Energy efficiency and surveillance applications in mobile sensor networks.

摘要

Sensor movement is a basic feature of mobile wireless sensor networks. It has been the driving force in the design of diverse network algorithms. In this dissertation, we characterize the fundamental statistical properties of a trip-based stochastic movement model and apply these movement properties in our study of two critical research issues in mobile sensor networks: (1) energy efficiency and (2) surveillance of network areas.;Energy efficiency in network communication is critical for wirelessly connected sensors which run on limited power supply. Meanwhile, transmission energy can be reduced significantly by reducing the communication distance between sender and receiver if the communication is postponed until they move near each other. In this dissertation, we develop a tight analytical lower bound (4% away from the measurement lower bound) of expected communication distance within given communication deadline constraints. An absolute performance measure shows that a previously developed least distance (LD) postponement algorithm can achieve an average communication distance reduction within 75% to 94% of the theoretical optimal lower bound. Since a mobile sender tracks its movement to postpone communication, we present an adaptive scheduler to determine an effective position sampling schedule under given changing operating conditions to further save energy on sensors. The proposed system has been implemented on an actual sensor network platform, and the measured total energy use of the system is reduced by up to 55%.;For the problem of mobile sensor based surveillance of geographical regions, we develop concepts of network coverage by a set of mobile sensors for given areas of interest (AOI). We further study the problem of a mobile target (the "mouse") trying to evade detection by the surveillance mobile sensors (the "cats") in a closed network area. We view this problem as a game between two groups of players: the mouse and the cats. We divide the problem into two cases based on the relative sensing capabilities of the cats and the mouse, and provide optimal movement strategies in each case for the mouse to maximize, and for the cats to minimize, the detection time.