CSE Doctoral Dissertation Defense: Kenji Aono

Dec 19, 2018
1 p.m.
3 p.m.
Jolley Hall, Room 309

"Nanopower Analog Frontends for Cyber-Physical Systems"

Kenji Aono
Adviser: Shantanu Chakrabartty

In a world that is increasingly dominated by advances made in digital systems, this work will explore the exploiting of naturally occurring physical phenomenon to pave the way towards a self-powered sensor for Cyber-Physical Systems (CPS). In general, a sensor frontend can be broken up into a handful of basic stages: transduction, filtering, energy conversion, measurement, and interfacing. One analog artifact that was investigated for filtering was the physical phenomenon of hysteresis induced in current-mode biquads driven near or at their saturation limit. Known as jump resonance, this analog construct facilitates a higher quality factor to be brought about without resorting to the addition of multiple stages and poles in the filter.

Exploiting this allows a filter that mimics mammalian cochlea using \SI{}{\nano\W} of power, and the viability of using such a filter was demonstrated in the application of speaker recognition. Features were extracted using a silicon cochlea analog frontend, which outperformed features from traditional linear filters when classification was done with a Gini-SVM.

To realize the measurement stage of the frontend, a previously reported technology, the Piezoelectric-Floating-Gate (PFG) was employed. The PFG matches physics of Impact-Ionized Hot-Electron Injection (IIHEI) in silicon metal-oxide field effect transistors with a piezoelectric transducer to drive nonvolatile data-logging measurements. The PFG implementation is self-powered in the sense that the energy required for sensing comes from the signal being observed, which allows for continuous, zero-downtime measurements of signals that exceed the IIHEI threshold and can drive \SI{}{\nano\W} loads. Moreover, since it directly matches the transduction stage to measurement, it obviates the need for an explicit energy conversion stage in the frontend. Multiple interfacing technologies were evaluated, including: wired, self-powered radio-frequency (RF) backscatter, periodic \SI{915}{\mega\hertz} active RF, and a hybrid model that uses energy scavenging to determine if an interrogator is within range before transmitting. A multi-year deployment of this sensor frontend for structural health monitoring is currently active on the Mackinac Bridge in northern Michigan and demonstrates successful transition from laboratory to practice for a CPS.

Finally, a modification to the PFG topology to include filtering aspects borrowed from earlier study was proposed and fabricated on a standard \SI{0.5}{\micro\meter} CMOS process. Measurements show that the PFG sensor can be endowed with frequency discriminating capabilities to better focus on signals of interest. The modifications also give rise to a means for higher sensitivity (input stimuli below IIHEI threshold) data-logging that would vastly expand the potential application space.