A GEM-BP based receiver for joint estimation and detection in MIMO systems

In this paper, we propose an iterative message passing receiver jointly performing channel, phase noise estimation and symbol detection for multi-input multi-output (MIMO) systems. The joint iterative receiver is designed using message passing techniques: general expectation maximization (GEM) and belief propagation (BP) on factor graphs. Therein, the nonlinear measurement function node and symbol detector are handled in E-step of GEM while the marginal posterior probability density function of channel parameter and phase noise are obtained by BP in M-step. Numerical simulation demonstrates that our proposed receiver outperforms the state-of-the-art soft-input extended Kalman smoothing approach in terms of mean square error, bit error rate and frame error rate with similar computational complexity.