Abstract: A new algorithm for blind channel estimation is proposed. A linear precoder is applied on each pair of OFDM symbols before they enter the OFDM systems. Due to the structure introduced by precoding, a new matrix is built to make the channel state information (CSI) be its one-dimension-subspace, and the CSI is estimated by using adaptive subspace tracking techniques. Simulation results demonstrate that the presented method can decreases the influence caused by the varying data distribution, channel, and noise, and can track the variation of channel, therefore, it lowers the mean squared error (MSE) of the channel estimation and the word error rate (WER) of the communication system.