Routing within a telecommunication network defines how the traffic flow is mapped on the network topology. The increasing commercial importance of the Internet together with a rising number of real-time applications make of the routing mechanisms as essential feature in the control of the network performance. The routing mechanisms involved allow to assign the network capacities, more or less efficiently, to the demands or requests of traffic. The routing choice has a direct impact on the existence and location of congestion within the network. A high level of congestion may decrease the grade of service (increased delays, packet losses, call blocking, etc). Today several mechanisms are proposed to increase the routing control and to optimize the network performance, without introducing complexity in the network management but keeping the actual routers configuration.
In this Thesis, we addressed the problem of local balancing in networks that use shortest path routing. In particular, starting from a given network topology, we are aimed in finding the link-distance metric assignment to all links in the network that minimize the maximum utilization and therefore minimize the congestion.
Since it has shown that optimizing the weight setting for a given demand of traffic between routers is NP-hard, we resorted to a local search heuristic. This method has been considered for OSPF and ECMP routing disciplines, and we have found a weight set that are within a few percent from that of the optimal general routing where the flow for each demand is optimally distributed over all path between source and destination.
2004. , 66 p.