An efficient sparse channel estimation method with predetermined sparsity

The performance of frequency-domain equalization in a SC-FDE system is affected by the precision of channel estimation results and estimation methods based on compressive sensing have better performance in sparse channel condition such as underwater acoustic communication channels. However, the commonly used greedy algorithm in sparse channel estimation requires the sparsity to terminate the recursive process. Unlike many existing methods in which the sparsity is treated as a known factor, we propose a sparse channel estimation method with sparsity predetermined by wavelet decomposition. Typical LS estimation method is applied first and wavelet decomposition results of the estimated channel impulse response are used to set the threshold for determining the channel sparsity. With the predetermined sparsity, sparse channel estimation technique based on compressive sensing can achieve a better performance.