Hierarchical Censoring Sensors for Change Detection

Devices in energy-limited wireless sensor networks remain in a low-communication `sleep' mode until an alarm event is detected. It has been proposed to use `censoring sensors' to reduce the probability that a sensor must transmit in this mode, thereby minimizing energy consumption when alarm events are not occurring, and lengthening sensor lifetime. Further, since devices in multi-hop networks are not usually in single-hop range of a fusion center, hierarchical distributed detection can lead to further energy efficiency. We report on a system that applies censoring in a hierarchical network to the CUSUM test of Page and Lorden, an online abrupt change detector. In this paper, we explore via simulation an example change detection problem and demonstrate that significant reduction in the number of sensor transmissions can be achieved at the cost of a small increase in mean detection delay compared to uncensored change detection performance