Approximation algorithms for constructing wavelength routing networks

Consider a requirement graph whose vertices represent customers and an edge represents the need to route a unit of flow between its end vertices along a single path. All of these flows are to be routed simultaneously. A solution network consists of a (multi)graph on the same set of vertices, such that it is possible to route simultaneously all of the required flows in such a way that no edge is used more than K times. The synthesis of wavelength routing network problem is to compute a solution network of minimum number of edges. This problem has significant importance in the world of fiber-optic networks where a link can carry a limited amount of different wavelengths and one is interested in finding a minimum cost network such that all the requirements can be carried in the network without changing the wavelength of a path at any of its internal vertices. In this paper we prove that SWRN problem is NP-hard for any constant K (K 2). Then we assume that GR is a clique with n vertices ...

• Type: Journal
• Year: 2002
• Where: NETWORKS
• Authors: Refael Hassin, Asaf Levin
• Keywords: Computer Networks, End Vertices, NETWORKS 2002, Requirement Graph, Solution Network