Graduate School and Research Center in Digital Sciences

Seminar: “Turning big data to small data: distributed scalable network telemetry with sketch and matrix methods”

Prof. Yongquan Fu - National University of Defense Technology

Date: November 29, 2019

Abstract: Internet provides reliable connections between billions of heterogeneous devices over the globe, which enable novel social-economical service applications. Network management is increasingly important to ensure the performance, utilization, availability, security. The first and foremost step is to gather network states via network telemetry. However, with the sheer volumes of network traffic and large-scale networked devices, real-time network telemetry is one of the grand challenges in the network field. Traditional methods either require device upgrades or heavy measurement cost, which poses challenges for commodity devices and edge networks. We present a suite of distributed network telemetry methods. First, we present two sketch methods to measure the one-way network delay and network flow counters, based on erasure codes and machine learning techniques. Next, we present two matrix-factorization methods to estimate network states, which can achieve self-stabilization and residual minimization under system churns. Moreover, these studies accurately infer global network states with linear measurements, which provides novel design directions for network management. Biography: Yongquan Fu received the B.E. degree from Shandong University, China in 2005, and the M.S. and Ph.D. degrees in computer science and technology from National University of Defense Technology, Changsha, China in 2007 and 2012, respectively. From 2010 and 2011, he was a joint PhD student with EURECOM, Sophia Antipolis, France. Since December 2012, he has been with the Science and Technology Laboratory of Parallel and Distributed Processing, College of Computer, National University of Defense Technology, where he is currently an associate professor. His research interests mostly include network machine learning, distributed systems, and data driven performance analysis and optimization. He had published over 30 peer-review papers at international conferences and journals, which include SIGCOMM, SIGMETRICS, CCGRID, WWW, CIKM, ToN, ACM TOMPECS, Computer Networks, Future Generation Computer Systems, Science China Information Sciences. Personal homepage: http://www.escience.cn/people/fuyongquan

“Turning big data to small data: distributed scalable network telemetry with sketch and matrix methods”

Search