Dissertation Defense

Improving RF Localization Through Measurement and Manipulation of the Channel Impulse Response

Benjamin Kempke

For over twenty years, global navigation satellite systems like GPS have
provided an invaluable navigation, tracking, and time synchronization service
that is used by people, wildlife, and machinery. Unfortunately, the coverage
and accuracy of GPS is diminished or lost when brought indoors since GPS
signals experience attenuation and distortion after passing through and
reflecting off of building materials. This disparity in coverage coupled with
growing demands for indoor positioning, navigation, and tracking has led to a
plethora of research in localization technologies. To date, however, no single
system has emerged as a clear solution to the indoor localization and
navigation problem because the myriad of potential applications have widely
varying performance requirements and design constraints that no system
satisfies. Fortunately, recently-introduced commercial ultra-wideband RF
hardware offers excellent ranging accuracy in difficult indoor settings, but
these systems lack the robustness and simplicity needed for many indoor
applications. We claim that an asymmetric design that separates transmit and
receive functions can enable many of the envisioned applications not currently
realizable with an integrated design. This separation of functionality allows
for a flexible architecture which is more robust to the in-band interference
and heavy multipath commonly found in indoor environments.

In this dissertation, we explore the size, weight, accuracy, and power
requirements imposed on tracked objects (tags) for three broadly representative
applications and propose the design of fixed-location infrastructure (anchors)
that accurately and robustly estimate a tag's location, while minimizing
deployment complexity and adhering to a unified system architecture. Enabled
applications range from 3D tracking of small, fast-moving micro-quadrotors to
2D personal navigation across indoor maps to tracking objects that remain
stationary for long periods of time with near-zero energy cost. Each
application requires careful measurement of the ultra-wideband channel impulse
response, and an augmented narrowband receiver is proposed to perform these
measurements. The key design principle is to offload implementation complexity
to static infrastructure where an increase in cost and complexity can be more
easily absorbed and amortized. Finally, with an eye towards the future, we
explore how the increasingly crowded RF spectrum impacts current ultra-wideband
system design, and propose an alternative architecture that enables improved
coexistence of narrowband and ultra-wideband transmissions.

Sponsored by

Prabal Dutta