Time Varying Sparse Channel Estimation by MSE Optimization in OFDM Systems

This paper addresses the problem of sparse channel estimation in the context of Orthogonal Frequency Division Multiplexing (OFDM) systems. We propose to extend a recently proposed tap-tuned threshold-based Channel Impulse Response (CIR) structure detection scheme to the case of fast varying channels, through the use of a uniform pilot subcarriers placement. The chosen thresholds minimize the estimation Mean Squares Error (MSE) per tap. We also incorporate a local measure of the sparsity level which relies on the coarse Least Squares (LS) CIR estimate coefficients energies. The structured Channel Frequency Response (CFR) estimate is then exploited to demodulate the data subcarriers. Simulation results demonstrate the good denoising capacity of the proposed scheme.