题目: Structured Interference --A Master Key for Network Capacity
时间: 2016年3月31日（星期四）下午3：00
地点：js33333金沙线路检测北楼四楼报告厅
主讲人: Hua Sun (孙砉), EECS, University of California, Irvine, USA
Abstract: Interference is a necessary evil for the communication networks that form the backbone
of modern society. Whether these networks are used for information storage, repair, retrieval or
transfer; whether they are optimized for secrecy, privacy, robustness, and/or throughput; and
whether they operate over idealized logical data links or the physical wireless medium, it is
nearly universally true that optimal solutions require intelligent mixing of information streams,
and can significantly outperform traditional approaches that try to keep information streams
separate. The intelligent mixing of information introduces structured interference. Indeed,
structured interference is the master key that unlocks many of the mysteries around the
questions of network capacity. This is the broad theme of my talk.
Within this broad theme, I will elaborate on the key structural elements of interference
management in three settings. First, I will focus on wireless interference networks,
both single-hop and multi-hop multiflow settings. For single hop I will highlight classes
where new insights into structured interference (onesided decomposability) are used to
settle an open problem. For 2 hops I will identify an intriguing rank matching phenomenon,
analogous to impedance matching in circuit theory. Second, I will shift focus to index coding,
which is a canonical problem that sits at the intersection of many open problems in network
information theory. I will identify three new classes of index coding -- those which are solved by translating ideas from wireless interference networks (interference
alignment), those where the structural requirements are trivial (orthogonal access is optimal),
and those where the structural requirements transcend our current understanding (non-Shannon information inequalities are necessary). Third, I will focus on the private information
retrieval problem traditionally studied in computer science and identify its capacity.
Short Bio: Hua Sun received the B.E. degree in Communications Engineering from Beijing
University of Posts and Telecommunications, Beijing, China, in 2011, the M.S. degree in
Electrical and Computer Engineering from University of California, Irvine, in 2013. He is
currently pursuing the Ph.D. degree at the University of California, Irvine. His research
interests include information theory and its applications to communications, networking,
and security.