Sensor Scheduling Under Energy Constraints.

Sensor Scheduling;Parameter Estimation;Target Tracking;Discrete Search;Electrical Engineering;Engineering;Electrical Engineering: Systems

Wang, YiVan Oyen, Mark Peter ;

University of Michigan

关键词: Sensor Scheduling;    Parameter Estimation;    Target Tracking;    Discrete Search;    Electrical Engineering;    Engineering;    Electrical Engineering: Systems;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/86303/yiws_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Recent advancement of wireless technologies and electronics has enabled the development of low-cost wireless sensor networks (WSN).The development of wireless sensor networks has also been motivated by military applications such as battlefield surveillance and target tracking. They are now used in various application areas, including habitat monitoring, industrial process monitoring and control, environment monitoring, health care applications, home automation, and traffic control.In this dissertation we investigate sensor scheduling problems under energy constraints through three scenarios: stationary parameter estimation, dynamic parameter tracking and discrete search. We first formulate a stochastic resource allocation problem for the stationary parameter estimation scenario with a sensor-dependent, parameter-dependent observation model. With the Gaussian assumption and linear observation model, the original problem is equivalent to a deterministic resource allocation problem. We propose a greedy algorithm and identify conditions sufficient to guarantee its optimality. Thereafter we formulate the parameter estimation problem with a sensor-dependent parameter-independent observation model as a static allocation problem. We derive lower bound on the optimal performance and propose a preprocessing algorithm to improve the lower bound. We use the improved lower bound to evaluate the performance of the proposed greedy strategy. Subsequently, we investigate the dynamic parameter tracking problem and discover the structure of an optimal strategy. For the discrete search problem with multiple sensors, we develop an easily implementable greedy strategy and identify conditions sufficient to guarantee its optimality. We discuss the relationship between each problem and the multi-armed bandit problem.