Doctoral Thesis: Wireless Communication and Localization Systems under Spatial and Temporal Channel Variations


Event Speaker: 

Peter Iannucci

Event Location: 


Event Date/Time: 

Thursday, January 10, 2019 - 10:00am


The combination of mobility and multi-path generates key challenges for wireless system design in communication and in sensing.  My thesis work contributes new link-layer protocols, network abstractions, fingerprinting algorithms, and evaluation tools for wireless systems.  Each explores a new problem or a new opportunity created by existing mitigations for motion and multi-path.  In this talk, I will discuss three aspects of this work:

Rateless codes mitigate the risk of sub-optimal coding behavior in the presence of multi-path fading.  In half-duplex wireless systems, however, the transmitter has to stop talking in order to find out whether the receiver has succeeded in decoding the message.  Pausing for feedback is costly, and we would like to maximize throughput.  The Rateless Link-layer Protocol is an inexpensive, essentially-optimal algorithm for determining when the transmitter should pause for feedback from the receiver.

Orthogonal frequency-division multiplexing enables high-bandwidth communications in the presence of multi-path, but it requires a rather extreme level of signal fidelity.  Existing techniques for reducing the peakiness, and hence sensitivity to non-linearity, of an OFDM signal exhibit high hardware complexity.  I present a new application of the alternating-direction method of multipliers that requires significantly fewer resources than previous schemes, providing new options for inexpensive, backwards-compatible peak-to-average ratio minimization.

Society faces stresses from the two-edged sword of anonymous behavior on the Internet.  As public policy evolves, technical standards will surely face pressure to follow.  In anticipation of an “arms race” between wireless anonymizers and de-anonymizers, we have explored a number of measures, countermeasures, and counter-countermeasures for tracking moving sources of wireless signals using physical-layer information.  Our effort to characterize the strengths and weaknesses of the most powerful de-anonymizer conceivable (under certain assumptions on signal structure) led us to pose a family of unsolved mathematical problems, which we tackle using tools from quantum physics.
Thesis Supervisor: Prof. Hari Balakrishnan