An MKF Based Blind Receiver for Fast Fading MIMO Channels

In this paper, an MKF(mixture Kalman filtering) based blind receiver over fast fading MIMO channels is proposed. The whole system is firstly modeled as a conditional dynamic linear model, where the channel coefficients are considered as state variables, and the fast-fading processes of channels are modeled by AR(2) process. After that, the MKF is employed to resolve the detection problem. The receiver only draws Monte Carlo samples to the indicator variables (the transmitted signal vector), and uses a mixture of Gaussian distributions to approximate the statistics of channels. The trial distribution expression and weight update expression are derived, respectively. The simulation shows that the new receiver can achieve good SER performance, and can timely track the variation of channels.