 |
Center for Nonlinear Studies |
In our research, networks of interest are those with sources and sinks such as power systems, for example. The survivability of a power system is defined as its ability to deliver power to loads in the amount sufficient to satisfy the loads’ demand in a presence of multiple simultaneous faults. A number of faults and their locations within a system are assumed to be unpredictable. Such faults are typically caused by adverse events. They can also be accumulated with time without repair or result from a combination of predictable and unpredictable factors. Regardless a cause of faults and their evolution within a system, only an ability of a system to withstand an arbitrary combination of faults is of interest in our study.
Intuitively, it is clear that there is a relation between the network’s survivability and the network’s topology: types of nodes, numbers of nodes of different types, and their connection with one another. A goal of our study is to quantify this relation. A probabilistic framework for the quantitative analysis of the topological network survivability will be presented. Computational challenges associated with the problem will be discussed. “Selfish” algorithm for reducing the computational cost of the survivability analysis will be introduced. An example of a bio-inspired network topology of enhanced survivability will be given.
Host: Ivan Christov