Local event-based sensor activation for occupant distribution estimation in buildings

Occupant distribution in buildings is important both for energy saving and for fast evacuation. This information can be monitored by wireless sensors, which are easy to install but subject to limited battery. In this paper, we consider the activation policy optimization of wireless sensor network for occupant distribution estimation under limited battery. Three contributions are made. First, we focus on policies that activate sensors based on events. We then formulate the sensor activation problem as an event-based optimization. Second, a nested structure of the event-based policy space is introduced, which represents policies with different complexities. Third, the performance of these policies are compared numerically. It is observed that policies that activate sensors based on local events achieve a good balance between the power consumption and the estimation accuracy. We hope this work brings insight to sensor activation problem in general.