A Hybrid GA for the Edge-Biconnectivity Augmentation Problem

In the design of communication networks, robustness against failures in single links or nodes is an important issue. This paper proposes a new approach for the NP-complete edge-biconnectivity augmentation (E2AUG) problem, in which a given graph G0(V, E0) needs to be augmented by the cheapest possible set of edges AUG so that a single edge deletion does not disconnect G0. The new approach is based on a preliminary reduction of the problem and a genetic algorithm (GA) using a binary vector to represent a set of augmenting edges and therefore a candidate solution. Two strategies are proposed to deal with infeasible solutions that do not lead to edge-biconnectivity. In the first, more traditional variant, infeasible solutions are detected and simply discarded. The second method is a hybrid approach that uses an effective heuristic to repair infeasible solutions by adding usually cheap edges to AUG until the graph augmented with AUG becomes edge-biconnected. The two GA-variants are empirica...

• Type: Conference
• Year: 2000
• Where: PPSN
• Authors: Ivana Ljubic, Günther R. Raidl, Jozef Kratica
• Keywords: Cheapest Possible Set, Distributed And Parallel Computing, Infeasible Solutions, NP-complete Edge-biconnectivity Augmentation, PPSN 2000