Time-varying filter estimation for the deconvolution of environmental reverberation from active sonar returns

The estimation and removal of the time-varying two-way impulse response to environmental scatterers from broadband reverberation data is considered for increasing the signal-to-noise ratio of sonar returns from targets in the water column. Spectrograms of simulated and real reverberation time series data from active sonars in the mid-frequency range show strong evidence of interference patterns which give clues to the number of important paths to environmental scatterers as well as their depth in the water column. In this talk we consider the estimation of a time dependent de-convolution filter for the removal of these environmental reverberation returns from active sonar data. Issues regarding the degrees of freedom required for the efficient implementation of this filter and the stability of these estimates are considered. Simulation results are shown which demonstrate the potential gain of using this approach to partially null the impact of environmental scatterers in active sonar data.