1-4244-1455-5/07/$25.00 c 2007 IEEE Fire Detection in the Urban Rural Interface through Fusion techniques E. Zervas Dept. of Electronics TEI-Athens 12210 Athens, Greece Email: zervas@ee.teiath.gr O. Sekkas, S. Hadjieftymiades, C. Anagnostopoulos Dept. of Informatics and Telecommunications University of Athens 15784 Athens, Greece Email: sekkas,shadj, bleu @di.uoa.gr Abstract Fires are a common, disastrous phenomenon that con- stitutes a serious threat. Thus, early detection is of great importance as the consequences of a fire are catastrophic. Towards this direction the SCIER 1 project envisages the deployment of Wireless Sensor Networks at the “Urban- Rural-Interface” (URI) and uses sensor fusion techniques to enhance the performance of the early fire detection and fire location estimation processes. 1 Introduction Fires are a common, disastrous phenomenon that con- stitutes a serious threat. Because of their speed of spread and intensity they often lead to property damages, personal injuries and loss of human lives. Thus, early detection is of great importance as the consequences of a fire (indoor or outdoor) are catastrophic. Fires that occur in wildland (forests, etc.) could also affect inhabited areas. These areas are widely known as “Urban-Rural-Interface” (URI), i.e., zones where forests and rural lands interface with homes, other buildings and infrastructures. In the URI, fires are fre- quent due to the special nature of these zones; a fire could be the result either of human on one side of the zone, or could arrive from the other side (wildland). In such cases early detection leads to an efficient control of the fire and makes feasible the immediate evacuation of the entire area if this is considered necessary. Great technology effort has been in- vested on the design of systems for fire detection and moni- toring. Most of them make use of temperature and humidity sensors, smoke detectors, infrared cameras, etc. In addition, aerial or satellite images are frequently used for outdoor fire detection and monitoring. In [3] a fire detection system is 1 SCIER (Sensor & Computing Infrastucture for Environmental Risks) is partially funded by the European Community through the FP6 IST Pro- gram. The work presented in this paper expresses the ideas of the authors and not necessarily the whole SCIER consortium. proposed based on multi-sensor technology and neural net- works (NNs). The sensed data include environmental tem- perature, smoke density and CO density. The use of NNs for automatic detection of smoke is also proposed in [10]. A system for wildfire monitoring using a wireless sensor network (WSN) that collects temperature relative humidity and barometric pressure is described in [4]. The wireless networked nodes communicate with a base station that col- lects the sensed data. Satellite based monitoring [11] is an- other method to detect forest fires but the scan period and the low resolution of satellite images make this method in- capable for real-time detection. The authors in [2] and [7] propose systems based on infrared (IR) technology for the detection of fires. The SCIER project will design, develop, and demon- strate an integrated system of sensors, networking and com- puting infrastructure for detecting, monitoring and predict- ing natural hazards (fires, etc.) at the URI. The overall goal of the SCIER system is to make the neglected URI zone safer against any type of natural hazards or accidents using wireless sensor network technologies, fusion techniques to assess dangerous situations, and predictive models to esti- mate the evolution of the hazardous phenomena. The fu- sion techniques used in SCIER are implemented in a spe- cial component of the SCIER system: the Local Alerting Control Unit (LACU). LACU controls a Wireless Sensor Network (WSN) and is responsible for the early detection of potential fires, the fire location estimation and the sub- sequent alerting functions. For the detection phase, sen- sor readings are evaluated and probabilities are assigned on each situation. If the probability of fire event exceeds a pre- determined threshold, the system shifts to fire location esti- mation phase. In this phase the centre of the fire outbreak and the radius of the fire spread are determined. The rest of the paper is organized as follows: in Section 2 the SCIER architecture is described and the requirements concerning the topology of the wireless sensors controlled by LACU are defined. Section 3 discusses fire detection and fire location estimation based on sensor readings (tem- perature) and sensor location (GPS coordinates). Finally in