Adaptive channel estimation based on sparsity detection

Recently, compressive sensing (CS) is applied to time domain synchronous OFDM (TDS-OFDM), which has higher spectral efficiency and faster synchronization than standard cyclic prefix OFDM (CP-OFDM), to improve the performance over multi-path fading channels. The performance heavily relies on the sparsity level of the channel, which is unknown in practise. In this paper, a signal-to-noise ratio (SNR) based sparsity estimation method is firstly proposed to detect the sparsity level of the channel. Then we propose the priori aided subspace pursuit (PA-SP) algorithm if the channel sparsity meets the theoretical CS model, otherwise we use an improved iterative channel estimation method with the help of the channel sparsity. Simulation results demonstrate that the proposed scheme could improve the robustness and performance of TDS-OFDM systems compared with the conventional iterative methods or recently proposed CS based methods.