A Connected-Component Improvement Procedure for Relative Isolation Probability Calculation
Abstract
Due to the prevalence of natural disasters, the resilience of networks that provide essential services to the population has been studied. A recent contribution is the probability of relative isolation, which measures the likelihood of a demand node to be unreachable by supply nodes. The existing procedure for calculation is an enumerative decision-diagram-based approach that may be resource-intensive especially on occasions when time and memory constraints are imposed. In response, the study reverse-engineers the current method towards increased efficiency. The proposed solution hinges on the breadth-first search, producing connected components based on the sources. The algorithm is then benchmarked on foundational networks from literature, presenting a 99.77% decrease in computation time on average.
