M. Wazenski
D. Alexandrou
Dept. of Elec. Eng., Duke Univ., Box 90291, Durham, NC 27708-0291
Active, wideband detection of targets in a dense multipath environment is approached via optimal detection and estimation theory in conjunction with a received signal model described by weighted and delayed replicas of the transmitted signal. The set of weights and delays are known to vary with acoustic medium properties including sound-speed profile, bottom composition and topography and surface area, which are largely unknown or uncertain. The optimum receiver is based on prior distributions of multipath amplitude and delay. An acoustic propagation model drives the optimum receiver by generating these distributions through Monte Carlo simulation. Unlike traditional matched field techniques, the acoustic medium is treated as a random process rather than a deterministic one for a given set of propagation conditions. The sonar system toolset (SST)/generic sonar model (GSM) software package provides the required simulation environment. Illustrative detection examples are presented along with receiver operator characteristic (ROC) performance. [Work supported by Digital System Resources.]