This NSF project is part of a grouping of related research projects that we informally refer to as the "Control over Wireless" or CoW projects. The goal overall across these projects is to think comprehensively about how to support the high-performance Internet-of-Things ecosystem that is going to be critical for the United States Economy, competitiveness, and jobs in the future.

This project CNS-1321155 focuses on wireless architecture and protocols as well as developing the underlying theoretical foundations of ideas that can be leveraged into protocols. It is supported by the NeTS program out of NSF's CISE directorate and has synergies with another project ECCS-1343398 that is supported by NSF's cross-cutting EARS program and particularly from within the ENG directorate and takes a spectrum-sharing angle and looks more closely at implementation-driven issues.

Synopsis

At the heart of this particular project have been two core prongs. One prong looks at the theoretical foundations of ideas connecting control to communication so that we can understand what the fundamental tradeoff are. The other prong takes theoretical ideas and combines them into architectures and protocols for wireless communication to support high-performance control of the Internet-of-Things. At a high level, the distinction is that the leading edge of the theoretical prong is looking at how to tolerate imperfections in communication and how much imperfection can be tolerated. Meanwhile, the protocol design work is focussed on seeing how to make wireless protocols that deliver nearly perfect communication from the perspective of control. We believe that in the relatively near term, almost perfect communications makes it easier to combine with existing control protocols that rely on almost perfect wired networks and are more suitable for being drop-in replacements for wired networks. Studying what it takes to make nearly perfect protocols also tells us when this is too "expensive" in terms of resources that are needed. That is when being able to relax our demand for perfection is important and the theoretical prong helps us understand what can be relaxed and what is fundamental.

Publications

Ranade, G. and Sahai, A. "Noncoherence in estimation and control," Allerton Conference on Communication, Control, and Computing, Oct 2013.

Park, S.Y. and Sahai, A. "A geometric slicing lower bound for average cost dynamic programming," IEEE Conference on Decision and Control, December 2013.

Park, S.Y. and Sahai, A. "Intermittent Kalman filtering with adversarial erasures: Eigenvalue cycles again," IEEE Conference on Decision and Control, December 2013.

Weiner, M.; Jorgovanovic, M.; Sahai, A.; Nikolic, B., “Design of a low-latency, high-reliability wireless communication system for control applications,” Communications (ICC), 2014 IEEE International Conference on , vol., no., pp.3829,3835, 10-14 June 2014.

Ramnarayan, G.; Ranade, G.; Sahai, A. "Side information in control and estimation," IEEE Symposium on Information Theory, June 2014.

Swamy, V. N.; Suri S.; Rigge, P.; Weiner, M.; Ranade, G.; Sahai, A.; Nikolic, B. “Cooperative Communication for High-Reliability Low-Latency Wireless Control,” 2015 IEEE Conference on Communications, June 2015.

Ranade, G. and Sahai, A. "Control capacity," IEEE Symposium on Information Theory, June 2015.

Educational Activities, Outreach, etc.

The ideas in this project have inspired rethinking of the Berkeley EECS Lower-Division curriculum and project alumna Gireeja Ranade took a lead role in the design and delivery of the new freshman-level EE16AB course sequence. Internet-of-Things ideas as well as wireless communication play a very visible role in these courses and furthermore, ideas developed in part in this project help spawn the "information as dimensions" theme that also plays a prominant role in EE16AB.

This project has also involved undergraduates, most notably Sahaana Suri. Her work on this project led to her being recognized as a CRA finalist for the Best Undergraduate Researcher Award.