Fine Frequency Offset Estimation for Frequency-Selective Channels

In this paper we address the problem of frequency offset estimation from non-periodic training sequences used in packet transmission over frequency-selective channels and additive white Gaussian noise. Based on polynomial approximations of the sin and cos terms in the joint likelihood function, we obtain closed-form expressions of the maximum-likelihood estimate of the frequency offset using standard optimization techniques. We show that, for small frequency offsets, the proposed solutions outperform conventional methods that use large size fast Fourier transforms (FFT). In addition, they can be used with small size FFT´s to cover a larger estimation range