Soft-Output-Sparse-Aware (SOSA) Detector for Coded MU-MIMO Systems

Abstract

In this correspondence, we present a novel soft-output-sparse-aware (SOSA) detector for an uplink multiuser multiple-input-multiple-output system. The proposed detector consists of two steps: First, a transmit vector is detected using a low-complexity detector; second, an improved soft-output is generated by exploiting the sparsity of a residual error vector (i.e., the difference between the transmit vector and the detected one). In contrast, the previous sparse-aware (SA) detectors used the sparsity to refine the hard-decision output of the underlying detector. Therefore, the SOSA detector can significantly outperforms the (hard-output) SA detectors for the coded systems since the former only can be well-incorporated into state-of-the-art soft channel decoders. Via simulation results, we demonstrate that the SOSA detector attains a nontrivial coded gain over the (soft-output) underlying detector with a similar complexity. Furthermore, the performance gap grows as signal-to-noise ratio and a system size grow.