PhD Defense by Nelson Gregory Andriano

Event Details
  • Date/Time:
    • Friday July 2, 2021
      2:00 pm - 4:00 pm
  • Location: Atlanta, GA
  • Phone:
  • URL: Bluejeans
  • Email:
  • Fee(s):
  • Extras:
No contact information submitted.

Summary Sentence: System of System Stakeholder Planning in a Multi-Stakeholder, Multi-objective, and Uncertain Environment

Full Summary: No summary paragraph submitted.

Nelson Gregory Andriano
(Advisor: Prof. Dimitri Mavris)

will defend a doctoral thesis entitled,

System of System Stakeholder Planning in a Multi-Stakeholder, Multi-objective, and Uncertain Environment


Friday, July 2nd at 2:00 p.m. (EDT)

The United States defense planning process is currently conducted in a pseudo-consolidated manner driven by the JCIDS process. Decisions to invest in technology, develop systems, and acquire assets are made by individual services with coordination at the higher joint level. These individual service’s decisions are made in an environment where resource allocation and need are influenced by external stakeholders (e.g. shared system development costs, additional levied requirements, and complementary system development). The future outcome of any given decision is subject to a high degree of uncertainty stemming from both the stakeholder execution of a decision and the environment in which that execution will take place.  Uncertainty in execution stems from TRL advancement, development timelines, acquisition timelines, and final deployed performance. Environmental uncertainty factors include future stakeholder resource availability, the future threat environment, cooperative stakeholder decisions, and mirrored adversary decisions.

The defense planning problem can be described as an acknowledged System of Systems (SoS) planning problem. Today, methodologies exist that individually address SoS Engineering processes, the evaluation of SoS performance, and SoS system deterministic evolution. However, few approaches holistically address the SoS planning and evolution problem at the level needed to assist individual defense stakeholders in strategic planning. Current approaches do not address the impact of multiple-stakeholder decisions, multiple goals for each stakeholder, the uncertainty of decision outcomes, and the temporal component to strategic decision making.

This thesis develops and tests a methodology to address defense stakeholder planning in a multi-stakeholder, multi-objective, and uncertain environment.  First, a decision space is populated and captured via sampling a game framework which represents multiple stakeholder decisions and the decision outcomes over time. A compressed Markov Decision Process (MDP) based meta-model is constructed using state-space consolidation techniques. The meta-model is evaluated using a risk-based policy development algorithm derived from combining traditional Reinforcement Learning (RL) techniques with mean-variance portfolio theory. Policy sensitivity to stakeholder risk-tolerance levels is used to develop state-based risk-tolerance sensitivity profiles and identify Pareto efficient actions. The risk-tolerance sensitivity profiles are used to evaluate both state spaces and decisions spaces to provide stakeholders with risk-based insights, or rule sets, to support immediate decision making and risk-based stakeholder playbook development.

The capability of the risk-based policy algorithm is tested using both elementary and complex scenarios. It is demonstrated that the algorithm can be used to extract Pareto efficient decisions as a function of risk-tolerance. The state space compression is tested via the comparison of the loss of information between the risk-based policy solutions for uncompressed and compressed state space. The full methodology is then demonstrated using a full-complexity scenario based on the joint development by France, Germany, and Spain of the SoS based Future Combat Air System (FCAS). The full complexity scenario is used to baseline the risk-based methodology against current optimal policy solution techniques. A significant increase in resulting derived insights relative to optimal policy solutions in a high uncertainty scenario is demonstrated.



  • Prof. Dimitri Mavris – School of Aerospace Engineering, Georgia Institute of Technology
  • Prof. Daniel Schrage – School of Aerospace Engineering, Georgia Institute of Technology
  • Dr. Kelly Griendling – School of Aerospace Engineering, Georgia Institute of Technology
  • Prof. Mariel Borowitz – School of International Affairs, Georgia Institute of Technology
  • Gen. Phillip Breedlove – School of International Affairs, Georgia Institute of Technology

Additional Information

In Campus Calendar

Graduate Studies

Invited Audience
Faculty/Staff, Public, Graduate students, Undergraduate students
Phd Defense
  • Created By: Tatianna Richardson
  • Workflow Status: Published
  • Created On: Jun 25, 2021 - 11:48am
  • Last Updated: Jun 25, 2021 - 11:48am