Abstract: Channel estimation is a very important problem that has been paid much attention in orthogonal frequency division multiplexing (OFDM) communication system. In this paper, an improved blind OFDM channel estimation algorithm based on maximum likelihood is proposed. The improved algorithm uses the cyclic prefix (CP) processing to implement channel blind estimation. It estimates the influence of the OFDM channel before DFT. The analysis and simulation results show that the method proposed both reduces the computation complexity obviously and improves its performance in theory.