684911 event 1758039015 1758039090 <![CDATA[Ph.D. Dissertation Defense - Shyam Krishnan Venkateswaran]]> Title:  Energy-Efficient Wi-Fi For The Internet of Things: Tools, Analysis, and Algorithms

Committee:

Dr. Raghupathy Sivakumar, ECE, Chair, Advisor

Dr. Karthik Sundaresan, ECE, Co-Advisor

Dr. Douglas Blough, ECE

Dr. John Barry, ECE

Dr. Sriram Vishwanath, ECE

Dr. Ashutosh Dhenke, CS

]]> This research addresses the fundamental tension between Wi-Fi's evolution toward higher performance and the critical energy efficiency requirements of Internet of Things (IoT) devices. While Wi-Fi dominates IoT connectivity, existing power-saving mechanisms face severe limitations under real-world conditions. Network congestion degrades orderly sleep schedules, forcing devices into prolonged active states. Meanwhile, the persistent "cost of listening" - requiring periodic wake-ups to check for pending traffic - establishes an energy floor that prevents multi-year battery life. These challenges demand a systematic investigation that moves beyond incremental improvements toward comprehensive solutions addressing when, how, and which devices are active in IoT networks. This thesis presents a multi-layered approach combining foundational tools, rigorous analysis, and algorithmic and architectural contributions. We develop a comprehensive open-source simulation toolkit with hardware-validated energy models and implementations of critical IEEE 802.11 features. This enables precise diagnosis of energy bottlenecks in both legacy and modern Wi-Fi. Building on these insights, we propose two algorithmic enhancements. First, a cross-layer client-side optimization significantly extends battery life for legacy power-saving features without infrastructure changes. Second, Accordion dynamically schedules modern Wi-Fi devices, adapting to varying network loads to deliver superior throughput and latency compared to static approaches. Finally, we introduce AuraWake to address the fundamental "cost of listening" barrier. This architecture superimposes control signals directly onto Wi-Fi data packets, enabling ultra-low-power wake-up radios to decode messages without dedicated spectrum. This eliminates the need for sleeping devices to wake their main radio for traffic checks. Together, these contributions provide a comprehensive approach for realizing energy-efficient Wi-Fi for IoT, demonstrating that high-performance wireless and ultra-low-power operation can be reconciled.

]]> <![CDATA[]]> 434381 1788 100811 1808