An improved reweighted zero-attracting NLMS algorithm for broadband sparse channel estimation

Sparse channel estimation has attracted more attention for various broadband wireless communication systems. Square error criterion based adaptive filter algorithms are extensively studied for broadband sparse channel estimations (SCE) such as zero-attracting (ZA) least mean square (ZA-LMS) and reweighting ZA-LMS (RZA-LMS) algorithms. However, these sparse LMS algorithms are usually sensitive to the scaling of their input signal. In this paper, an improved sparse algorithm is proposed on the basis of the normalized LMS (NLMS) algorithm, reweighted ZA (RZA) techniques and compressed sensing concepts. The proposed SCE technique is implemented by using an error sequence to redesign the step-size (SS) of the NLMS to modify the RZA-NLMS algorithm. The complexity is also discussed based on the trace calculation strategy. The behaviors of the proposed SCE algorithm are verified over a broadband sparse multi-path wireless channel. The proposed results obtained from the simulation indicate that the presented SCE algorithms are superior to the conventional LMS, NLMS, ZA-LMS, RZA-LMS, ZA-NLMS and RZA-NLMS algorithms with reference to the convergence rate and the estimated error.