Research On Signal Detection Algorithm For Indoor High-Speed And Large-Capacity MIMO System

Recently years,a new generation of applications represented by artificial hologram and intelligent interaction continue to emerge,which puts forward higher requirements for the throughput and reliability of the communication system.Multiple-Input Multiple-Output（MIMO）is widely used as an important method to improve the transmission rate.However,it also makes it difficult to detect the signal at the receiving end.For this issue,the signal detection methods with high throughput,high reliability,and easy parallelizability are proposed.First of all,K-Best suitable for MIMO system with a high number of antennas is studied.Sorting Relaxed Parallel QR Decomposition is proposed to reduce the high complexity of the sorting QR decomposition in the channel matrix.For sorting,with the column vector l¹-norm as reference,each sorting can decompose k columns in parallel.For decompositing,Coordinate Rotation Digital Computer（CORDIC）is introduced to eliminate the element value of the channel matrix,and grouping matrix and scaling CORDIC are adopted to improve parallelism.The simulation results show that when k is equal to 4,the accuracy and complexity of QR decomposition can be compromised.On this basis,for the problem of doubling the number of constellation points in each layer in the real domain while searching,Dynamic Parallel K-Best is proposed.For reservating the paths,the nodes of each layer are sequentially expanded according to the size of Partial Euclidean Distance with Schnorr&Euchner（SE）optimal solution as the center.The number of reserved nodes is dynamically adjusted according to the probability that the partial solution coincides with Maximum Likelihood（ML）solution.For parallelizing,the channel matrix is reshaped to eliminate the correlation between adjacent layers.The nodes expanded from the two layers are combined,and then the final surviving paths are retained by threshold filtering.The simulation results show that the difference between the proposed algorithm and ML is only0.15d B when the bit error rate is 10^-5 in a 16×16,256-Quadrature Amplitude Modulation（QAM）system.The number of adders and multipliers required are 18944 and 19008,respectively,within acceptable complexity.Therefore,Dynamic Parallel K-Best Based on Relaxed Sorting Parallel QR Decomposition is suitable for MIMO system with a high number of antennas.Subsequently,Depth Sphere Decoder suitable for MIMO system with high-order QAM modulation is studied.To reduce the high computational complexity and avoid the empirical configuration of the initial search range,Optimization for Sphere Center and Dynamic Radius is proposed.For the center of the sphere,the channel matrix is inverted through each column.For the initial radius,the range of selectable points is narrowed by adding a weighted noise factor to the channel expansion matrix.The simulation results show that the column-wise inversion can obtain the accurate result while simplifying the complexity,and when the weighting coefficient is set to 0.4,the traditional search performance only loses0.01d B.On this basis,Double-Layer Iterative Depth Sphere Decoder is proposed to solve the problem of doubling the search depth in the real domain.For traversal order,the backtracking direction is determined by the principle of minimum Euclidean Distance in every two layers and the nodes are progressively selected into subsequent searches.For parallelizing,the search range of the two layers can be determined at one time in the real domain.At the same time,the number of updating radius is reduced to half of that of the traditional algorithm.The proposed algorithm improves the search efficiency.The simulation results show that in a 8×8,1024-QAM system,when the bit error rate is 10^-5,the difference between the proposed algorithm and ML is only 0.09d B.The number of adders and multipliers required are 12459 and 12140,respectively,within acceptable complexity.Therefore,the Double-Layer Iterative Depth Sphere Decoder Based on Optimization for Sphere Center and Dynamic Radius can support the signal detection of MIMO system with high-order QAM modulation well.After that,the architecture is designed for the above two algorithms,which provides a reference for the implementation.Finally,the outcomes are summarized,which demonstate the proposed algorithms are both reasonable and advanced,and the research direction in the future is prospected.