My focus is enabling the "Internet of Things"; research to enable smaller,
smarter sensor nodes for a connected world. As these devices proliferate,
networks that "just work" are critical. Administering and managing thousands
of microscopic nodes is an untenable problem.

My research with the Michigan Micro Mote (M3) project fulfills all these goals.
M3 is an ambitious project to develop a cubic-millimeter sensor node. This
energy-harvesting node will support complex sensing primitives, such as motion
detection and image capture. My focus and interest are the software and system
level research questions: How do nodes with such extremely constrained power
budgets form a dynamic network? What density of nodes is required for a
deployment, what sensing coverage can be obtained? In anticipation of the
hardware, I have written a complete software emulator of the nodes (M-ulator)
that can dynamically shift to in-circuit emulation, replacing pieces of the
emulated system with real hardware as they become available, enabling both
component and system testing. I am confident that there are myriad
applications for a general purpose millimeter scale computing and sensing
platform at the DoD, as well as the M-ulator technology for enabling
concurrent hardware/software development.

The current corner of the M3 project---and the greater sensor network
community---that I wish to tackle is communication among ultra low power nodes.
Recent work by Yerva et al (Gecko) is the closest effort towards communicating
on an M3-scale power budget, and it failed. The fundamental problem is the
conflation of _communication_ and _synchronization_. Efficient wireless
communication protocols rely on nodes being synchronized, yet must also rely on
inefficient fallbacks using the same communication channel to bootstrap
synchronization and power-hungry real time clocks to maintain it. I propose
decoupling _synchronization_ from _communication_, using an alternative
low-power low-bandwidth communication medium to provide synchronization for
highly efficient RF communication protocols. Leveraging a 91 bps, 695 pW optical
receiver (FLOW, Kim et al) developed for M3, visual light becomes the
synchronization medium. I propose Smart Bulbs, LED bulbs that beacon
imperceptible synchronization messages to nodes. From previous work of
myself and of Chauvenet et al, the Smart Bulbs can form a mesh network using
powerline communication as a backhaul. Using Lu et al's Smart Blueprints, the
Smart Bulbs can automatically self-localize within a building.  Now any
FLOW-enabled node---or any device with a camera---can freely synchronize and
localize within a building.

My life goal is to become a professor. Beyond research, I have taught both
operating systems and embedded systems. I developed a novel class-wide project
built on M-ulator, requiring the entire class to work together as one cohesive
team. I love working at the forefront of technology and wish to do so for the
rest of my life.