| Lecturer Profile |
Dr. Zuqing Zhu received the PhD degrees from the Department of Electrical and Computer Engineering, University of California, Davis, in 2007. In Jul. 2007, he joined the Service Provider Technology Group of Cisco Systems, San Jose, as a senior software engineer. In Jan. 2011, Dr. Zhu joined the University of Science and Technology of China (USTC), where he is now a full professor. He has published more than 150 papers on the peer-reviewed journals and conferences. Dr. Zhu is an editorial board member of IEEE Communications Magazine, and others. He has received the Best Paper Awards from IEEE ICC 2013, IEEE GLOBECOM 2013, IEEE ICNC 2014, and IEEE ICC 2015.He is a Senior Member of IEEE and a Senior Member of OSA. |
| Lecture Abstract |
It is known that by incorporating network function virtualization (NFV) in inter-datacenter (inter-DC) networks, we can use the network resources more intelligently to deploy new services faster. This paper considers an inter-DC elastic optical network (IDC-EON) and studies how to orchestrate the multicast-oriented NFV trees (M-NFV-Ts) in it efficiently. We first consider an offline scenario in which all the M-NFV-Ts are known and need to be served in the network. A mixed integer linear programming (MILP) model is formulated to solve the problem exactly, and we also propose a heuristic based on path-intersection (PI) to reduce the time complexity. With extensive simulations, we show that the proposed heuristic can approximate the MILP's performance on low-cost M-NFV-T provisioning but only requires much shorter running time. Next, the online scenario where the M-NFV-Ts can come and leave on-the-fly is addressed, and we leverage PI to design two online algorithms for orchestrating dynamic M-NFV-Ts in IDC-EONs, i.e., with either the batch (B-PI) or sequential (S-PI) scheme. Simulation results indicate that compared with S-PI, B-PI can reduce blocking probability effectively. |
