Performance Enhancement of OMP Algorithm for Compressed Sensing Based Sparse Channel Estimation in OFDM Systems

摘要

Long duration of the channel impulse response along with limited number of actual paths in orthogonal frequency division multiplexing (OFDM) vehicular wireless communication systems results in a sparse discrete equivalent channel. Implementing different compressed sensing (CS) algorithms enables channel estimation with lower number of pilot subcarriers compared to conventional channel estimation. In this paper, new methods to enhance the performance of the orthogonal matching pursuit (OMP) for CS channel estimation method is proposed. In particular, in a new algorithm dubbed as linear minimum mean square error-OMP (LMMSE-OMP), the OMP is implemented twice: first using the noisy received pilot data as the input and then using a modified received pilot data processed by the outcome of the first estimator. Simulation results show that LMMSE-OMP improves the performance of the channel estimation using the same number of pilot subcarrier. The added computational complexity is studied and several methods are suggested to keep it minimal while still achieving the performance gain provided by the LMMSE-OMP including using compressive sampling matching pursuit (CoSaMP) CS algorithm for the second round and also changing the way the residue is calculated within the algorithm.