| Recent progresses in distributed robotics and low power embedded systems have led to the creation of mobile sensor networks. Controlled mobility, moving sensors intentionally, enables a new set of possibilities in sensor networks. One of the most important applications is in target tracking area. In this dissertation, we consider the problem of target tracking using different types of mobile sensors, ones that measure received signal strength (RSS) and others that measure the signal direction of arrival (DOA) from the target. For tracking, we propose the use of sequential Monte Carlo (SMC) methods, also referred to as particle filters (PFs), where the positioning of the mobile sensors is based on the predicted target positions. In deciding how to deploy the sensors, we have used the Cramer-Rao lower bound (CRLB) that we have derived for our scheme of RSS sensors tracking and measurement uncertainty derived from geometric dilution of precision (GDOP) for DOA sensors tracking. Besides standard particle filtering (SPF) methods, which have been successfully applied to a variety of highly nonlinear problems, for target tracking with mobile sensors we also propose a new class of SMC methods named cost-reference particle filters (CRPFs). When there is no knowledge about the probability distributions of the noise in the system, CRPFs have been shown to be a flexible and robust alternative to PFs. Also, we extend the proposed work to tracking of multiple targets. Both association free and association based models corresponding to RSS and DOA sensors respectively are applied with different implementations. The performance of the methods has been investigated by simulations and compared to tracking with traditional static sensor network. It is shown that introducing mobility to the sensors improves the tracking performance. Also, the robustness of CRPFs has been confirmed when there is uncertainty in the system probabilistic model. |
