Variational-Bayes-Assisted Joint Signal Detection, Noise Covariance Estimation, and Channel Tracking in MIMO-OFDM Systems

This paper introduces a robust variational Bayes (Robust-VB) receiver algorithm for joint signal detection, noise covariance matrix estimation, and channel impulse response (CIR) tracking in multiple-input-multiple-output (MIMO) orthogonal frequency-division multiplexing (OFDM) systems over time-varying channels. The VB framework and the turbo principle are combined to accomplish the parameter estimation and data detection. In the proposed Robust-VB receiver, a modified linear-minimum-mean-square-error interference cancellation (MLMMSE-IC) soft detector is developed based on the VB theory, which adaptively sets the log-likelihood ratio (LLR) clipping value according to the reliability of detection on each subcarrier to mitigate the error propagation. Following the signal detection, an adaptive noise covariance matrix estimator is derived for effective noise covariance estimation. Furthermore, to track time-varying channels, a VB soft-input Kalman filter (VB-Soft-KF) is first derived. However, unreliable soft symbols introduce outliers, which degrade the performance of VB-Soft-KF. To tackle this problem, we propose a robust VB-Soft-KF (VB-Robust-KF) based on the Huber $M$ estimation theory. Finally, the performance of the proposed algorithm is assessed via simulations, showing the superior performance of the Robust-VB receiver compared with the other benchmark receiver algorithms.