Accelerated sphere decoding for multipleinput multiple-output systems using an adaptive statistical threshold

The authors present a complexity reduced near-maximum-likelihood (ML) scheme for the decoding of multiple-input multiple-output (MIMO) systems, which is targeted at a recently proposed fixed-complexity sphere decoder (FSD). The proposed decoder that the authors call the statistical threshold-based FSD (ST-FSD) combines a threshold constraint strategy with the FSD search region, thus speeding up the FSD algorithm by avoiding unnecessary search paths. As a consequence, higher efficiency and lower complexity can be obtained. The optimum threshold is derived through analysis of the statistical distributions of the correct and erroneous estimates. Furthermore, a tight lower bound on the threshold has been obtained by using the singular value decomposition (SVD) method and applied to the FSD. From simulation results, the proposed scheme is shown to be able to achieve a significant reduction in computational complexity with almost no performance degradation compared to the original FSD algorithm. Moreover, a novel grouped architecture for efficient hardware implementation of the proposed ST-FSD algorithm is motivated through simulation results and shown to compare favourably with the alternative options. This confirms that the ST-FSD is advantageous with respect to the original FSD in terms of the overall complexity.