Towards a Tailored Sensor Network for Fire Emergency Monitoring in Large buildings

Abstract

In this presentation, we describe some of the ongoing efforts in developing a wireless sensor network tailored specifically for fire emergency monitoring. Network simulations of a dense sensor network with a flat architecture and using off the shelf protocols such as Wi-fi and Zigbee reveals the serious performance degradtion due to power consumption and packet losses caused by simultaneous collisions. By performing computational simulations of several fire scenarios, we show that spatio-temporal collisions exist in the sensed data. These can be leveraged in communication protocols for more efficient transmissions. While we do not propose algorithms for achieving this, our study highlights the need to consider statistical properties of the fire data in the communication protocols.