A Unified Framework for Delay-Sensitive Communications

Fangwen Fu
Advisor: Mihaela van der Schaar

Wednesday, May 19, 2010 at 3:00pm
Engr. IV Maxwell Room 57-124

Abstract:
Delay-sensitive communications (e.g. multimedia transmission) are booming over a variety of wireless networks. Current solutions often lead to an unsatisfactory experience for delay-sensitive applications since they ignore the stringent delay requirements and the heterogeneous features (e.g. importance, delay deadlines and dependencies) of the delay-sensitive data. This problem is becoming increasingly more serious when multiple delay-sensitive applications coexist in wireless networks and share the scarce network resources. In this dissertation, we develop a unified foresighted optimization framework which explicitly considers both the heterogeneity of the delay-sensitive data and the dynamics of the wireless networks in order to optimize the long-term utilities of the delay-sensitive applications.

In the proposed unified framework, we establish three separation principles which are theoretically important for designing delay-sensitive communication systems. First, by introducing the post-decision states, we separate the foresighted decisions from the underlying network dynamics, which enables us to explore the structures of the optimal solutions and design low-complexity algorithms. Second, in order to explicitly consider the heterogeneity of the multimedia traffic, we prioritize the delay-sensitive data (expressed as direct acyclic graphs) and separate the multi-data unit foresighted decision into multiple single-data unit foresighted decisions, which can subsequently be performed from the high priority to the low priority. Third, when multiple delay-sensitive applications coexist in the wireless network, by introducing a resource price and relaxing the network resource constraints imposed in the future transmission, we separate the multi-user foresighted decision into multiple single-user foresighted decision, thereby significantly reducing the computation and communication complexity.

Implementing the above framework in practice requires statistical knowledge of the network dynamics, which is often unavailable before transmission time. To overcome this obstacle, we propose novel structure-aware online learning algorithms derived from the above three separation principles. The proposed online learning algorithms have low complexity and fast convergence, and achieve ?-optimal solutions, which can significantly improve the delay-sensitive communication performance in the unknown environments.

Biography:
Fangwen Fu received the bachelor's and master's degrees from Tsinghua University, Beijing, China, in 2002 and 2005, respectively. He is currently pursuing the Ph.D. degree in the Department of Electrical Engineering, University of California, Los Angeles. During the summer of 2006, he was an Intern with the IBM T. J. Watson Research Center, Yorktown Heights, NY. During the summer of 2009, he was an intern with DOCOMO USA Labs, Palo Alto, CA. He was selected by IBM Research as one of the 12 top Ph.D. students to participate in the 2008 Watson Emerging Leaders in Multimedia Workshop in 2008. He received Dimitris Chorafas Foundation Award in 2009. His research interests include wireless multimedia streaming, resource management for networks and systems, stochastic optimization, applied game theory, video processing, and analysis.