Gibbs sampling MIMO detection with maximum ratio combining

In this paper, Gibbs sampling multi-input multi-output (MIMO) detection with maximum ratio combining (MRC) is proposed and it is applied to the uplink with a massive MIMO receiver. In conventional Gibbs sampling schemes, the Gibbs sampling algorithm is directly applied to received signals. When it is applied to signals with higher order modulation symbols, a stalling problem occurs. Though a random restart scheme is proposed to solve the stalling problem, it needs adaptive restart criteria that makes hard to estimate the total amount of computational complexity. In the proposed scheme, each candidate symbol is updated with a metric calculated by multiplying corresponding MRC coefficients to the received signal vector. In this case, the convergence speed of the metric for the received signals even with higher order modulation symbols improves and the complexity of metric calculation for the update of a transmit candidate symbol vector can be reduced. Numerical results obtained through computer simulation have shown that the proposed scheme reduces the number of multiplication operations by a factor of 1/6.7 when it is applied to signals modulated with QPSK. It is also shown that the proposed scheme improves the convergence speed of the metric for E_b/N₀ ≥ 10dB when 16QAM symbols are transmitted. Furthermore, the proposed mixed Gibbs sampling achieves the same BER performance with smaller complexity as compared to that of a QR decomposition with M-algorithm (QRM)-maximum likelihood detection (MLD) when the number of transmit signals is more than 32.