Compressed Sensing Based on Doubly-selective Slow-fading Channel Estimation in OFDM Systems

In order to improve the reconstruction accuracy of smoothed l0-norm (Sl0) algorithm in the presence of noise, a modified algorithm named smoothed l0-norm regularized least-square (l2-Sl0) is proposed in this paper, which permits a small perturbation. Further, through placing a finite grid in the planar time-frequency bounded region, the problem of doubly-selective slow-fading channel estimation in OFDM system is modeled as the problem of sparse signal reconstruction in compressed sensing framework, and then thel2-Sl0 algorithm is applied to reconstruct the channel parameters. A number of computer-simulation-based experiments show that reconstruction accuracy of thel2-Sl0 algorithm is improved by approximately 10 dB as compared with theSl0 algorithm in the presence of noise. The performance of the proposed doubly-selective slow-fading channel estimation method usingl2-Sl0 algorithm is nearly close to that of the ideal Least Square (ideal-LS) method. Moreover, the proposed method has higher estimation uccuracy well in the case of low SNR.