (2o + 1)-th column to the 2N-th column in the upper triangular matrix R and multiply it with its own row vector to form an intermediate matrix; where the value of o is equal to the number of iterations, and the value of N is equal to the number of transmit antennas in the MIMO system; Step 2.3, update the matrix after removing the decomposed elements by using the relevant elements of the intermediate matrix to obtain the updated iterative matrix; Step 2.4, determine whether there are still elements in the current iterative matrix that have not been removed. If so, execute Step 2.1; otherwise, execute Step 2.5; Step 2.5, generate an equivalent unitary matrix Q through the upper triangular matrix R and the channel real matrix; Step 3, the K-Best parallel search module parallelly searches for the paths of adjacent two layers by using the K-Best method: Step 3.1, select an undetected transmit antenna from all the transmit antennas of the MIMO system; Step 3.2, divide the reserved search paths into three types of paths according to the sorting of the cumulative Euclidean distance, and expand the sub-nodes of adjacent two layers simultaneously according to the SE enumeration rule. The number of expanded sub-nodes is determined by the path type and the value of K; Step 3.3, add the expanded lower-layer nodes to the end of the corresponding reserved path to obtain the expanded path, then copy the expanded upper-layer nodes and add them to the end of the corresponding expanded path to perform a second expansion on the path to obtain the path after the second expansion; Step 3.4, calculate the cumulative Euclidean distance of each path after the second expansion through the upper triangular matrix R and the equivalent unitary matrix Q, and select the K paths with the smallest cumulative Euclidean distance among them as the reserved paths; Step 4, the detection result decision module determines whether all the transmit antennas of the MIMO system have been selected. If so, execute Step 5; otherwise, execute Step 3; Step 5, the detection result decision module determines the detection value of the transmitted signal: Select the path with the smallest cumulative Euclidean distance among the reserved paths. Starting from the root node, select two values each time, and use the two selected values as the real part and the imaginary part respectively to form a detection value of the complex value of the transmitted modulation signal until reaching the leaf node. The set of complex numbers composed of the detection values of the complex values of the transmitted modulation signals is the signal detection value of all the transmit antennas.

3. The signal detection system for MIMO system based on parallel Cholesky matrix factorization according to claim 1, wherein The conversion of the channel complex matrix of the MIMO system into a channel real matrix described in Step 1 is realized by the following formula: Among them, H′ represents the real channel matrix, Re(·) represents the real part symbol, H1 represents the first column vector of the complex channel matrix, and H N represents the Nth column vector of the complex channel matrix. The value of N is equal to the total number of transmit antennas in the MIMO system, and Im(·) represents the imaginary part symbol.

4. The signal detection system for MIMO system based on parallel Cholesky matrix decomposition according to claim 1, wherein The generation of the iteration matrix from the channel real matrix described in Step 1 means obtaining the values of the elements in the odd rows of the iteration matrix from the channel real matrix, selecting the elements in the odd rows of the iteration matrix that are in the odd column positions, dividing the element and its right, lower, and lower-right elements into a group, and obtaining the values of the elements in the even rows one by one within the group in such a way that the values of the elements in the main diagonal direction are the same and the values of the elements in the anti-diagonal direction are opposite to each other. Among them, the values of the elements in the odd rows of the iteration matrix are obtained by the following formula: where a i,j represents the element at the i-th row and j-th column in the iteration matrix, i is an odd value selected within the interval [1, 2N], the value of N is equal to the total number of transmit antennas in the MIMO system, j is a real value selected within the interval [i, 2N], H i ′ represents the vector composed of the i-th column elements of the channel real matrix, (·) T represents the transpose operation on the matrix.

5. The signal detection system for MIMO system based on parallel Cholesky matrix decomposition according to claim 1, characterized in that, The obtaining of the elements in the (2o - 1)-th row and the 2o-th row of the upper triangular matrix R from the element to be decomposed described in Step 2.1 means performing a square root operation on the element in the first row and the first column of the element to be decomposed to obtain the value of the element in the (2o - 1)-th row and the (2o - 1)-th column of the upper triangular matrix R. After obtaining the values of the remaining elements in the (2o - 1)-th row of the upper triangular matrix R from the element to be decomposed, for the elements in the odd column positions of the (2o - 1)-th row, the value is taken as the opposite of the value of the upper-right element, and for the elements in the even column positions, the value is taken as the value of the upper-left element. Among them, the values of the remaining elements in the (2o - 1)-th row of the upper triangular matrix R are obtained by the following formula: where r 2o-1,2o+l represents the element in the (2o - 1)-th row and (2o + l)-th column of the upper triangular matrix R, where the value of o is equal to the number of iterations, 1 ≤ o ≤ N, and the value of N is equal to the total number of transmit antennas in the MIMO system, and l is a real value selected within the interval [2, 2N - 2o], a 1,l represents the element in the 1st row and l-th column of the element to be decomposed, r 2o-1,2o-1 represents the element in the (2o - 1)-th row and (2o - 1)-th column of the upper triangular matrix R.

6. The signal detection system for MIMO system based on parallel Cholesky matrix decomposition according to claim 1, characterized in that, The update of the matrix after removing the decomposed element using the relevant elements of the intermediate matrix in Step 2.3 is achieved by the following formula: a″ 2x-1,2y-1 = a' 2x-1,2y-1 -(t 2x-1,2y-1 + t 2x,2y ) a″ 2x-1,2y = a′ 2x-1,2y -(t 2x-1,2y -t 2x,2y-1 ) where a″ 2x-1,2y-1 represents the element in the (2x - 1)-th row and (2y - 1)-th column of the updated iterative matrix, where x is a real value selected in the interval [1, N - o], N is equal to the total number of transmit antennas in the MIMO system, and y is a real value selected in the interval [x, N - o], and a′ 2x-1,2y-1 represents the element in the (2x - 1)-th row and (2y - 1)-th column of the matrix after removing the decomposed element, and t 2x-1,2y-1 represents the element in the (2x - 1)-th row and (2y - 1)-th column of the intermediate matrix, and t 2x,2y represents the element in the 2x-th row and 2y-th column of the intermediate matrix.

7. The signal detection system of the MIMO system based on parallel Cholesky matrix decomposition according to claim 1, wherein The generation of the equivalent unitary matrix Q from the upper triangular matrix R and the channel real matrix in Step 2.5 is achieved by the following formula: Q = (R -1 ) T H′ T where Q represents an equivalent unitary matrix, R -1 represents the inverse matrix of the upper triangular matrix R, (·) T represents the transpose operation on a matrix or vector, and H′ represents the real channel matrix.

8. The signal detection system for a MIMO system based on parallel Cholesky matrix decomposition according to claim 1, characterized in that, The calculation of the cumulative Euclidean distance of each quadratic-expanded path in Step 3.4 is achieved by the following formula: where d m-2 represents the cumulative Euclidean distance of the quadratic extended path with the number of nodes to be searched being m - 2, where the value of m is equal to the number of nodes to be searched, 1 ≤ m ≤ 2M, Q m represents the vector composed of the elements of the m-th row of the equivalent unitary matrix, represents the vector of the real-valued received signal, represents the signal estimation value corresponding to the m-th node of the current path, and its value range is determined by the signal modulation method used in the current MIMO system, r m,m represents the element of the m-th row and the m-th column of the upper triangular matrix R, and |·| represents the absolute value operation.

Images