2023_programme: Improving sonar array performances with environment-informed sub-array processing

• Session: 16. Trends and Advances in Array Signal Processing
  Organiser(s): Angeliki Xenaki, Peter Gerstoft and Eliza Michalopoulou
• Lecture: Improving sonar array performances with environment-informed sub-array processing
  Paper ID: 1983
  Author(s): L'Her Alexandre, Drémeau Angélique, Le Courtois Florent, Real Gaultier, Cristol Xavier, Stéphan Yann
  Presenter: L'Her Alexandre
  Abstract: The wide range of environmental fluctuations present in underwater acoustic propagation affects sonar performance. In particular, internal waves can cause sonar arrays to lose signal coherence, reducing their gain. \n\nOne of the most direct ways to mitigate this effect is to use sub-array processing. By taking the coherence radius (the average distance at which two sensors can be assumed to be coherent) as the parameter determining the length of a typical sub-array, this method allows for more conventional processing methods to be used while assuming a coherent field on each sub-array. The limitation with this method comes from the need to have a previous estimation of the coherence radius. In this work, we use the Canonical Correlation Analysis (CCA) method to learn a model between the Empirical Orthogonal Functions (EOFs) of the environment (e.g. temperature) and the coherence of the array. This model is used to infer the coherence of the signal from environmental EOF at a latter time. We can then take the inferred coherence radius to parameterize the length of the sub-arrays.\n\nHowever, smaller sub-arrays result in a loss of angular resolution that can be detrimental to source detection and tracking. Many methods can be used to compensate for this loss, such as sparse DOA estimation methods. We propose here an implementation of SoBaP (for Soft Bayesian Pursuit), a Bayesian algorithm from the literature, on each sub-array to offset this resolution loss.\n\nWe show on experimental data from the ALMA 2017 campaign that using sub-array processing and SoBaP together match the performances of conventional beamforming (CB) on the sub-arrays while gaining angular resolution. Additionnally, it gives us better ROC curve statistics than using CB or SoBaP on the full array. This study is a first step for an environment-informed array processing method.
    Download the full paper
• Corresponding author: Mr Alexandre L'Her
  Affiliation: ENSTA Bretagne, Thales DMS
  Country: France
  e-mail: