This project focuses on holistic decision making for network optimization.
MIT, along with Applied Communication Sciences (ACS), Apogee Research, University of Pennsylvania, and Texas A&M University are developing a solution called “Distributed Enclave Defense Using Configurable Edges” (DEDUCE) to address the problem of maximizing cumulative network utility (CNU) for the DARPA program, EdgeCT. MIT’s component of this project is the solver RADMAX: Risk and Deadline-Aware Planning for Maximum Utility. RADMAX models CNU maximization under uncertain network conditions as a distributed resource allocation problem that explicitly accounts for task priorities and deadlines. The RADMAX solver builds on technologies developed by MIT to map EdgeCT task priorities and deadlines (expressed within utility functions) to bounded levels of risk that the task will fail to achieve acceptable performance.