Multiple restarts mixed Gibbs sampling detector for large-scale antenna systems

This work proposes a low-complexity detector for medium- and high-order modulation large-scale multiple-input multiple-output (LS-MIMO) systems based on the set of Markov chain Monte-Carlo techniques. Such efficient signal detection algorithm is based on the mixed Gibbs sampling with multiple restarts (MGS-MR) strategy with sample-averaged approach during the coordinate updating process, named averaged MGS (aMGS). The proposed strategy applies multiple samples average procedure to restrict the range of the random solution, which comes from the mixture proposed by the original MGS. Numerical simulation results considering higher-order M -QAM demonstrated that the proposed detection method can substantially improve the convergence of the MGS-MR algorithm, while no extra computational complexity is required. The proposed aMGS-based detector suitable for medium- and high-order modulation LS-MIMO further exhibits improved performance when the system loading is high, i.e. when ( K / N ) ≥ 0.75. In addition, the proposed numerical simulation analyses have shown that the optimal value of the mixing ratio parameter can vary regarding system and channel configuration scenarios, resulting somewhat different from the 1/2 K value disseminated in the literature.