Resilience and Cascades in Transportation Networks
IDSS PIs: Damon Acemoglu, Munther Dahleh, Emilio Frazzoli
Collaborators: Giacomo Como, Ketan Savla
Technological advancements in terms of smart sensors, high speed communication enabling the transfer of massive data sets in almost real time, and real-time decision capabilities have resulted in the emergence of large scale, complex, and optimized interconnected systems. While such systems perform well under normal operations, they can exhibit fragility in response to certain disruptions that can lead to system breakdown and cascades of failures. This phenomenon, referred to as systemic risk, emphasizes the role of the system interconnection in causing such, possibly rare, events. The ﬂash crash of 2010, the ﬁnancial crisis of 2008, the New England power outage of 2003, or simply the cascaded delays in air travel because of unexpected weather in a hub city like Chicago are just a few of many examples of systemic risk present in complex interconnected systems. The propagation of failure in such systems is not caused just by malicious interventions, but also it often results from the interaction of the interconnected subsystems as they respond to small disruptions.
Recognizing that there is no existing methodology that addresses this grand and holistic challenge, the research objective of this multi-disciplinary eﬀort is to create a foundational science that allows for measuring, predicting, and containing systemic risk. Such a theoretical development will emerge from an in-depth understanding of systemic risk in three critical complex systems: the future power grid, the transportation system, and the ﬁnancial market.
Such a transformative research will hinge on the following thrusts: 1) understanding propagation mechanisms of failure in large scale interconnected systems, 2) development of a theoretical foundation for inference and early detection of such mechanisms, 3) development of a methodology for reconﬁgurable decision systems, and 4) development of scientiﬁc methods for robust architectures.
By starting in the application domains, we will develop powerful abstracted models that will enable us to characterize systemic risk, develop techniques for prediction and early detection, and ﬁnally act to contain the risk. These models will be validated by both domain experts and experimental testbeds.
The proposed program embraces the nascent but important collision and synthesis of topics that have been driven by the widespread diﬀusion and adoption of networked systems over the last two decades. While new technologies have spawned no shortage of powerful and fascinating products and services that now pervade much of contemporary life, these developments have now run ahead of science and education, as is often the case during periods of extraordinary technological innovation.
These technologies have created complex systems through the real-time interaction between physical, cyber, and social networks that can exhibit unpredictable behaviors under seemingly benign disruptions. Although ideas and methods from systems and control theory, Economics and Finance, Operations Research, Network Science, and many other disciplines are all relevant to understanding such phenomena, they are inadequate in providing a framework for measuring, predicting, and containing the systemic risk of catastrophic failures. The foundational science developed within this framework will provide new paradigms and tools for the analysis and synthesis of man-made complex interconnected systems.
References and Related Content:
“Robust Distributed Routing in Dynamical Flow Networks – Part I: Locally Responsive Policies and Weak Resilience” – IEEE Transactions on Automatic Control, 2013.
“Robust Distributed Routing in Dynamical Flow Networks – Part II: Strong Resilience, Equilibrium Selection and Cascaded Failures” – IEEE Transactions on Automatic Control, 2013.
“Bayesian learning in social networks” – Review of Economic Studies, 2011.