Using Wireless Sensor Network for Wildfire detection . A discrete event approach of environmental monitoring tool. Thierry Antoine-Santoni Jean François Santucci Emmanuelle de Gentili Bernadette Costa University of Corsica University of Corsica University of Corsica University of Corsica UMR CNRS 6134 UMR CNRS 6134 UMR CNRS 6134 UMR CNRS 6134 20250 Corte 20250 Corte 20250 Corte 20250 Corte FRANCE FRANCE FRANCE FRANCE antoine-santoni@univ-corse.fr santucci@univ-corse.fr gentili@univ-corse.fr bcosta@univ-corse.fr Abstract – In the Mediterranean area, the wildfire is a very important and critical problem. Environmental monitoring by using of sensors network seems to be a significant alternative of current detection techniques. Wireless sensor networks have been recognized as one of the most important technologies for this Century. Wireless sensor network represents a real and adapted solution to this problem. Test showed the sensor capacity to transmit the data wildfire in real time. However, we don’t know if this capacity is reliable with a large number of nodes. We introduce the DEVS discrete event formalism, and to apply it to the modelling and simulation of a great sensor network. The improvement of routing protocol and data transmission quality in the wildfire phenomenon is the base of our research. I. INTRODUCTION The wildfire is a critical problem which causes each year ecological and material disasters. In fire protection, the France government services take three types of actions: fire prevention education for the public, vegetation management and surveillance of forested areas for early detection. In the Mediterranean climate defined as a “red” zone of high fire risk, detection is based on a system of fire towers, fire spotting planes and mobile ground units of firemen patrolling the forest. The need of a reliable and fast detection tool really exists. A wildfire application and a development of detection tool appear essential for many reasons: - Prevention: a technology to collect environmental data - Detection (fight): mesh network architecture to square a zone and inform firemen. - Monitoring: detecting critical areas with birth conditions of forest fire. Environmental monitoring must be environmentally appropriate, which requires easy to install, low maintenance, non-toxic and inexpensive instrumentation. In order to monitor wildfires and impending wildfire conditions, the Firebug project [1][2]decided to use wireless, low-power sensor technology to collect environmental data (temperature, humidity, barometric pressure). Sensor network are composed of a large number of sensing devices, which are equipped with limited computing capabilities [3]. A sensor nodes, using the recent advances in micro-electro- mechanical systems (MEMS), combines the abilities to compute, communicate and sense. Using the Crossbow GPS motes, the wildfire monitoring system of Firebug shows that is possible to follow the fire behaviour with a sensor network of ten nodes and to collect many data and its characteristics (origin point, birth conditions) of this phenomenon. In a sensor network, different functionalities can be associated with the sensor nodes[4]. In earlier works, all sensor nodes are assumed to be homogenous, having equal capacity in terms of computation, communication and power[4][5]. However, depending on the application a node can be dedicated to a particular special function such as relaying, aggregation[4]. These first experiments are very important but limited by many points. Indeed, we don’t know the real behaviour sensor network in a great area with a large number of nodes. The rate of failure, the time of transmission data, the evolutionary architecture are blurs concepts which are not verifiable with such experiments. Tests with thousands of nodes on hundreds of hectares seem difficult to implement, financially and humanly, for dubious results. It is in this precise case that simulation takes a particular importance. Capacity to model and simulate a large number of nodes under particular conditions is essential to validate a future wildfire detection tool. In all networks, routing protocol is essential in information diffusion. However sensor network are different because nodes are very prone to failure. In order to build a real detection tool, it is necessary to define a routing protocol reliable in wildfire conditions. In order to define such protocol a set of simulation involving real wildfire conditions are necessary. The only efficient way to perform such simulation is to use a modeling approach based on the DEVS formalism. This approach will allow us: - a modular representation of routing protocols - the interconnections of heterogeneous models representing the behavior of the different elements involved in the protocols architecture." - an automatic generation of the simulation algorithms from the routing protocols models. It is a modular formalism for deterministic and causal systems. It allows for component-based design of complex systems. The hierarchical nature of DEVS makes it ideal for describing a system like sensor network. The discrete-event nature improves the execution performance of a model like this, due to the asynchronous nature of the events occurring in sensor network. This paper is organized as follows: Section 2 is a survey of the Crossbow technology, hardware and software, which we choose. We present the MTS420 motes equipped by GPS material and TINYOS, an operating system for embedded systems developed by University of Berkeley. Section 3 is an overview of most popular routing protocol with a focus on two particular techniques. In Section 4 we introduce our sensor model with DEVS approach. The section 5 concludes this paper with an outlook on future research activities.