Distributed Online Optimization in Dynamic Networks
Office of Naval Research
09/01/2012 - 09/30/2017
Multi-agent networked systems consisting of dynamic nodes are envisioned to play a crucial role in missions of great interest to DoD. A number of such missions, such as reconnaissance and search and rescue operations, can be approached by expressing their critical mission tasks as large-scale optimization problems whose data is only revealed over time. In this project we build on the recent research advances in multi-agent networked systems on one hand, and online algorithms on the other, to propose a comprehensive analytic framework for reasoning about the role of information geometry across time and space for the efficient operation of multi-agent networks in uncertain environments.