Low Complexity Adaptive Beamforming Applied to Sonar Imaging (Invited)

Abstract

Modern sonars provide high-resolution acoustic imaging for a range of applications, including pipeline inspection, harbor surveys and seabed mapping. Central in the signal processing of the sonar data is the beamforming, where conventional (fixed) aperture tapering (or weighting) typically is used to set the equipment’s resolution and robustness to appropriate values. This means that there are predefined weights, not necessary optimally chosen, that express a compromise between resolution and sensitivity to noise and interference. In this work, we show how we can increase the resolution of bottom detections, without reducing the robustness, by applying Low Complexity Adaptive (LCA) beamforming. LCA can be viewed as a low computational cost version of the minimum variance distortionless response (MVDR) beamformer. It is easy to implement and improves the imaging process by adaptively choosing an appropriate weight for any point in time and space. We apply LCA beamforming on measured and simulated data. The simulated data are generated by the freely available ultrasound simulator “Field II”. The measured data was collected by a commercial echosounder. LCA beamforming gives significant reduction of the mainlobe width and sidelobe level over conventional beamforming. LCA also gives clear improvements for amplitude detections, and phase detections on measured data. The phase detections on simulated data are partially worse. We expect to get the same improvement for both datasets after minor modifications of the LCA algorithm.