Talk: Predictive Wireless Systems across Earth and Space

LIVE STREAM: https://uwmadison.zoom.us/j/95610095898?pwd=mtG5KgcFprodWOilazT6cEIpmB8BSW.1

Abstract: Wireless networking systems and protocols traditionally operate in a reactive manner – reacting to changes in network conditions, user mobility, and environmental variations. In my talk, I will argue that it is time to rethink this reactive design paradigm and move towards predictive wireless system design. My argument is motivated by two recent developments. First, large-scale satellite deployments in low Earth orbits promise global Internet connectivity and frequent Earth imagery. However, these satellites move at 15000 miles per hour with respect to users on Earth, have few minutes of connectivity to ground stations, and have rapidly fluctuating link capacity. These characteristics represent extreme environments for networking. Yet, the motion of these satellites is largely predictable due to their orbital dynamics. I will demonstrate that exploiting such predictability leads to new techniques for ground station scheduling, computing in space, and direct-to-satellite IoT connectivity. Second, recent advances in Machine Learning (ML) enable enhanced predictability that improves the performance of traditional wireless primitives like channel estimation and localization. In practice, real-world adoption of ML-based wireless systems faces challenges due to the need for frequent data labeling effort and the lack of robustness to adversarial and non-adversarial noise. I will discuss our group’s work on quantifying the vulnerabilities in ML-based wireless systems and on the path towards mitigating these vulnerabilities. Through the talk, I will highlight how an interplay of Physics-based models in wireless systems and ML-based techniques leads to new predictive wireless systems for communication and sensing.

Bio: Deepak Vasisht is an Assistant Professor in Computer Science at the University of Illinois at Urbana Champaign. His research is focused on mobile computing and internet-of-things (IoT) systems, including in-body, in-farm, and in-space systems. Deepak’s research has been published in multiple top venues, including SIGCOMM, MobiCom, NSDI, and Ubicomp. His research on IoT systems for agriculture has been deployed across farms in United States, India, and Africa. His work has been featured in the Economist, IEEE Spectrum, BBC, MIT News, Daily Mail, and CBC among others. Deepak received his undergraduate degree from IIT Delhi where he was awarded the President of India gold medal. In 2021, Deepak was named as a Graduate of the Last Decade by IIT Delhi. He has a Ph.D. from MIT, where his research won the ACM SIGCOMMM Doctoral Dissertation Award, two best paper awards, and the Microsoft Research PhD Fellowship. Deepak is a recipient of an NSF CAREER award and the VMWare Systems research award in 2023.