CISE Seminar: Les Servi, Former Chief Scientist at The MITRE Corporation

A Framework for Identifying Critical Nodes in a Mission: Mission Dependency Network Analysis

In recent decades, two trends have accelerated in tandem: optimization methods and computing power have advanced dramatically, while operational missions increasingly depend on complex, tightly coupled systems of systems. The result is both opportunity and risk – opportunity to reason quantitatively about resilience and tradeoffs, and risk that small disruptions can cascade into outsized mission impact. These trends motivated the need for a broad modeling and optimization framework for identifying critical elements in such environments. This entails representing hierarchical and cross-layer dependencies—how component operability “rolls up” to subsystem and mission performance—using flexible dependency rules, and then translating those semantics into tractable mixed-integer linear programs (MILPs).

The analysis goes beyond a single notion of importance. “Criticality” is examined through multiple lenses (e.g., single points of failure vs. small interdiction/repair sets vs. expected marginal degradation). The talk also addresses imperfect knowledge of the dependency structure by incorporating uncertainty in the mapping using robust optimization ideas. Finally, it explicitly accounts for time to detect, diagnose, mitigate, and recover, so that criticality reflects not only what can fail, but what can be managed quickly enough to affect outcomes. Small illustrative examples demonstrate how these modeling choices materially change what appears “most critical,” and how optimization can support robust, defensible decisions under uncertainty.

This work aims to (i) help senior leaders see the “art of the possible,” (ii) provide practitioners with usable methods that can be adapted to real settings, and (iii) stimulate technical extensions by the research community. This talk is based on the book, Mission Dependency Network Analysis: Mathematical Foundations for Modeling and Practice | Springer Nature Link (formally SpringerLink)

Dr. Les Servi is an operations research scientist and former Chief Scientist for Cyber Operations Research at The MITRE Corporation. He is currently an Adjunct Professor at Boston College and member of the Board of Directors of INFORMS. Earlier, he conducted research at MIT Lincoln Laboratory, Bell Laboratories, and GTE Laboratories (now Verizon), and spent a sabbatical year as a visiting scientist at Harvard University and MIT.

His public-service contributions include membership on Defense Science Board task forces on counterinsurgency (2010–2011) and constrained military operations (2016). During the COVID-19 epidemic, he led the optimization-modeling effort for a multi-tier medical supply chain within the Under Secretary of Defense for Acquisition & Sustainment (USD(A&S)) Joint Acquisition Task Force. He received a Certificate of Appreciation from U.S. Director of National Intelligence, James Clapper. His research record includes five papers with 200+ citations each and 12 patents. He is an INFORMS Fellow (elected 2004), served six years on the INFORMS Board of Directors, and delivered a keynote at the 2023 INFORMS Annual Meeting. Within the Military Operations Research Society (MORS), he served six years on the Board, was elected President in 2024, delivered a keynote address in 2016 and was named a Fellow in 2025.

He has held editorial roles with Operations Research, Management Science, and the INFORMS Journal on Computing, and has served on Ph.D. committees at Harvard, MIT, and Boston University. He earned his Ph.D. in engineering from Harvard University.

Faculty Host: Christos Cassandras

Student Host: Lizzy Yue