In this paper, a new modified iterative version of adaptive Kalman filtering algorithm is introduced which uses the short training sequences to adjust its filter weights with respect to time varying channel environment. Robustness against time varying channel is on the bases of Kalman gain. The modified algorithm also uses a measurement noise covariance that leads to fast convergence with respect to Signal to Noise (SNR) ratio. Simulation results show that the modified iterative algorithm presents robustness as well as minimum mean square error and less error probability when compared to Recursive Least Square (RLS) algorithm and Kalman channel equalizer. © 2013 IEEE.