Improving Resilience in Large Scale Cyber-Physical Networks
Resilience in cyber-physical systems such as electrical power systems is of paramount importance for the socio-economic welfare of the society. Based on North American Electric Reliability Council (NERC), power systems are designed to be single fault tolerant systems. However, multiple failures do occur and result in cascading failures that cause severe system blackouts. These failures can result from both physical faults and cyber-effects that can be easily triggered via cyber-attacks which have become more prevalent and feasible. Due to the large number of autonomous components, smart grids are becoming more vulnerable to such failures that increases system complexity and makes resilience a complex problem. Several analysis models have been developed to analyze these failures however; they have their own limitations. These models focus mostly on time-independent physical failures and analyze the system from only one aspect which greatly limits the analysis. In addition, considering the large scale of the power system networks, several multiple contingencies do occur as a result of both physical failures and cyber-attacks which can be static or dynamic in nature that causes severe system damage. However, due to the computational complexity, it is very difficult to identify the critical components using an exhaustive analysis. Moreover, we also face the challenge of effectively deploying the limited defense resources to protect the critical points in the network under financial budget constraints. Therefore, in this work we present heuristical and game-theoretical based approaches to address the above-mentioned challenges. As a first step in the process, we have developed models and tools that can include cyber failures in addition to physical failures and provide the capability to initiate faults at different time instants. Next, we designed a framework to perform multi-platform analysis using Domain-Specific Modeling Language (DSML) based approach that integrate various exogenous tools together to perform a richer analysis. Further, we created mechanisms to optimally identify higher order critical contingencies using heuristical approaches. Finally, we developed game-theoretic based approaches towards identifying the effective deployment of the limited defense resources in order to improve the overall system resilience.