EFFICIENT EVALUATION OF NETWORK ROBUSTNESS WITH A GRAPH

摘要

Likely disconnections or system-level-impact producing slowdown in a flow parameter in a network can be identified by annotating a graph representing the network with a weight to edges of a graph representing the network. The weight can be linearly or non-linearly proportional to the inverse of the severity of the impact, and/or linearly or non-linearly proportional to the inverse of the likelihood of disconnection. Minimal cut sets needed to generate a disconnection in the network are identified from the annotated network based on the weights on the edges. Subsets of each minimal cut set are generated and evaluated during a simulation. The subsets can correspond to almost isolated scenarios for the network. The scope of the simulations can be reduced by selecting the minimal cut sets employing the weight.