A Bayesian estimation method for time-varying channel with the color noise was proposed.Based on random matrix theory and Gibbs sampling

the covariance matrix of the color noise and the channel parameters were firstly esti-mated.Then

the proposal distribution（or importance distribution） of the sequential monte carlo（SMC） sampler was modified to track the time-varying channel parameters blindly（with no pilot signals）.Therefore

the proposed method may online semi-blindly estimate the time-varying channel parameters in the background of color noise.Compared with existing approaches

which are based on Kalman filtering and particle filtering

the proposed method is more robust and more accurate while the color noise is considered.The simulation results show the effectiveness of the proposed method.