Employing simple FFT-interpolation for improved complex tone detection and fine estimation

Frequency offset estimation is an important task in mobile communication receivers. FFT-based approaches have been shown to have a very good performance at low siggal-to-noise ratios (SNR). They usually consist of two steps: A coarse estimation by searching for the maximum FFT-value and a subsequent fine estimation for which many different approaches exist. In this paper, we demonstrate how simple interpolation in frequency domain can be used to improve the search for the maximum FFT-value without having to rely on zero-padding. The concept is then applied and tested in an acquisition scheme where large frequency offsets occur. Further, the interpolation scheme is used to reduce the complexity of a previously proposed frequency offset estimator. The basic idea of the interpolation scheme is that the spectrum can be evaluated at any position just by using the FFT-coefficients. Since most of the energy is stored in the coefficients close to the actual position, it is sufficient to consider a few of them.