A simple channel estimation via biasing OFDM symbols

Abstract

In this paper, blind channel estimation for Orthogonal Frequency-Division Multiplexing (OFDM) systems is considered. Based on a novel idea for biasing transmitted symbols, a simple estimation method is proposed. This method, which exploits first order statistics instead of higher orders, has a very simple structure that can be considered as the optimum solution in practical applications. The interesting point is that using the first order expected value instead of calculating covariance matrix, which is prevalent in the most former approaches, addresses many concerns in channel characterization such as estimation ambiguity and the necessity of using precoders which cast doubt on the blindness of the former methods. We investigate the asymptotic performance of the proposed estimator, i.e. normalized mean square error (NMSE), for large numbers of independent observations. The numerical results clearly show the effectiveness of our proposed algorithm in estimating channel elements as well as its improvement over the existing techniques.