Abstract
In this paper, we investigate soft-output K-best and fixed-complexity sphere decoder (FSD) detectors for multiple-input multiple-output (MIMO) systems, where high-order mod-ulations are considered. When using the conventional parallel candidate adding aided (PCA-aided) soft-output algorithm, the FSD and K-Best detectors provide similar bit error rate (BER) performances. In contrast, when using the proposed candidate-list-aided soft-output algorithm, the FSD detector outperforms the K-best detector and delivers a near-ML (maximum likelihood) performance. Therefore, a soft-output FSD detector architecture, which guarantees a feasible implementation complexity level, is devised for the MIMO system which consider large-scale antennas and modulation orders. Compared to the works reported in liter-ature, the proposed detector is able to provide a notably enhanced error performance with feasible implementation complexity.