|
Abstract:
|
Networks, be they social, communications, or transportation, can experience failures. Such failures can be due to adversarial or natural actions. Assessing the survivability of networks is therefore an activity that is germane. Graph and network theorists have assessed network reliability by assuming that the failure propensities of the network's nodes are known, and that the failures are independent. This is an idealization. In this talk, I shall make a distinction between network survivability and network reliability and then point out issues that need to be addressed to obtain the former. I shall make the argument that node interdependence is hierarchical and that independence can only be claimed under independence at both the two stages of the hierarchy. The role of Markov chain Monte Carlo for putting my ideas to work so computable results can be obtained and highlighted.
|
