A Reduced Complexity, Improved Frequency Offset Estimation for Ofdm-based Wlans

This paper presents a reduced complexity frequency offset estimation method as a modification to the nonlinear least square (NLS) method from [3]. The complexity reduction is achieved by utilizing the Chirp-Z transform. At the same estimation performance, the complexity gain, defined as the complexity ratio between the NLS method and the proposed method, can be approximately from 2 to 9 depending on the implemented complexity of the NLS method. At similar complexity, the estimation performance gain, defined as the mean square error ratio between the NLS method and the proposed method, can be approximately from 1.5 dB to 8.5 dB depending on the operating SNR value. We also present a modified training structure that can give approximately from 2.5 dB to 10.5 dB estimation performance gain over the training structures of [3] and IEEE 802.11a.