Sparse channel estimation in OFDM systems using improved smooth L0 algorithm

The channel response of data communication over orthogonal frequency division multiplexing (OFDM) systems is often required to be known at the receiver. Recently, many wireless multipath channels have been verified to exhibit sparse structures. To exploit the inherent sparsity, many channel estimation methods based on compressive sensing are proposed. In this paper, we introduce a channel estimation strategy using improved smooth L0 (SL0) algorithm. Instead of zero-mean Gaussian function which is the smooth function in smooth L0 algorithm, we utilize the hyperbolic tangent function as the smooth function to get better approximation of L0 norm, then the quasi-newton method is used to recover the signal. Numerical simulations results illustrate the performance gains achieved by the improved method.