University of Victoria

Published purpose

We envision developing methods and automated tools for characterizing RF environments using machine learning techniques. This includes compact representations of signals, identifying groups (clusters) of signals, blind class discovery, classification, and assessing novelty of signals. We envision developing a common open format for the distribution of these datasets, with a focus on compatibility with pertinent standards such as IEEE 1900. The sparse representation will allow for inter-site collaboration on RFI detection and mitigation techniques which will establish us as the global leaders in combating detrimental effects of new interference sources (satellite mega constellations, 5G, etc.). The signal classification and novelty detection will enable rapid response to new interfering events ensuring DRAO remains a unique location able to carry out transformative radio science in Canada.