Abstract: According to performance of the blind channel identification algorithms based on second-order statistics deteriorates under the small received data and is sensitive to the error of channel order. This paper proposes an improved blind channel identificattion algorithm based on FFT, This algorithm is expoliting the property of the MCR algorithm requiring minimum redundancy information to obtain the channel vector,then effectively reduces the computational complexity of the original algorithm(BI-FFT) but the performance is quite. It has been found that the extra estimated channel common zeros introduced by overestimating the channel order based on FFT-MCR algorithm are gathered around the unit circle, at the same time puts forward R-FFT-MCR algorithm that is robustness to channel order overestimation, We can search the common zeros around the unit circle through the idea of clustering, remove them and realize the correct channel estimation. Theoretical analysis and simulation results verify the effectiveness of the proposed algorithm.