Evaluating Heuristic Optimization, Bio-Inspired and Graph-Theoretic Algorithms for the Generation of Fault-Tolerant Graphs with Minimal Costs

Abstract

The construction of fault-tolerant graphs is a trade-off between costs and degree of fault-tolerance. Thus the construction of such graphs can be viewed as a two-criteria optimization problem. Any algorithm therefore should be able to generate a Pareto-front of graphs so that the right graph can be chosen to match the application and the user’s need. In this work algorithms from three different domains for the construction of fault-tolerant graphs are evaluated. Classical graph-theoretic algorithms, optimization and bio-inspired approaches are compared regarding the quality of the generated graphs as well as concerning the runtime requirements. As result, recommendations for the application of the right algorithm for a certain problem class can be concluded.