Thèse soutenue

Optimisation dans les graphes avec contraintes sur le degré : application aux réseaux des télécommunications

FR  |  
EN
Auteur / Autrice : Ignasi Sau Valls
Direction : Xavier Muñoz LópezJean-Claude BermondDavid Coudert
Type : Thèse de doctorat
Discipline(s) : Informatique
Date : Soutenance en 2009
Etablissement(s) : Nice en cotutelle avec Barcelone, Universitat politécnica de Catalunya
Ecole(s) doctorale(s) : École doctorale Sciences et technologies de l'information et de la communication (Sophia Antipolis, Alpes-Maritimes)

Résumé

FR  |  
ES

The first part of this thesis is devoted to traffic grooming, which is a central problem in optical networks. It refers to packing low-rate signals into higher-speed streams, in order to improve bandwidth utilization and reduce the network cost. The objective is to minimize the number of Add-Drop Multiplexers (ADMs), which are devices that insert/extract low-rate traffic to/from a high-speed stream. In graph-theoretical terms, the problem can be translated into finding a partition of the edges of a request graph into subgraphs with bounded number of edges, the objective being to minimize the total number of vertices of the partition. We first focus in Chapter 1 on a general request graph when the topology is a ring or a path. We provide the first inapproximability result for traffic grooming for fixed values of the grooming factor C, answering affirmatively to a conjecture in the literature. We also provide a polynomial-time approximation algorithm for traffic grooming in rings and paths, with an approximation ratio independent of C. We introduce in Chapter 2 a new model of traffic grooming in unidirectional rings, in order to design networks being able to support any request graph with bounded maximum degree. We show that the problem is essentially equivalent to finding the least integer M(C,D) such that the edges of any graph with maximum degree at most D can be partitioned into subgraphs with at most C edges and each vertex appears in at most M(C,D) subgraphs, and we establish the value of M(C,D) for almost all values of C and Delta. In Chapter 3 we focus on traffic grooming in bidirectional rings with symmetric shortest path routing and all-to-all unitary requests, providing general lower bounds and infinite families of optimal solutions for C=1,2,3 and C of the form k(k+1)/2. In Chapter 4 we study traffic grooming for two-period optical networks, a variation of the traffic grooming problem for WDM unidirectional ring networks with two grooming factors C and C' that allows some dynamism on the traffic. Using tools of graph decompositions, we determine the minimum number of ADMs for C=4, and C'=1,2,3. The study of the traffic grooming problem leads naturally to the study of a family of graph-theoretical problems dealing with general constraints on the degree. This is the topic of the second part of this thesis. We begin in Chapter 5 by studying the computational complexity of several families of degree-constrained problems, giving hardness results and polynomial-time approximation algorithms. We then study in Chapter 6 the parameterized complexity of finding degree-constrained subgraphs, when the parameter is the size of the subgraphs. We prove hardness results in general graphs and provide explicit fixed-parameter tractable algorithms for minor-free graphs. We obtain in Chapter 7 subexponential parameterized and exact algorithms for several families of degree-constrained subgraph problems on planar graphs, using bidimensionality theory combined with novel dynamic programming techniques. Finally, we provide in Chapter 8 a framework for the design of dynamic programming algorithms for surface-embedded graphs with single exponential dependence on branchwidth. Our approach is based on a new type of branch decomposition called surface cut decomposition, which generalizes sphere cut decompositions for planar graphs. The existence of such algorithms is proved using diverse techniques from topological graph theory and analytic combinatorics.