A Pre-Whiten Method for Active Sonar in Reverberation Background

Matched filter, the optimal receiver in white noise, is not always optimum in reverberation background. A pre-whiten filter to restrain the reverberation is proposed. On the premise of Gaussian color noise model of reverberation with local stationarity, a method using all-pole partition pre-whiten filters is proposed based on autoregressive(AR) model, which is adaptive. Priori knowledge of the noise and reverberation are not required. An algorithm called reverse partition pre-whiten is also brought forward versus the order partition pre-whiten algorithm. Simulation results are presented to show that the method is effective in enhancing the echo-to-reverberation ratio, and the performance of reverse partition pre-whiten algorithm is better than the order partition pre-whiten algorithm.