Abstract: To overcome the disadvantages of Doppler estimation algorithm based on channel sparsity detection,an improved method is proposed to achieve better performance under ice-water mixed regions and reduce computational complexity.This algorithm utilizes time domain resampling technique to compensate the signal with different compression factors,which avoids the high-order fast Fourier transformation operation in the frequency-domain resampling technique,thereby reducing the complexity of the algorithm.The pilot subcarriers are extracted more accurately by combining the comb pilot position vector,thereby obtaining more accurate channel estimation and channel sparsity detection results,and improving the precision of Doppler estimation.The simulation results and under-ice experiments demonstrate that this improved algorithm is able to effectively track the doppler under ice-water mixed regions and reduce the computational complexity,ensuring the robust mobile orthogonal frequency division multiplexing communication.