Joint Sparse-AR Model Based OFDM Compressed Sensing Time-Varying Channel Estimation

Abstract

In this paper, a time-varying channel estimation method based on compressed sensing (CS) is studied to reduce the pilot overhead for orthogonal frequency division multiplexing (OFDM) system. By taking advantage of the dynamic characteristic and temporal correlation of time-varying channel, we propose a novel channel estimation scheme based on joint sparse-autoregressive (AR) model. The proposed method performs the following two steps in a sliding window strategy. Firstly, the channel delay structure is estimated using the proposed sparsity adaptive simultaneous orthogonal matching pursuit (SASOMP) algorithm. Secondly, with the channel delay estimation, a reduced order Kalman filter (KF) is performed to obtain the amplitude of channel. Simulation results indicate that the proposed method is capable of recovering the time-varying channel with much lower pilot overhead than conventional CS-based channel estimators with a superior estimation performance.