Adaptive Beamforming and Autocalibration for Swath Sonars

Accurate knowledge of the seabed is of vital importance for many human endeavors. Applications range from safe navigation to climate change models. Swath sonars are a key tool for efficient and high-resolution mapping of the seabed. This thesis aims to improve the quality of swath sonars by improving the beamformer, which is a key part of current signal processing. We explore two methods: Adaptive beamforming and autocalibration. Adaptive beamforming improves the beamforming process by adapting the beamforming to the received signal. We investigate how the adaptive Capon and Low Complexity Adaptive (LCA) beamformers can improve swath sonar beamforming on both simulated and field data, and their effect on the water column image and bathymetry. The Capon beamformer is well-tested and can give high performance, but has a high computational load and may have robustness issues. LCA is a recently developed and related adaptive beamformer which may be more robust and faster, with similar performance in many ways. We find that both beamformers improve resolution, edge definition and sidelobe level in the water column, and give more accurate amplitude detections. This leads to better defined features, better separation of features from the background, and sometimes detection of features not visible with the conventional delay and sum (DAS) beamformer. Capon has better resolution, somewhat better edge definition, and somewhat higher sidelobe level than LCA. We also find that an adaptive beamformer may improve interference rejection for phase detection, but generally reduces accuracy in the current configuration. This seems to be a side effect of the improved edge definition, and the effect can be reduced by reconfiguring the beamformers. Autocalibration estimates calibration errors without external reference sources. The errors, which particularly limit the sidelobe level, may then be compensated for. We model the errors by a complex factor per element and estimate them using data available during normal surveys. The method is based on the Generalized Interferometric Array Response. On simulated data, we are able to lower the sidelobe level below 50 dB. On field data, the sidelobe level is generally reduced, but the effect is much smaller. However, some sidelobes are unchanged and new sidelobes occasionally appear. We suggest that the reduced performance in the field is due to an insufficient calibration model.