A Compressed Sensing Estimation Technique for Doubly Selective Channel in OFDM Systems

Abstract

In this work, an estimation algorithm based on the compressive sensing (CS) technique is proposed for an orthogonal frequency division multiplexing (OFDM) system applied to multi-path time-varying channels. A fitting polynomial is used to approximate each channel path, and full-band training symbols are adopted. The variation of the channel response can then be evaluated according to the observation on the inter-carrier interference (ICI) from the determinate signals. The CS-based technique is introduced to explore the sparsity of the double-selective channel and the usage of high-order fitting polynomials is allowed. The orthogonal matching pursuit (OMP) algorithm is designed to cooperate with the fitting polynomial model, and more accurate channel estimation results can be provided compared to those provided by conventional least-square (LS) algorithms. When multiple transmission antennas are required, the advantage of the proposed polynomial-fitting OMP (PF-OMP) algorithm is more obvious as more transmission antennas are used.

Keywords

• Compressed sensing
• channel estimation
• OFDM systems