Wireless sensors are often deployed in remote areas to monitor and detect interesting events. For long-term monitoring of these events, it is necessary for sensors to have perpetual operation. Hence, they are equipped with batteries that recharge using renewable resources. The work in this thesis considered the problem of detecting changes in the state of event process (referred to as state transitions) so that the number of redundant transmissions is reduced. The objective was to maximize the number of transitions detected and transmitted under energy constraints. Two types of transitions were considered: transition transmitted immediately and transition transmitted with a delay. Transitions transmitted immediately reap the maximum reward, while late transmissions are modeled to reap a reward that decreases exponentially with the delay in transmission. The problem was formulated as a partially observable Markov decision process(POMDP), and the optimal policy (maximizes the average reward over time) was evaluated using value iteration. An approximate solution for the optimality equation was formulated, and the applicability of the approximate solution under various state space categories was discussed. Motivated by the structure of the optimal policy, a simple near-optimal policy that is asymptotically optimal was proposed.

Thesis (M.S.)--Wichita State University, College of Engineering, Dept. of Electrical Engineering and Computer Science.