Research On Low-complexity Precoding Algorithm For Downlink In Massive MIMO Systems

Massive multiple-input multiple-output(massive MIMO)technology can achieve significant system capacity gain and energy efficiency without increasing the transmission power and spectrum.It is an indispensable key technology in the current 5th Generation(5g)Mobile Communication System.However,the complexity of downlink precoding algorithm in large-scale MIMO systems increases dramatically due to the excessive number of users and base station antennas.Although the channel hardening phenomenon can make the traditional linear precoding such as Minimum Mean Square Error(MMSE)approach the optimal performance of the system,it still involves a large-scale high-order Hermite matrix inversion process.In view of the high complexity of the current precoding algorithm,In this paper,two kinds of precoding schemes are studied,which can reduce the computational complexity while maintaining the performance of the algorithm:In order to solve the problem of high complexity in the inversion process of traditional MMSE precoding,a new algorithm is proposed after introducing the weight factor and acceleration factor on the basis of the Gauss-Selder algorithm.Firstly,the forward and backward iterative results are obtained by the symmetry of iterative matrix,and then use the acceleration factor to further accelerate the convergence speed of the algorithm.Then,based on the least square criterion,the approximate selection formula of the optimal weighting factor and the selection range of the acceleration factor are given.Finally,the simulation results prove that the proposed algorithm can effectively reduce the complexity of an order of magnitude on the basis of traditional MMSE precoding.It can not only avoid that the system is too sensitive to the weight factor,but also can meet the required inversion accuracy with a small number of iterations.In large-scale MIMO system,Newton iteration method is used in the traditional MMSE precoding inversion,but the calculation of the initial value is complicated.An improved algorithm is proposed to solve this problem.Firstly,an intermediate algorithm is proposed on the basis of SOR algorithm,and then combined with Newton iterative algorithm,the inverse of high-order matrix is directly estimated by the intermediate algorithm,and the result is taken as the initial value of Newton iteration method to accelerate the convergence speed.Simulation results show that compared with the traditional Newton iterative method,the improved algorithm can approach the performance of MMSE algorithm with less iterations and approximate the same complexity.