Adjustable Iterative Soft-Output Detection for Massive MIMO Uplink

Minimum mean square error (MMSE) detection algorithms in massive multiple-input multiple-output (MIMO) uplink suffer from prohibitively high complexity of exact matrix inversion. Conventional detection algorithms based on Neumann series expansion (NSE), which approximates the matrix inversion, can only be efficient with small number of NSE terms. Worse still, conventional NSE-based approaches are facing convergence problems in poor propagation environments. In this paper, we introduce an adjustable iterative detection algorithm based on NSE to solve the complexity and convergence problems. Furthermore, we propose an efficient approach to approximately compute the log-likelihood ratios (LERs) with low-complexity. Both the analytical and numerical results have shown that the proposed approach has benefits in terms of computational complexity and convergence rate, especially in the case of poor propagation environments.