Sensor scheduling, target detection and estimation in sonar applications.

摘要

This thesis is organized in two parts. First we deal with the problem of wide area surveillance with a limited set of sensors. A model of the traffic dynamics and a framework to optimize the resources for tracking the regional traffic are developed. Using simulated annealing and a performance evaluator, the available sensors are scheduled to make measurements at optimal times. This methodology is analyzed along with some alternative algorithms. The second part of the thesis is focused on target tracking in sonar applications. Because of the intrinsic nonlinearity of the problem, several algorithms are developed and analyzed for different target movement behaviors. Data association problem in multi target environments is also considered and a detection splitting algorithm is developed for this purpose. It is compared with PDA (Probabilistic Data Association).; These new algorithms are also applied to real data and the performance is analyzed.