Massachusetts Institute of Technology
Friday, March 16, 2018
10 AM - 11 AM
The Internet of Things will transform industries, but its adoption is challenged by resource constraints and fears about data privacy and system security. In this talk, I define the Internet of Things and explore key challenges before presenting an architecture for connectivity capable of minimizing constrained devices' power and bandwidth use while simultaneously improving security and gaining user trust. This approach employs "Data Proxy" models to digitally mimic human's contextual understanding in order to efficiently generate rich digital twins from sparse inputs. These Proxies further anticipate the consequences of incoming commands to check them for safety and feasibility, as well as supervise a system's state evolution to detect and respond to monitoring or modeling faults. This architecture, in conjunction with user- centric visualization tools, comprehensively addresses challenges in resource efficiency, security, and privacy, removing barriers to IoT's growth. Improved security and efficiency facilitate information collection at scale. I discuss how the resulting abundant data allow engineers to optimize vehicles during design, manufacturing, use, and maintenance. I focus in-depth on how pervasively-sensed smartphone audio can diagnose automotive engine misfires as part of an effort to build an intelligent "tricorder" for cars and other systems.
Josh Siegel is a research scientist at the Massachusetts Institute of Technology, the founder of two automotive startups, and the lead instructor/organizer for MIT's Internet of Things Bootcamp. He received S.B. (2011), S.M. (2013), and Ph.D. (2015) degrees in mechanical engineering from MIT for his work creating platforms and applications for connected vehicles, developing "cognitive" architectures for the Internet of Things, and applying smartphone sensor data to proactively detect automotive maintenance needs. Josh and his companies have been recognized with numerous accolades including the Lemelson-MIT Student Prize and the MassIT Government Innovation Prize. He has multiple pending and issued patents and has published in top scholarly venues including IEEE Transactions on Intelligent Transportation Systems, IEEE IoT Journal, and the Proceedings of the European Conference on Machine Learning. His work has been featured in popular media outlets such as WIRED, ArsTechnica, Technology Review and MIT News. Dr. Siegel's ongoing research focuses on developing architectures for secure and efficient connectivity, applications for pervasively sensed data, and intelligent diagnostics and prognostics for electrical and mechanical systems.
Dr. Xiaoming Liu