MIMO-OFDM系统中基于变分Bayes EM算法的联合符号检测与鲁棒Kalman信道跟踪

Abstract

A new joint symbol detection and channel estimation algorithm is proposed for MIMO-OFDM systems over fast fading channels based on the Variational Bayes Expectation-Maximization (VBEM) algorithm and turbo principle. By applying VBEM algorithm,a modified List Sphere Decoder (LSD) and soft-input Kalman Filter (KF) are derived for symbol detection and channel estimation, respectively. The covariance matrices of channel estimation are considered in the symbol detector, and the covariances of symbols are taken into account in channel estimator. In fast time-varying channels, soft symbols with low reliability may introduce outliers in soft-input kalman filter. Since the KF lacks robustness to outliers, filter divergence may occur and gives rise to propagation and error floor phenomenon. In order to mitigate the outliers effect, a robust soft-input kalman filter under the VBEM framework is proposed based on the theory of robust statistics, which can keep robust channel-tracking ability in fast fading channels. Simulation results show that the proposed VBEM algorithm has better performance over the conventional VBEM algorithm and EM algorithm in fast time-varying multipath channels.