A Novel Fast Linear Iteration Detection Algorithm in MU-Massive MIMO Systems

Abstract

Massive MIMO is an extremely attractive technology for the potential of improving transmission capacity significantly. However, the minimum mean square error (MMSE) detection algorithm has also high complexity because it involves matrix inversion in obtaining the near-optimal detection for uplink multiuser massive MIMO (MU-Massive MIMO) systems. Iteration detection is effective in avoiding matrix inversion, and several iteration methods have been proposed by academia as well. In this paper, we propose a parallel fast linear iteration detection (FLID) algorithm for uplink MU-Massive MIMO systems. The proposed FLID algorithm can accelerate the convergence rate through reducing the spectral radius $$\rho ({\mathbf {B}})$$ of the re-derived iteration matrix B, and reach almost the same near-optimal detection as MMSE through a parallel iteration. The computational complexity of FLID reduces from $$O(K^{3})$$ to $$O(K^{2})$$ . Simulation results show that the proposed FLID algorithm can also achieve the same detection performance to the traditional MMSE algorithm with lower complexity.