From Sensors to Video: Information Theoretic Foundations of Source Communication over Wireless Channels
Communication over wireless channels is challenging because of random time variations and multiuser interference. Many wireless applications include source signals whose properties can be exploited for improved performance: Sensor data are typically correlated across time and space; video signals are scalable and can be represented at various qualities. This lecture provides an information theoretic foundation of communicating such source signals over wireless channels.
We start by reviewing Shannon's source-channel coding theorem in the context of wireless channels by showing that in simple multiuser networks, and in channels with random time variations, it is no longer optimal to separate source and channel coding. We then show some simple building blocks of optimal joint source-channel codes that maintain a certain level of modularity. We discuss how some of these principles can be applied to wireless video delivery and how availability of memory at the end users can augment video quality by joint design of caching and delivery strategies.
Biography: Elza Erkip is a Professor of Electrical and Computer Engineering at New York University Tandon School of Engineering, Brooklyn, NY, USA. She is an IEEE Fellow, a member of the Science Academy Society of Turkey and is among the Thomson Reuters Highly Cited Researchers. She received the NSF CAREER Award in 2001, the IEEE Communications Society Stephen O. Rice Paper Prize in 2004, the IEEE ICC Communication Theory Symposium Best Paper Award in 2007, and the IEEE Communications Society Award for Advances in Communication in 2013. She has served on various editorial boards and conference organizing committees, and will be the technical program chair of WCNC in 2017.