Robust scheduling for target tracking with a wireless sensor network (WSN), focuses on the deployment of a WSN in a remote area to monitor a set of moving targets. Each sensor operates on a battery and is able to communicate with reachable sensors in the network. The targets are typically moving vehicles (planes, trains, cars,…) passing through the area. In order to monitor the targets, an activation schedule is sought such that the sensor network is continuously collecting data about the targets. Additionally, the transfer of the data collected to a base station deployed near the network also has to be planned. In this work, we consider that the trajectories of the targets are estimated. i.e., during the mission, at each time instant t, there is a given position where the target is expected. However, such estimations are inaccurate and deviations can occur. In this work, we formulate the problem of spatial robust scheduling. The aim is to produce an activation schedule for the sensors such that the targets are covered as long as they remain no farther from their estimated positions than a maximized value, called the spatial stability radius of the schedule. Afterwards, we formulate the spatio-temporal robustness problem. It is a bi-objective problem, with a spatial stability radius and a temporal stability radius for covering delays and advances. Two algorithms are proposed to solve these problems, and we show their efficiency through several numerical experiments.

使用无线传感器网络（WSN）进行目标跟踪的稳健调度着重于将WSN部署在偏远地区以监视一组移动目标。每个传感器都依靠电池工作，并能够与网络中的可达传感器通信。目标通常是经过该区域的移动车辆（飞机，火车，汽车等）。为了监视目标，寻求激活时间表，以使传感器网络不断收集有关目标的数据。另外，还必须计划将收集的数据传输到部署在网络附近的基站。在这项工作中，我们认为目标的轨迹是估计的。即，在任务期间，在每个时刻t，则在给定位置可以实现目标。但是，这样的估计是不准确的，并且可能发生偏差。在这项工作中，我们提出了空间鲁棒调度的问题。目的是为传感器生成激活计划，以便覆盖目标，只要目标距其估计位置的距离不超过最大值（称为计划的空间稳定性半径）。然后，我们提出时空鲁棒性问题。这是一个双目标问题，具有空间稳定半径和时间稳定半径以覆盖延迟和进展。提出了两种算法来解决这些问题，并通过几个数值实验来证明它们的效率。