Estimation of NON-WSSUS Channel for OFDM System: Exploiting Support Correlations through a Novel Adaptive Weighted Predict-Re-Estimate L1 Minimization

—It is challenging to estimate the wireless channel of the Orthogonal Frequency-Division Multiplexing (OFDM) broadband system under a changing communication environment. The difficulty is mainly attributed to this wireless channel’s Non Wide Sense Stationary Uncorrelated Scattering (Non-WSSUS) which has an implication that the delay and Doppler shift of such a channel are non-stationary and correlated. A Non-WSSUS channel is very different from the classical time-varying channel with constant delay and Doppler shift. In this paper, we propose an estimation method for the Non-WSSUS Channel Impulse Response (CIR) of the OFDM system. Based on the sparsity property of the delay-Doppler spread function, the delay and Doppler shift of Non-WSSUS channel can be extracted through a Compressive Sensing (CS) approach. Then a novel CS algorithm referred as Pre-Re L1 is proposed. The proposed CS algorithm exploits the correlations of the sparse supports to obtain adaptive weights for 1 L minimization. Numerical Simulation results show that the proposed CS method improves the performance of the NonWSSUS wireless channel estimation.