Complex Exponential Basis Expansion Model(CE-BEM) has the advantage of simple computation

independence on channel statistics and so on

it is used to time varying of channel estimation widely.But Gibbs effect produces unignored model error of CE-BEM.The paper analyzes the model error and the impact of baseline tilting method

then presents an channel estimator which mitigates the model error

the method estimate the slope of block period

then derives the new data model combined with the baseline tilting method and the LS algorithm and LMMSE algorithm.Computer simulation show that suggested method decreases MSE of critical-sampled CE-BEM and over-sampled CE-BEM in case of higher Doppler frequency shift extension

especially for the critical-sampled CE-BEM

performance is improved considerably.