Multistage sequential Gaussian Approximation for MIMO systems

This paper presents a reduced complexity detection scheme, based on sequential Gaussian Approximation (SGA), called multistage SGA detection (MSGA), for multi-input multi-output wireless communication systems with high-order QAM modulation. In the SGA detection, only part of transmit symbol combinations contributing significantly to posterior probabilities are retained, having a complexity linear with retained branches number. We develop a multistage SGA for QAM MIMO system with a modulation order Mc by decomposing the original QAM symbol into Mc/2 4-QAM symbols, resulting in a complexity reduction from exponential in the modulation order to linear, compared to basic SGA detection. Simulation results show that the multistage SGA can approach the performance of the basic SGA detection.