System design and fusion techniques for multistatic active sonar

Recent advances in construction technology have led to the commission of new classes of submarines which are very silent and highly manoeuvrable. Their detection by purely passive means reaches only limited performance. Active sonar systems, instead, offer sufficient performance, especially when operating in a multistatic setup, i. e., using multiple and spatially distributed sources and sensors to activate and receive echoes. However, the ocean, and in particular in areas of shallow water, which are of operational interest, offers only noisy and fading channels for sound transmission. Furthermore, reverberation generates clutter contacts that add to the set of contacts originated by the target. Therefore, when applying data fusion and tracking to multistatic sonar data, it is necessary to take account of these stochastic features. In this paper, we discuss preliminaries for the construction of optimal fusion techniques reflecting the physical knowledge of the stochastic features in the underwater sound channel. We apply two different tracking techniques to data sets gathered from sea experiments: a version of the multi hypothesis tracking (MHT) algorithm, which has been extended for the multistatic measurements, and a version of the cardinalized probability hypothesis density (CPHD) filter, also extended for multistatic measurements and, especially, for the fusion of data from qualitatively different receivers.