Presenter : Anais DURAND - MC LIMOS, Université Clermont Auvergne
Abstract : Mobile networks, robot fleets, swarms of drones are example of highly dynamic networks, i.e., networks whose topology changes overtime. In these networks, topological changes cannot simply be considered as a perturbation but are truly inherent to the system. However, fault-tolerant distributed computing researches mainly focused on networks whose topology is static or systems where topological changes are rare and sparse.
Self-stabilizing systems are able to recover a correct behavior in finite time and without any external intervention, after some perturbations (e.g., memory corruption, message losses) put the system in an arbitrary state. Self-stabilization is a very promising solution to withstand faults in highly dynamic networks.
I will present some ongoing work aiming to explore the power of self-stabilization in highly dynamic networks (what problems can we solve? what hypotheses on the dynamic of the system are necessary?, etc.). We consider Time-Varying Graphs to model the dynamic of the network topology.