Faculty Recruiting Support CICS

Software-defined Infrastructure for IoT-based Energy Systems

18 Jun
Tuesday, 06/18/2019 10:00am to 12:00pm
PhD Thesis Defense
Speaker: Stephen Lee

Smart Internet of Things (IoT) devices, sensor data analytics, and intelligent control have begun to permeate through every aspect of our society. These advances are poised to revolutionize domains such as smart energy, health, and transportation.

In this talk, I will argue the need for software-defined infrastructure in the context of managing IoT energy systems and discuss the four different components of my thesis. First, I will briefly discuss two approaches - (i) design of smart renewables that can modulate their power output in a grid-friendly manner, and (ii) virtualization of  IoT-based energy systems to share a larger system across a set of users.

Next, I will present a data-driven analytics approach that can be used to plan and deploy these energy systems at city-scale. Our deep-learning based approach identifies ideal rooftop locations for installing solar panels using satellite imagery. We validate our results with solar experts and show that our approach is more scalable than existing LIDAR-based approaches. Finally, I will discuss a Hyperledger Fabric-based blockchain system for peer-to-peer energy trading platform to share renewable energy. I will specifically address the performance bottlenecks that arise in these systems and reduce the computation and I/O overheads. Our approach improves the transaction throughput of the resulting system by over 4x, without any interface changes to Hyperledger Fabric.

Advisor: Prashant Shenoy