Blind resampling parameter estimation for doubly selective underwater acoustic channels (Invited Paper)

Underwater acoustic channels are well modeled by different Doppler scaling per path, a generalization of the commonly employed model with equal Doppler scaling on all paths. The path dependent Doppler shifts and wideband channel, destroy carrier orthogonality and yield severe inter-carrier interference in an Orthogonal Frequency Division Multiplexing system. Resampling is typical front-end processing for signals over a common Doppler shift channel and this work examines the choice of resampling factor for the distinct Doppler per path scenario. Two criteria are derived to evaluate the optimal resampling parameter, one using sufficient statistics and the second using a matched filtering interpretation of resampling. The filtering interpretation is then used to derive a blind estimator for the optimal resampling parameter. Simulation results show the for small to moderate Doppler spreads, the blind estimator significantly outperforms the classical packet length based Doppler scaling estimator in many operating regimes.