Group-centrality and impact of large-scale localized faults in spatial networks

Major catastrophes such as hurricanes and weapons of mass destruction are expected to introduce large-scale and localized faults on a network. Analyzing the impact of such faults in the network is extremely difficult and can also get computationally expensive in large complex networks. In this work, we address the issue of predicting the impact of large-scale localized faults by evaluating the centrality of the fault region. This work proposes mapping a network to network of overlapping regions to create failure scenarios. Well-known centrality measures are used over the region network to determine the centrality indices of the regions. The physical impact of the failures is measured by the relative loss in network measures such as connectivity and efficiency. The performance of the centrality measure is evaluated by the accuracy of its prediction. The experimental results confirm that the region-based centrality index of a region gives a fair prediction of the impact of the region's failure.