System Reliability for Stochastic Flow Networks and Applications

Abstract

This presentation will focus on system reliability evaluation for a stochastic flow network where each arc has several possible capacities with a (arbitrary) probability distribution. Many real-life systems such as computer systems, telecommunication systems, urban traffic systems, logistics systems, electric power systems, and etc can be regarded as stochastic flow networks. System reliability is the probability that the system can transmit a given units of data/commodity from the source(s) to the sink(s) under some constraints. These constraints may be budget, time, or length. In this talk, several models and algorithms are introduced for solving different practical flow systems. The minimal paths and minimal cuts approaches are involved in these algorithms. System reliability can be represented as a union of subsets, and a method named RSDP can then be utilized to evaluate system reliability efficiency. Some applications are introduced as well.