Space Cognitive Communications: Characterizing Radiofrequency Interference to Improve Digital Space Domain Awareness

Radiofrequency interference (RFI) is a persis-tent challenge for space system communications. The Rus-sianlUkrainian conflict has heightened awareness of the signif-icant impact of RFI. The dramatic increase in space vehicle launches in recent years has exacerbated this issue given the challenges of enforcing spectrum regulation. Space domain awareness (SDA) involves providing space vehicles and their operators visibility to their surrounding environment so as to avoid collisions - both physical and digital. Given the potentially malicious nature of radiofrequency interference, it becomes mission-critical to precisely characterize RFI. This paper presents a novel application of a combined convolutional neural network and a k-nearest neighbor algorithm to enable space cognitive communications. Our models, designed for edge processors, can identify specific types of radiofrequency interfer-ence including crosstalk, jamming attacks, spoofing attacks and replay attacks with a high degree of accuracy. We developed and tested these models for Global Navigation Satellite System (GNSS) signals which are among the most common mission-critical communications for space vehicles.