A low-complexity iterative channel estimation (ICE) algorithm is proposed with the promise of improved error performance. The new algorithm operates the LMS filter both in the forward and the backward directions along a block. The feedback from the decoder to the estimator is in the form of soft decisions. The pilot symbol assisted modulation (PSAM) is used as the transmission technique. The effect of code choice on various ICE algorithms is also explored by considering the blockwise concatenated codes initially offered for block-fading channels. The performance of the new estimation algorithm with the proposed coding is shown to outperform the conventional estimation algorithms over a fast time-varying Rayleigh fading channel beside its low complexity structure.