Generalized frequency offset estimation in OFDM systems using periodic training symbol

Abstract

For orthogonal frequency division multiplexing (OFDM) systems, the bandwidth is divided into many subcarriers, requiring a highly accurate frequency offset estimator. In this paper, a novel frequency offset estimator utilizing the periodicity embedded in a training symbol is proposed. We generalize the well known correlation-based methods such that the estimation range of all the correlations embedded in a training symbol can be extended to the maximum. These correlations can thus be utilized freely without degrading the estimation range. Since only one training symbol is needed, the transmission efficiency is increased. The proposed method needs only the periodicity of the training symbol which leaves the design of pilot symbol for other considerations such as timing synchronization and channel estimation. To implement this estimator, a low-complexity algorithm is introduced. Performance analysis and simulation results are both given to compare with the previously proposed best linear unbiased estimator (BLUE), and the superiority of our proposal is witnessed.