Towards a robust algorithm for distributed monitoring of network topology changes

Dynamic sensor networks applied as health-monitoring assistance systems require a constant supervision to assure the availability of essential network channels during emergencies. Topology changes due to connection and hardware failures have to be observed and reacted to. For that, an algorithm is needed that monitors topology changes within the network automatically without human interaction. The Fiedler vector is a powerful heuristic for topology inference, e.g. to detect communication bottlenecks. Current algorithms for the distributed calculation of the Fiedler vector do not consider topology changes. In this paper, we provide a stable algorithm for the distributed, indulgent derivation of the Fiedler vector which is applicable within dynamic networks. Simulation results show its robustness to topology changes.