Robust sparse channel estimation for OFDM system using an iterative algorithm based on complex median

In this paper, we present a robust approach to estimate communications channel in OFDM systems exploiting the sparsity of the channel impulse response (CIR) commonly found in multi-path channels. The CIR is found as the solution of a regularized optimization problem where we minimize the least absolute deviation of a residual signal while, at the same time, encourage sparsity in the solution by an ℓ₀ pseudo-norm regularization term. The proposed approach reduces to estimate iteratively each tap of the CIR using a complex median based operator followed by a relevance test that forces sparsity in the solution. A blanking filter at the front end of the receiver is used to further mitigate the impact of the impulsive noise. Extensive simulations show that the proposed approach performs better than conventional approaches for AWGN channels and for situations where the additive noise follows a heavier-than-Gaussian tail distribution.