Signal recovery in a reverberation-limited environment

The problem of recovering signals masked by reverberation is considered. Reverberation data from a shallow-water active sonar experiment in conjunction with simulated echoes are used to examine the potential for signal recovery offered by adaptive filtering and prediction. The deterministic least squares lattice filter is the central adaptive estimator of choice. The prediction error lattice is used to selectively "whiten" the composite process by controlling the algorithm adaptation speed. This is shown to result in significant signal enhancement for low-Doppler echoes masked by reverberation. Adaptive noise canceling with multiple reference beams is shown to be successful in extracting even zero-Doppler echoes from the reverberation background.