Low-Power Distributed Event Detection in Wireless Sensor Networks

In this paper we address the problem of energy-efficient event detection in wireless sensor networks (WSNs). Duty cycling is a fundamental approach to conserving energy in WSNs. However, it brings challenges to event detection in the sense that an event may be undetected or undergo a certain delay before it is detected, in particular when sensors are low duty-cycled. We investigate the fundamental relationship between event detection and energy efficiency. Based on a simplified network model, we quantify event detection performance by deriving the closed forms of detection delay and detectability. We also characterize the intrinsic tradeoff that exists between detection performance and system lifetime, which helps flexible design decisions for WSNs. In addition, we propose a completely localized algorithm, CAS, to cooperatively determine sensor wakeups. Without relying on location information, CAS is easy to implement and scalable to network density. Theoretical bounds of event detection are also studied to facilitate the comparative study. Comprehensive experiments are conducted and results demonstrate that CAS significantly improves detection performance.