A new Kalman-ML based channel tracking for correlated mimo fading channels in assistance with G. A.

Time varying nature of uplink/downlink channels in wireless communications, calls for prior knowledge about the channel parameters in the receiver side. This accounts for different techniques of estimating the parameters of the so called channel between the transmitter and the receiver. In this paper beside using a channel model for a typical multi-input multi-output (MIMO) wireless communication system, we propose a new modified Kalman filter based method for estimating and tracking the channel variations in time. An auto regressive (AR) model is fitted to the channel fluctuations, based on which the forthcoming variations is predicted via a one step predictive Kalman estimation. Then using the recently estimated channel state, the transmitted symbol is estimated via maximum likelihood (ML) estimation, the quantity which is needed by the Kalman estimator for the next one step prediction. In the proposed algorithm the parameters of the Kalman filter is updated at each training instant by defining a cost function which is minimized by a genetic algorithm based numerical method. In our proposed method the periodicity of sending training symbols in order to inform the receiver about the channel state in each transmitting interval is fairly increased. The efficiency of the method for channel matrix estimation is studied through simulation.The results show satisfactory match between the true and the estimated channel.