An OFDM Symbol Design for Reduced Complexity MMSE Channel Estimation

Abstract

In this paper we revisit the minimum mean square error (MMSE) pilot-aided channel estimation for broadband orthogonal frequency division multiplexing (OFDM) systems. The careful design of OFDM symbol leads to the proposal of a simplified time-domain (TD) MMSE estimator. By exploring the Fourier properties of the symbol, the investigated method eliminates the need to use direct or inverse discrete Fourier transforms (DFT/IDFT) before the channel estimation, by using the TD received symbol samples as the input to the MMSE filter. Moreover, performing the estimation in TD (where the channel impulse response (CIR) energy is mainly limit to a small set of samples), makes way to a simple, nonetheless efficient estimation of the filter parameters. The performance of the channel estimation scheme, using the estimated parameters, presents a tolerable degradation when compared with the ideal situation (a-priori knowledge of the channel correlation and noise variance), while exhibiting a reduced computational load. The feasibility of the investigated method is substantiated by system simulation using indoor and outdoor wireless channel models.