Research On Low Complexity Detection Algorithm Based On Linear Iteration And Deep Learning In Massive MIMO

The minimum mean square error(MMSE)detection method in the Massive MIMO system has a problem that the complexity of matrix inversion is too high.In recent years,there have been many studies to reduce the complexity.On the premise of ensuring low complexity,improving convergence speed and detection performance has always been the focus of academic circles.Therefore,this thesis will use two algorithms of mathematical iteration and deep learning to jointly optimize the MMSE algorithm.The main research content of the thesis is summarized as follows:1.A detection algorithm based on symmetric accelerated over-relaxation(SAOR)is proposed,which uses an iterative algorithm to approximate the high-dimensional matrix inversion operation in the MMSE algorithm.The algorithm avoids complicated matrix inversion calculation,and the implementation complexity is reduced by an order of magnitude compared with the MMSE method.The simulation results show that the SAOR-based detection method can approach the detection performance of the MMSE algorithm with fewer iterations,which provides a better implementation method for the fast detection of received signals in Massive MIMO systems.2.The Jacobi-neural network(J-Net)detection algorithm is proposed,which uses traditional iterative algorithms and deep learning to jointly optimize the MMSE algorithm.This algorithm expands the traditional Jacobi iterative algorithm to the deep neural network.Through the simulation results,we can conclude that,compared with the Jacobi algorithm,the J-Net algorithm retains the advantage that the complexity is one order of magnitude lower than that of the MMSE algorithm,and even has a faster convergence speed and better BER performance.Under low SNR conditions,the algorithm even exceeds the performance of the MMSE algorithm.In addition,the algorithm also optimizes the problem that the performance of the Jacobi algorithm is limited by the size of the antenna.