Markovian agents models for wireless sensor networks deployed in environmental protection

Wireless sensor networks (WSNs) are gaining popularity as distributed monitoring systems in safety critical applications, when the location to be controlled may be dangerous for a human operator or difficult to access. Fire is one of the major thread in urban as well as in open environments, and WSNs are receiving increasing attention as a mean to build effective and timely fire protection systems. The present paper presents a novel analytical technique for the study of the propagation of a fire in a wide open area and the interaction with a WSN deployed to monitor the outbreak of the fire and to send a warning signal to a base station. For the complex scenario under study, an analytical modeling and analysis technique based on Markovian agents (MAs) is discussed. It is shown that, even if the overall state space of the models is huge, nevertheless an analytical solution is feasible, by exploiting the locality of the interactions among MAs, based on a message passing mechanism combined with a perception function. - Highlights: • We present a revised theory of Markovian agent models, detailing the analysis techniques and its complexity • We a target a complex application of a wireless sensor network (WSN) that monitors forest fire. • The model captures the propagation of fire, heat, and the detection by the WSN. • We compute key performance indices such us the fire propagation front, and message travel time. • We perform an extensive set of experiments to study the effectiveness of the WSN in detecting forest fire