CRISP 2.0 Type 1: Collaborative Research: Distributed Edge Computing to Improve Resilience of Interdependent Systems

点击翻译按钮获取中文摘要

This Critical Resilient Interdependent Infrastructure Systems and Processes (CRISP) research advances the body of knowledge on interdependent infrastructure resilience of systems through utilizing distributed assets to minimize cascading failures under extreme events. It is hypothesized that the domino effect of service disruptions is rooted in the vulnerability of the backbone-versus-edge relationship among the systems. For example, a backbone component in one system, such as natural gas-fueled power plants, is only at the end of the supply chain of natural gas (termed as the edge). Consequently, a backbone failure in one system (such as natural gas pipeline outage) can create the domino effect of failures through the entire interdependent systems. One way to alleviate this backbone-vs-edge tension is to bring assets to the edge (referred to as distributed resources), hence releasing the reliance of one infrastructure system on the others. This research will establish a new framework to effectively coordinate among the distributed resources, without requiring centralized coordination. Such a framework will be tested under various hazards including urban droughts, hurricanes and earthquakes. In addition, economic benefits of the added resilience will be quantified to help policy makers with more efficient solutions for improving resilience without sacrificing economic growth. The research will be widely disseminated to scientific communities and public via publishing in scientific outlets as well as leveraging press releases and media tools. Moreover, this research-integrated program and commitment to enhanced diversity promises to inspire underrepresented groups in STEM, and train the next generation of interdisciplinary scholars.To effectively control distributed resources across multiple interdependent systems, a novel distributed optimization algorithm will be established. Most of the existing distributed optimization algorithms cannot deal with complicated (and possibly non-convex) network constraints. To bridge this knowledge gap, there is an algorithm which leverages successive convex approximation, coupled with suitably designed message passing protocols, to allow an optimization problem to be solved in a distributed manner at a large number of computational nodes (i.e., the distributed resources), connected by a network with arbitrary topology and time-varying links. This is particularly useful in modeling outages in physical or communication networks, such as electricity network interruptions. To quantify the benefits of the distributed assets under extreme events, a multi-dimensional resilience quantification framework will be developed to simultaneously characterize multiple performance metrics of systems as opposed to measuring a single performance metric which is the prevalent approach today.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.