Estimation of Integer Carrier Frequency Offset in OFDM Systems Based on the Maximum Likelihood Principle

An algorithm is derived for the estimation of the integer part of the carrier frequency offset in OFDM systems. The algorithm employs correlation of consecutive OFDM symbols in the frequency domain, is formulated for the general case of a system using data subcarriers and/or pilots and approximates the maximum likelihood (ML) correlation-based estimator for the AWGN channel. The cases of pseudo-pilots, boosted pilots and virtual subcarriers are also examined. It is shown that, compared to previous approaches based on symbol correlation, the algorithm makes better use of the information contained in the correlated symbols, leading to a more accurate, approximate ML estimate of the integer carrier frequency offset at operating signal-to-noise ratios. Simulations indicate that, even in the case of multipath channels, the proposed algorithm is very likely to outperform other techniques.