Towards improved sonar performance using environment-informed sparse sub-array processing

In underwater acoustics, environmental fluctuations can cause loss of coherence of sonar arrays. Among the various approaches that can be considered to mitigate their effects, sub-array processing is a popular and intuitive one. However, it naturally leads to a sub-optimal performance. In this paper, we propose to enrich the approach in two ways. A first improvement is made by linking the parameters of the approach to those of the environment through a canonical correlation analysis. This leads in particular to an environment-informed calibration of the sub-arrays. A second improvement, more classical, is considered by using a sparse regularization on each sub-array processing. We show on real data the relevance of these choices and discuss the perspectives they open.