Improved single snapshot beamforming in a noisy environment using the cubic autoproduct

Abstract

The presence of random noise is a challenge for remote sensing tasks as success typically diminishes with decreasing signal-to-noise ratio (SNR). To improve performance in noisy environments, signal processing techniques commonly implement snapshot averaging, requiring multiple signal samples. This presentation investigates the improvement offered by the cubic autoproduct for low SNR beamforming with a single snapshot. Prior work has demonstrated the utility of a quadratic product of complex field amplitudes within the signal bandwidth for beamforming and matched field applications. This quadratic field product, known as the frequency-difference autoproduct, is a synthetic estimate of an acoustic field at the difference frequency of the two constituent fields. That formulation is extended here to a cubic product of three complex field amplitudes within the signal bandwidth, termed the cubic autoproduct, capable of mimicking field content at frequencies below, above, and within the signal bandwidth. Important features and the mathematical description of the cubic autoproduct are reviewed before discussing the beamforming approach. Single snapshot beamforming results from experimental data, acquired in a noisy underwater environment, are directly compared between the cubic frequency-difference autoproduct and the acoustic field. [Work supported by ONR and by the US DoD through an NDSEG Fellowship.]