The objective of this project is to undertake a comprehensive study of synergistic gains for multi-state communication networks and develop strategies that can exploit them. The project is organized into three complementary thrusts, focusing on 1) dynamic channels, 2) dynamic channel knowledge, and 3) mixed channel knowledge. The first thrust seeks ways to exploit channel variations starting with simple and optimal joint coding schemes over infinite horizons and leading, from a practical perspective, to efficient coding schemes over finite horizons, and, from a theoretical perspective, to alternative canonical models for information-theoretic insights into networks where the static setting has been intractable. The second thrust seeks ways to exploit alternating patterns of channel knowledge, either arising naturally out of channel dynamics or deliberately enforced through alternating feedback from different groups of users, to identify the most efficient forms of channel knowledge spread over time. The third thrust explores ways to exploit the simultaneous availability of distinct forms of channel knowledge, that have previously been studied only in isolation, and are therefore associated with coding schemes that are mutually incompatible.
With existing wireless networks struggling to keep up with the exploding data rate demands of an increasingly mobile and information-dependent society, it is imperative to uncover new opportunities for expanding the throughputs of next-generation systems. Any such effort must grapple with the enormous complexity of wireless networks, such as time-varying fading, dynamic network states, and the availability of many forms of network state information across users. The classical approach is to study the essential elements of these networks in isolation: decomposing a fundamentally multi-state network into its constituent states and studying each state individually. The motivation for this reduction typically comes from the conventional wisdom that 1) studying each element in isolation is easier than studying them together, and 2) understanding each element individually will lead to an understanding of their collective behavior. Surprisingly, recent work has shown that this wisdom can be misleading on both counts: multi-state networks are often not only more tractable, practical, and insightful, but also demonstrate significant synergistic gains that are not accessible through isolated studies of the constituent states. By shifting the focus from static to dynamic models, the proposed research will advance our understanding of information-theoretic limits of wireless interference networks where the intractability of the static setting has long stunted progress. Because joint coding across multiple states is often simpler, more insightful, and practical, it will bring the information-theoretic results closer to practice. A joint consideration of multiple states will provide the foundation for proper accounting of the overheads and benefits of channel knowledge. These research efforts are complemented by outreach efforts that include a mixture of traditional activities, such as tutorials at international conferences and summer schools, and novel ventures, such as the publication of an e-book on interference management.
