Fachvortrag in englischer Sprache

Todays societies heavily depend on functioning networks for traffic, telecommunication, electricity, and logistics, just to mention a few. The reliable and cost-effective functioning of these infrastructures has become an important factor for our economies and people, making efficient planning methodologies for these networks indispensable. In this talk, we discuss some of the recent advances in the methodologies for optimizing the access parts of these networks, which typically account for the bulk of the overall network cost. Two of the major mathematical challenges in planning these networks arise from the often complex technical and operational restrictions that need to be respected and from the enormous size of the networks to be planned, which makes classical access network planning models invalid or computationally intractable. In the first part of the talk, we consider new variants of combined facility location and network design problems that include the most important structural constraints arising in telecommunication and logistics applications, such as modular and shared capacities and length restrictions. Combining techniques for approximating shallow-light Steiner trees, simultaneous approximation of shortest paths and minimum spanning trees, buy-at-bulk network design, and greedy coverings, we show how to efficiently compute solutions with proven quality guarantees for several fundamental variants of these problems. In the second part of the talk, we present a computational solution approach based on integer linear programming and Lagrangian decomposition for one particular problem arising in the design of fiber optical telecommunication access networks. This approach enables practitioners to (approximately) solve very large problem instances with very little computing time. This of particular interest in the early stages of the long-term strategic network planning, when numerous planning scenarios with varying technological assumptions and demand, cost, or revenue predictions are evaluated in order to identify the most important parameters and make strategic decisions concerning technology vendors, the use of existing or new infrastructures, or the long-term evolution of the network. At this stage, fast computational methods are required to make optimization methodologies accessible to the practitioners.

Speaker: Prof. Dr. Andreas Bley, University of Kassel