Optimum signal processing with suboptimum realizations to reject turbulent boundary layer noise

Abstract

Optimum beamforming or filtering of a line array of either two- or three-element closely spaced hydrophone clusters is considered. The particular topic of the paper is the cancellation of Turbulent Boundary Layer (TBL) noise but the signal processing approach applies to the general subsonic noise problem. Array gain for clusters of unweighted hydrophones and clusters of optimally weighted hydrophones are compared. Array gain improvement for two-hydrophone optimally processed clusters with high interelement TBL noise coherence relative to conventional (time delay and sum) single element beam-forming is shown to increase linearly with the TBL-to-uncorrelated noise ratio and quadratically with the magnitude coherence of the TBL process. Suboptimum realizations of the optimum cluster-element combining schemes such as noise cancellation are proposed. Those realizations are shown to result in very simple hardware implementations. The general noise cancelling approach for high TBL-to-uncorrelated noise ratios uses outputs of a subset of hydrophones in a cluster to estimate TBL noise in a single hydrophone. The estimate is subtracted from the single hydrophone resulting in a subtractor output that is ideally insensitive to the TBL process. It is shown that using three or more hydrophones in an optimally processed cluster is relatively less effective than the two-element cluster as the TBL noise level increases. This suggests that a simple two-element optimally processed cluster would be a cost effective technique for suppressing high level TBL noise because processing complexity increases as the square of the number of hydrophones in a cluster.