An important source of complexity in emerging wireless networks is the increasingly diverse set of wireless devices and their applications. Interference management is a key challenge in the design and operation of such networks, and forms the overall theme that connects the speaker’s work in diverse settings. This talk provides a brief overview of some aspects of this work and is organized as follows. Part one situates his work in the overall context of interference management and pro- vides a bird’s eye view of his work both in academia and in industry. Parts two and three will discuss selected topics from his work in industry and as an academic researcher, respectively. The talk concludes with some thoughts on the challenges posed by the increased device heterogeneity expected in future wireless networks. The talk will touch upon the speaker’s contributions to three challenging problems in wireless LAN (WLAN, commonly called “Wi-Fi”) system design that form some of the core technology in flagship wireless LAN SoC products. The first two topics pertain to some of his work at Broadcom on the general problem of wireless coexistence, where devices with very different communication requirements and constraints share a common spectrum. The roles of interference avoidance, mit- igation and adaptation are discussed in the context of specific examples. It is pointed out how a holistic perspective on the system-level impact of interference is necessary for significantly improved coexistence. The third topic is selected to present some of his work at Redpine Signals, Inc., that concerns the design of low-complexity soft-output MIMO demodulators. It is shown how reasonably accurate Likelihood-Ratios (LLRs) can be obtained without the need for matrix inversion when the number of spatial streams is small, a scenario of practical interest. As part of his doctoral research, the speaker has worked on multiple theoretical and exper- imental topics. One representative topic from each methodology will be presented here. The experimental work will discuss the development of the first known prototype of a superposition coded wireless system that was implemented on a software radio platform. In addition to sys- tematically developing experimentally robust analogs and methodologies to theoretical constructs (such as the information-theoretic rate region), this line of work also investigates some approaches that can leverage superposition coding in practice. The theoretical topic will discuss his work on deriving asymptotic scaling laws for the convergence behavior for a class of distributed signal pro- cessing algorithms called average consensus algorithms when message exchange between network nodes occurs over an interference-limited wireless network. By combining the mathematical tools from Markov chain theory, geometric random graph theory and information theory, this line of work was the first to establish convergence results on this algorithm in the presence of interference. In particular, it challenges the long-held view that increased communication range always improves the rate of convergence to consensus. The talk ends with some thoughts on the emergence of device heterogeneity, and some challenges this poses in the context of the emerging Internet of Things (IoT).
Sundaram Vanka received the B. Tech. and M. Tech. degrees from the Indian Institute of Technol- ogy, Madras in 2003, and the Ph. D. degree from the University of Notre Dame, Indiana, United States of America in 2012, all in Electrical Engineering. Since 2012 he has been with the wireless LAN (WLAN) systems architecture group at Broadcom Ltd. (previously Broadcom Corporation), a market-leading designer, developer and supplier of WLAN chips on diverse platforms, where he is currently Principal R&D engineer. He has been a lead architect into systems R&D on new system architectures and protocols for interference management in WLAN systems. His work spans the design, implementation, laboratory characterization and productization of the same on Broadcom’s flagship WLAN/Bluetooth combo System-on-Chip (SoC) products that are shipped to leading con- sumer electronics companies.