Predictive Scheduling for Spatial-dependent Tasks in Wireless Sensor Networks

Recent advances in wireless energy transfer technology propels the developments of renewable sensor networks. To sustain the operation of a sensor network, a mobile charger is used to recharge each node. Consider each node’s recharge request as a task with a soft due time, the mobile charger needs to dynamically schedule these tasks. This scheduling problem is very challenging, since both the nodes’ due times and their locations have to be considered. Existing solutions to this problem usually assume fixed paths generated by the travelling salesman or Hamilton cycle algorithms. These solutions suffer from high deadline miss ratios. In this paper, we investigate maximum response ratio based scheduling algorithms that can dynamically select the path of chargers for higher network coverage ratio. To further improve the performance, we exploit the dependencies of different recharge tasks and predict the impacts of executing one charging task to the others. With extensive trace-driven analysis, our algorithms significantly outperform existing algorithms. Keywords—Vehicle scheduling, spatial dependency, wireless charging