High-resolution broadband beamforming and detection methods with real data

Abstract

IntroductionAs the acoustic noise sources become quieter and quieter, detection and localization of the acoustic source become more and more difficult. The noise sources are often many dB quieter than the acoustic noise floor and the source noise field is unknown in that the signal strength and spectra of the source are unknown. In addition, the target is often submerged in multiple noise sources. On the other hand, spatial discrimination capability (the ability to separate two planewave signals whose arrival angles are closely spaced) depends on the size of the spatial aperture. We are constrained by the geometric and performance limitations of the available acoustic array. It is necessary to find good beamformer and broadband detection methods with high resolution to detect and track low signal-to-noise ratio (SNR) multiple targets.High-resolution adaptive beamforming techniques for underwater applications have been extensively studied for past several decades [1]. However, not many applications have been reported in the public domain. Furthermore, most of the works reported in the literatures were for narrowband array processing. Although broadband beamforming implement procedure has been reported [2-4], most of them are for conventional beamforming.The goals of broadband detection are to condense the information contained in an array's beamformer output and aid operators in detecting acoustic targets. Conventional Energy Detection (CED) is often implemented as optimal signal detection in uncorrelated noise fields. It provides good detection ability with uncluttered targets and the best theoretical Minimum Detectable Level (MDL) than all other energy detectors. However, it has wide blurry traces for loud targets and poor bearing resolution. It also has impaired detection ability in real world (cluttered) noise environments.Subband Energy Detection is another new class of important broadband processing method that provides improved performance to sonar systems [5]. One of them is called Subband Peak Energy Detection (SPED). SPED can provides thinner target traces, improved bearing resolution, and better detection advantage in real world (cluttered) acoustic environments. The author's group have given the principle and simulation results for SPED earlier in reference [6], results on real data will be presented in the paper.