Analytical methods for planning for, and recovery from, multiple network or system failures due to nature or sabotage

摘要

Planning for recovery from exogenous or endogenous failures or outages of complex networks, including communications networks, power stations and entire grids as well as physical-transport networks such as pipelines, trucking, airways and railroads—for the purpose of smartly executing recovery—present planning difficulties because of the combinations of events which can occur, whether random, weather-induced, cascading or the effects of well-planned terrorism. These are poorly modeled by simple “Monte Carlo” runs and probability calculations because multiple events are rarely statistically independent, especially when (1) smart terrorists are the protagonist, or (2) when cascade effects occur, as at Fukushima. We discuss a mathematical model based on this author's analogous communications network model, CAINS (Communications and Information Network Solver) for assisting in the forward and contemporaneous evaluation and planning of backups and responses to such events. Abjuring Monte-Carlo simulation for stated reasons, the model is analytical and statistical in nature, avoids the problems with Monte-Carlo run length and chaotic algorithms, and thus runs very rapidly when planning for large numbers of failure modes.