The Mean-Square-Error (MMSE) detection achieves near-optimal performance in signal detection for massive Multiple-Input-Multiple-Output (MIMO) systems. But MMSE detection still suffers from high complexity of matrix inversion. In this paper, an efficient and flexible architecture is proposed based on the modified version of the Symmetric Successive Over Relaxation (SSOR) method.
A Reconfigurable Computing Array (RCA) is used to implement the SSOR Method. In order to speed up the iteration, an initial solution is adopted. Approximated LLRs computational method is used to scale down the computing load of Log-Likelihood Rate computations. FPGA implementation results show a superior performance over the state-of-the-art designs.