Connected tours for sensor networks using clustering techniques

We consider a new problem related to data gathering in sensor networks based on multiple mobile sink elements. The problem involves the planning of a set of tours (cycles or closed paths) that cover all sensors and are connected: Each tour must be connected to at least one other tour by including a common network location. These connection points are used as data relays. The mobile element gathering data from its assigned path can leave all collected data on the designated relay point, so that it is picked up by the mobile element on the overlapping tour. This way, all data can eventually be transported to the network gateway, or any other node in the network. We formally define this new connected tour cover problem, which is a generalization of the travelling salesman problem (TSP) and therefore inherits all the known hardness properties of TSP. Then we propose a heuristic that solves this problem based on clustering, and establish its practical performance using an extensive experimental evaluation.