In this talk, we consider the application of the theoretical framework of self-organization to transport networks. “Transport networks” is understood in a quite general sense, namely network topologies that can be interpreted as graphs, on whose links entities are forwarded from one node to the next, until they reach their destination node. Both links and nodes are subject to capacity restrictions. This class of networks includes among others telecommunication (IP) networks as well as urban traffic networks where vehicular traffic traverses intersections controlled by traffic lights. The goal is to investigate the self-organizing optimization of such networks, namely the minimization of entity travel- or waiting times. Without depending on a central authority, each node (router, traffic light) uses local rules to determine when to allocate its processing capacity to which entities (IP packets, vehicles). Current research indicates that such a self-organizing transport network might not perform much worse than optimal solutions computed based on global knowledge (for example, a so-called green wave traffic flow determined by a central controller); moreover, the self-organizing network is adaptive, scalable and robust.
Johannes Göbel is a visiting student from the University of Hamburg at the Computer Science division. He is currently busy with his PhD, and this colloquium will discuss his proposed work for his doctoral studies.