Symbol Detection of Phase Noise-Impaired Massive MIMO Using Approximate Bayesian Inference

In this letter, we investigate the symbol detection of an uplink massive multiple-input multiple-output system impaired by phase noise at the transmitter and receiver sides. We propose a low-complexity iterative algorithm using approximate Bayesian inference based on the framework of generalized expectation consistent signal recovery to recover the symbol vector from nonlinear noisy measurements. Numerical results show that the proposed algorithm outperforms the existing algorithm and approaches the symbol error rate limit of a genie detector in high signal-to-noise ratio (SNR) regime, while the performance loss is very small in medium SNR. In particular, the complexity of proposed algorithm is quadratic, which makes it particularly suitable for large systems.