Blind adaptive equalizer for underwater communications

This paper presents a novel blind adaptive equalizer for underwater communications. By exploiting the cyclostationary property of modulated signals, the proposed equalizer uses the phase-locked loop (PLL) to estimate the carrier signal and generates a transformed signal having spectral lines at carrier and baud-rate related frequencies. In order to update coefficients of the feedforward filter of the equalizer, the mean square error between the transformed signal and the carrier is minimized using the least-mean-square (LMS) algorithm. For decision-feedback equalization, a feedback filter is followed to remove that part of the intersymbol interference from the present estimate caused by previously detected symbols. As compared to traditional decision-feedback equalizers, the proposed equalizer requires no training sequence and exhibits faster convergence speed and lower bit-error rate beyond a certain threshold signal-to-noise ratio. Simulation results are provided to evaluate the effectiveness of the proposed equalizer for various scenarios involving stationary and nonstationary channels.