Randomized Iterative Sampling Decoding Algorithm For Large-Scale MIMO Detection

In this paper, the paradigm of the traditional iterative decoding schemes for the uplink large-scale MIMO detection is extended by sampling in an Markov chain Monte Carlo (MCMC) way. Different from iterative decoding whose performance is upper bounded by the suboptimal linear decoding scheme like ZF or MMSE, the proposed iterative random sampling decoding (IRSD) algorithm is capable of achieving the optimal ML decoding performance with the increment of Markov moves, thus establishing a flexible trade-off between suboptimal and optimal decoding performance. According to convergence analysis, we show that the Markov chain induced by IRSD algorithm experiences the exponential convergence, and its related convergence rate is also derived in detail. Based on it, the Markov mixing becomes tractable, followed by the decoding optimization with respect to the standard deviation of the target distribution. Meanwhile, further decoding performance enhancement and parallel implementation are also studied so that the proposed IRSD algorithm is well suited for various cases of large-scale MIMO systems.