Research On Low Complexity Methods In Massive MIMO System

Massive MIMO technology has attracted more and more attention as it can simultaneously improve the capacity and spectrum efficiency of wireless communication systems.In order to provide services for multiple users at the same time,the base station is usually equipped with large-scale Antenna arrays,which also bring severe user interference.Precoding technology can effectively suppress multiuser interference.And as the number of antennas increases,linear precoding can also achieve near-optimal precoding performance in massive MIMO systems,thus it has attracted more attention.Although the complexity of linear precoding is generally lower than that of nonlinear precoding,the complexity is still very high due to the need for matrix inversion operation in massive MIMO systems with large-scale channel matrix.The complexity needs to be further reduced.This article mainly studies low-complexity precoding algorithms in massive MIMO system,and proposes two new iteration-based low-complexity precoding algorithms.In this paper,we first improve the existing SAOR iterative algorithm.Via adding weighted coefficients,the weighted symmetric accelerated overrelaxation(WSAOR)algorithm is obtained and used for precoding.This algorithm includes various iterative algorithm forms,which can be degenerated to various existing iterative algorithms such as WSSOR iteration by parameter selection,thus it has better flexibility.Compared with the SAOR algorithm,the WSAOR algorithm has a great improvement in performance with a little increase in complexity.Among the various existing iterative algorithms,its spectrum radius of the iterative matrix is the smallest and the convergence speed is the fastest.Then we further accelerate the convergence speed at the beginning of the iteration by means of preconditioning and get the preconditioning weighted symmetric acceleration overrelaxation(PWSAOR)algorithm,its performance is verified through simulation.The performance is better than some existing precoding algorithms under the same complexity.When the signal-to-noise ratio is low or the channel estimation quality is not good enough,performance almost the same as ZF precoding can be obtained after one iteration,and the complexity is reduced by an order of magnitude.Finally,the FBAOR algorithm is used in precoding,the complexity is consistent with the SAOR algorithm and the performance is improved a lot.After combining with the weighting coefficient,the precoding algorithm based on the weighted forward and backward acceleration overrelaxation(WFBAOR)is obtained.The backward iteration parameters are different from the WSAOR algorithm,which further increases the degree of freedom.Compared with the WSAOR algorithm,the obtained WFBAOR algorithm can improve bit error rate and sum-rate performance under the condition of the same complexity.Among various iterative algorithms,its spectrum radius of the iterative matrix is the smallest and the convergence speed is the fastest.Finally,we parallelized the WFBAOR algorithm,which is originally solved one element by one element,and the performance is slightly reduced while the parallelism is enhanced,and it still has a fast convergence speed.