MMOSPA-based direction-of-arrival tracking with a passive sonar array — An experimental study

We consider the directions-of-arrival tracking of multiple narrowband waveforms impinging on a sonar array. As the captured measurements do not provide any information about the labeling of the objects, a challenge is to resolve the directions-of-arrival coming from closely-spaced objects. In order to prevent track coalescence, we pursue the recently introduced concept of Minimum Mean OSPA (MMOSPA) estimation and propose a Gaussian particle filter that is tailored to the Optimal Sub-Pattern Assignment (OSPA) distance for sets. The main contribution of this work is to illustrate the benefits of MMOSPA estimation with a real data set from a large-aperture passive sonar array mounted in the east of Ft. Lauderdale in Florida, USA.