Vulnerability analysis of complex network infrastructures using a genetic algorithm

This paper proposes a method for analyzing the vulnerability of network infrastructures. The method uses a genetic algorithm for finding cross-sections that block delivering resources from their sources to consumers. The well-known approaches to solving network problems based on combinatorial and evolutionary approaches are considered. A feature of the proposed method is the fitness function chosen as an algorithm for calculating the number of paths in the graph when isolating the graph vertices that make up the individual. The graph reachability matrix and simple mathematical operations are adopted to optimize the fitness function and calculate the number of paths. The efficiency of the genetic algorithm compared to combinatorial methods is shown: multiple failures are found significantly faster than using exhaustive search algorithms.