Towards Optimal Architectures for Hazard Monitoring Based on Sensor Networks and Crowdsensing

Abstract

The monitoring of hazards through the ability to detect events and predict spatial and temporal evolution of dynamical hazards still remains a big challenge for dynamic disaster risk assessment and mitigation. The goal of this paper is to show how well established methods arising from the control theory can positively contribute to dynamic risk assessment improvement through an effective hazard monitoring. More precisely, the objective is threefold. Firstly, the design of an optimal monitoring architecture is proposed based on the combination of optimal sensor placement and receding horizon observer design. In this paper, the focus is only made on model-based and data-driven approaches. The benefit of using sensor networks and crowdsensing techniques is also discussed. Secondly, the paper seeks to identify the application areas that can benefit from both optimal sensor location techniques and receding horizon observers, while reviewing already existing references. Thirdly, some personal contributions illustrating the proposed approach are presented. In particular, two case studies are presented: one considers the dynamic positioning of drones for monitoring air pollution, the other is dedicated to the early detection of a wildfire outbreak.