The goal of our program is to study the effects of a dynamic mobile network environment on the computational performance of certain classes of decentralized algorithms in the context of variable volatility of wireless links and time-varying network topologies. These are algorithms that are inherently critical for various high-level distributed autonomy tasks, for instance online real-time machine learning, control, planning, localization and mapping in collaborative multi-agent platforms such as teams of Unmanned Aerial Vehicles (UAVs), or other autonomous robotic systems. The main focus will be on algorithms for decentralized numerical optimization, but other classes, such as average consensus, or decentralized auction algortihms, will also be considered. We are particularly interested in evaluating and quantifying the algorithms' performance in terms of throughput, robustness, and resilience under different realistic scenarios, as well as in developing rigorous empirical models and guidelines for future deployment on similar architectures. This research is similar in spirit to the seminal study of the performance of thirteen fundamental parallel algorithm classes conducted at the UC Berkeley a decade ago .
In order to ensure robust, reproducible experimental environement, an integrated mobile edge computing system emulator based on EMANE (network communications) and SCRIMMAGE (multi-agent autonomous systems) will be developed and will serve as a virtual testbed protoype. A state-of-the-art NVIDIA Jetson TX2 platform will be used for hardware-in-the-loop testing. This research will involve algorithm development and implementation, conducting mobile network experiments and evaluations, innovative data analysis, model extraction and validation, as well as presenting and communicating findings.
 A. Nedic, A. Olshevsky and M. G. Rabbat, "Network Topology and Communication-Computation Tradeoffs in Decentralized Optimization," in Proceedings of the IEEE, vol. 106, no. 5, pp. 953-976, May 2018.
 Boyd S, Parikh N, Chu E, Peleato B, Eckstein J: "Distributed optimization and statistical learning via the alternating direction method of multipliers." Foundations and Trends in Machine Learning 3(1): 2010
 Asanovic K, et al: "The Landscape of Parallel Computing Research: A View from Berkeley." EECS Department, University of California, Berkeley, Technical Report No. UCB/EECS-2006-183: December, 2006
Collaborative autonomy; Distributed computing; Computational performance evaluation; Mobile computing; CUDA architecture; Decentralized learning; Consensus-based distributed optimization; Decentralized algorithms; Networked UAVs; Robotic communication; Multi-agent autonomous systems