PublisherThe University of Arizona.
RightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
AbstractThe goal: To expedite cleaning of maritime distress signals without manual filtering. Maritime distress transmissions received by the U.S. Coast Guard are subject to environmental noise and communication channel distortion, and must be filtered manually using current Rescue 21 software, which can be time consuming and imprecise. This project uses deep learning to improve signal filtration. The core of the software is a de-noising auto-encoder, a machine learning model that recognizes and removes distortion from input audio files. Noisy distress signals are input into the team’s software and automatically filtered by the auto-encoder, which generates a coherent, noiseless version of the input signal. Testing required the software to process a 5-minute audio segment in under a minute while using less than 2 GB of local memory. The software is compatible with all operating systems used by the Coast Guard, including the existing Rescue 21 software.
Degree ProgramHonors College
Electrical and Computer Engineering