CAREER: Harnessing Interference Structure in Networks

Sponsor: National Science Foundation (NSF)

Award Number: 1253918

Co-I/Co-PI: Bobak Nazer

Wireless networks are the fabric of the mobile Internet. High-speed, ubiquitous wireless access is increasingly an enabling technology for important applications ranging from communication to commerce, medicine, and education. It is thus critical to create a pathway for sustainable wireless network growth in terms of the number of users and their data rates. A major challenge facing this effort is interference: since the wireless channel is a shared medium, simultaneous transmissions interfere with one another. This phenomenon is conventionally viewed as harmful by theorists and practitioners alike, and modern wireless protocols are designed to avoid interference. Within this paradigm, the wireless channel is a fixed resource with a finite capacity, meaning that as more users join a network, the maximum data rate per user plummets. However, interfering signals are not just additional noise; rather, they represent data from other users and possess considerable structure which can be exploited.

This research pursues a novel approach to multi-user communication based on harnessing the algebraic structure of interference. The key insight is that it is possible for a receiver to first decode linear combinations of all transmitted codewords and only afterwards solve for its desired codeword in the digital domain. In many important network scenarios, this powerful technique yields significantly higher data rates compared to conventional approaches. This project leverages this technique to uncover the fundamental capacity limits of canonical wireless networks subject to interference as well as establish architectural principles for coding strategies that can efficiently approach these limits. It draws upon modern tools from combinatorial optimization to solve the resource allocation problems that emerge when this technique is applied to complex wireless networks, such as distributed multiple antenna systems. More broadly, this research lays the foundation for an algebraic network information theory to tackle challenging problems in decentralized information processing, compression, and decoding. This project incorporates several outreach efforts including interactive presentations on cellular communication for high school students, tutorials, and workshop organization.

For more information, click here.