Extraction of Static Features of RFI for Cancellation From Multiple Sources

Radio Frequency Interference (RFI) from ubiquitous expansion of communication networks poses a significant threat to the scientific and societal goals of passive radio users, like radio astronomy. Such RFI can be mitigated with blind cancellation approaches with limited accuracy or using collaborative methods with high communication overhead. In this work, we propose a collaborative data-driven RFI mitigation method in RAS that exploits unique properties of the most abundant interferer - modulated signals, by extracting features that depend on two key factors: static, based on waveform structure and dynamic, that captures temporal variation. This allows us to make the static features of the RFI available to the RAS in the deployment stage whereas the dynamic component is shared periodically. This allows us to significantly reduce the communication overhead and computation complexity of the RFI cancellation protocol. We also propose an RFI mapping approach that translates the shared RFI information into a reconstructed version of the RFI incident on a radio telescope. This mapped RFI is used to perform a time-domain cancellation and desired signal recovery. Extensive simulations of this approach are performed using real-world astronomical signals captured using the Deep Synoptic Array (DSA-110) at the Owens Valley Radio Observatory (OVRO) and simulated RFI from multiple base stations under different channel conditions, and are compared to state of the art collaborative RFI cancellation tools. Results show up to 70.42% reduction in communication overhead and astronomical signal recovery accuracy of 97.52%.