Distributed Algorithms in Network Systems: Algorithms, Fundamental limits, and Applications

November 17, 2017, Webb 1100

Na (Lina) Li

Harvard University, Electrical Engineering


One central goal of multi-agent network systems is to achieve desired collective behaviors of the networked system through the design of admissible local agent control algorithms. This is nontrivial due to various challenges. In this talk, I will mainly focus on two challenges: one is local connectivity and one is limited communication. In particular, I will mainly present two lines of our recent work. In the first one, we consider the distributed optimization where a group of agents need to solve an optimization problem using local information and communication. I will present our work on bridging the gap between the convergence speeds of centralized gradient methods and distributed gradient methods. To the best of our knowledge, the algorithm is faster than any other distributed gradient-type algorithm that has been proposed so far. In the second one, I will present our work on distributed resource allocation under limited communication, with a particular application to energy management in power grids. I will first show how we can extract information from physical measurements and develop fast and closed-loop decentralized control algorithms. Then I will present how to recover information from local computation by studying a general class of quantized gradient methods and its use for resource allocation problems. We also investigate communication complexity of the problem, the minimal number of communicated bits needed to solve some classes of problems regardless of the used algorithms.

Speaker's Bio

Na Li is an assistant professor in Electrical Engineering and Applied Mathematics of the School of Engineering and Applied Sciences in Harvard University since 2014. She received her Bachelor degree in Mathematics in Zhejiang University in 2007 and PhD degree in Control and Dynamical systems from California Institute of Technology in 2013. She was a postdoctoral associate of the Laboratory for Information and Decision Systems at Massachusetts Institute of Technology 2013-2014. Her research lies in distributed optimization and control of cyber-physical networked systems. She received NSF career award (2016) and AFSOR Young Investigator Award (2017). She entered the Best Student Paper Award finalist in the 2011 IEEE Conference on Decision and Control.