A novel method for blind frequency offset correction in an OFDM system

Orthogonal frequency division multiplexing (OFDM) is believed to be an important technique for future wireline and wireless communications. One of the major disadvantages of such a multi-carrier modulation is the sensitivity of the performance to a frequency offset. The frequency offset can result from a Doppler shift due to mobile movement as well as from a carrier frequency synchronization error. Such a frequency offset causes a loss of the carriers' orthogonality, and hence inter-carrier interference (ICI). A previously proposed method to estimate and correct the frequency offset uses the retransmission of a data symbol, thus reducing the bandwidth efficiency. Another method, which merely reduces the effects of the ICI instead of completely correcting for effects of the frequency offset is based on repetition coding in the frequency domain, hence also reducing the throughput. In this paper, we propose a blind method using the correlation of the samples at the output of the discrete Fourier transform (DFT) to estimate the frequency offset. As no training sequence or symbol retransmission is required there is no loss in bandwidth efficiency. Additionally, we develop an adaptive algorithm for the frequency offset correction. The derived algorithm is basically a stochastic gradient algorithm which minimizes the mean squared frequency offset error.