최상호(Sangho Choe) ; 유정화(Jeong-Hwa Yoo) ; 폰수게수라니(Ponsuge Surani Shalika Tissera) ; 강조인(Jo-In Kang) ; 양희경(Hee-Kyung Yang)
In this paper, we propose an event-driven low-power low-cost wireless sensor network (WSN) for real time wildfire monitoring. The proposed system will be mainly deployed and operated in a fire alert (or monitoring) area that could be divided into several sectors, each sector consisting of a controller (AP: access point) and its controlled nodes. For the evaluation of the proposed system, we implement a test bed (TB) using Arduino Uno board that has a smoke detector and a ZigBee module. The presented sensor nodes that are operated in power saving mode (PSM) during most of idle period have a longer battery life. Once a node detects smoke sample (which is termed as sensData) that is larger than a given threshold, it will generate a transmission message, called msgSensInfo (msgSensInfo1 for node fusion or msgSensInfo2 for AP fusion), and transmit that message to the AP and then the remote server. When a sensing event or a wake-up message event occurs, the presented node system is woken up from PSM, and gives a prompt (real-time) response including sensing message msgSensInfo generation and its transmission to AP. The proposed system has a data fusion algorithm collecting and synthesizing either multiple spatial samples (in case of AP fusion) or multiple time samples (in case of node fusion) that improves the fire detection performance further. Therefore, compared to existing low-cost WSN schemes, the proposed low-cost WSN scheme would be better at the perspective of real time performance, cost-effectiveness, and life-cycle such that it would be a good candidate for wildfire prevention as well as initial fire fighting.