Abstract: This paper presents a method of improving the least square channel estimator accuracy without increasing the pilot density. The Kalman filter process equation can be represented as Autoregressive model, and the least square channel estimate is seen as a noisy measurement of the true channel state, so the Kalman filter measurement equation can be represented as the least square estimated channel. The process equation and the measurement equation jointly form a state space model of the dynamic of the channel. Thus the Kalman filter can be used to estimate the state variable, i.e., the time varying channel. The Kalman filter based channel estimator leads to a significant gain in performance as compared to the least square channel estimator.