Risk is New Robustness: A Systemic Perspective
Abstract
Statement of Work:Develop a novel unified theory that is based on an axiomatic general theory of systemic measures towards an understanding of performance and risk in large networks of autonomous systems. Develop data-driven algorithms to assess global performance, risk, and fragility features of a network of LORCA Unmanned Surface Vehicles (USVs) by computing the value of its systemic measure in real-time.Objective:The goal of this project is to develop a novel unified theory based on an axiomatic general theory of ~systemic~ measures towards an understanding of performance and risk in large networks of autonomous systems. Systemic refers to the study of the effects of uncertainties that spread throughout the network affecting all subsystems. The notion of systemic risk describes fragility in interconnected systems that result in global or cascading failures due to either relatively small shocks at the subsystem level or larger and more malicious types of disruptions affecting the whole network.Approach:Developing the theory of systemic risk is a very challenging and needed research effort; especially since traditional approaches to optimal performance and uncertainty are not satisfactory for many application domains. The new theory will have a very broad range of applications that include networked autonomous vehicles, distributed emergency response systems, interconnected transportation networks, distributed smart energy networks, metabolic pathways,and social and financial networks. The classes of systemic measures in this proposal quantify the ways that the effects of external disturbances/shocks, various modeling uncertainties, and structural changes to the dynamics of the network propagate system-wide and affect performance and risk of the network as a whole. The PI will explore the connections between existing gold standard performance and risk measures that have been used in disciplines such as control anddynamical systems, physics, and biology, and construct a unified systems theoretic framework for characterization of such measures and their inherent fundamental limits and tradeoffs. The proposed research builds on the PI~s extensive preliminary work which has led to significant breakthroughs in distributed control and dynamical systems. The development of the theory involves integration of concepts from control theory, dynamical systems, optimization, probability, operator and graph theories. The primary focus of the proposed plan is on revealing foundational role of underlying dynamical structure and sparse information structure in dynamical networks in emergence of severetheoretical hard limits on the resulting performance and risk in such networks. To test the theoretical work, the PI plans to develop data-driven algorithms to assess global performance, risk, and fragility features of a network of LORCA Unmanned Surface Vehicles (USVs) by computing the value of its systemic measure in real-time.Overall Merit and ONR Mission/Relevance:This research addresses ONR~s Autonomy and Unmanned Systems focus areas. This work is expected to develop theory and algorithms for understanding failures of networked autonomous agents. This research is expected to develop a novel approach and omputational methods for studying potential risks andfailures of systems composed of large numbers of networked autonomous agents, and ways of mitigating such risks.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141612645
Entities
People
- Nader Motee
Organizations
- Lehigh University
- Office of Naval Research
- United States Navy