An Accurate and Very Low Complexity LMMSE Channel Estimation Technique for OFDM Systems

In this paper, we propose a very low linear minimum mean square error (LMMSE) channel estimation technique for orthogonal frequency division multiplexing (OFDM) systems. The conventional LMMSE, based on the autocorrelation channel matrix, requires O(NP3)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$O(N_P^3)$$\end{document} operations (NP\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N_P$$\end{document} represents the total number pilot in one OFDM symbol). By exploiting the structure of the channel autocorrelation matrix, we propose a very low complexity LMMSE channel estimator which requires only O(NPlogNP)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$O(N_PlogN_P)$$\end{document} operations. Simulation results show that the proposed technique yields the same performances as well as the classical LMMSE in terms of bit error rate and mean square error.