Abstract (EN)

A wireless sensor network (WSN) is a set of sensors, randomly deployed in an area often hard or dangerous to access and without any infrastructure. Hence,the batteries of the sensors are not refillable which limits the lifetime of the network, i.e., how long it can operate. There are several types of sensors for different applications, and we focus in this work on the target tracking. In such applications, the network aims to monitor a set of moving targets (planes, trains,terrestrial vehicles,...), whose spatial trajectories are estimated. It means that, at instant t in the time horizon, we have an estimation of the position of each target. However, this estimation may not be accurate, and the difficulty of the problem is to cover the targets considering the highest possible deviation from their estimated trajectories. Moreover, in order to preserve energy in the net-work for future mission, at most one sensor per target should be used at anytime. Finally, all the data collected by the sensors have to be transmitted to abase station. The problem is to find a robust schedule to continuously monitor the targets and to transfer the data. This schedule is robust because it has to maximize a spatial stability radius, such that, it stays feasible as long as the targets are not deviated for more than the value of the stability radius from their estimated position. The targets are covered at every instant t as long as,they are located in the disc of radius equals to the stability radius and centered on the estimated position of the target. In this work, we propose (i) a discretization method on the geometric data, (ii) two upper bounds on the value of the stability radius, and (iii) a method that uses the discretized data and the upper bounds to compute a robust schedule.