Writing · 2024
Analyzing the Resilience of UAV Communication Systems
How UAV control links degrade under interference and EMI, and a machine-learning approach to testing noise tolerance systematically instead of anecdotally.
Sole-authored, Cal Poly Pomona · 2024
Signal integrity is a flight-safety property in a UAV, not a comms nicety. Competing signals degrade link quality into control latency or outright loss of signal, and the onboard electronics are themselves a source of the problem — power supply fluctuation and electromagnetic interference corrupt sensor accuracy and actuator response from inside the airframe.
The paper works through UAV flight controller architecture, the mechanisms by which signal integrity is lost, and the testing methodologies available for deliberately disrupting a link under controlled conditions.
Its argument is that resilience testing is usually anecdotal — fly it and see — and that machine learning offers a way to characterise a vehicle's noise tolerance systematically, by exploring the disruption space rather than sampling a few points in it.