A Study of GNSS Jamming Detection Using Deep Learning

The Global Navigation Satellite System is being widely used for positioning, navigation, and timing information. The satellite orbit in this system is around 20 km above the surface of the earth. The signal acquired by the receiver at the earth’s station is, therefore, very low. This makes it highly vulnerable to malicious attacks including jamming, which involves deliberate attacks on signals. It causes significant threats to reliability, security, and functionality of the system. Moreover, availability of low-cost jammers is causing serious threats to the communication links. In this paper, we report on the development of a long short-term memory based deep learning network for jamming detection. It successfully identifies the interference caused by the jamming, which may enable the mitigation of the interference. Our numerical results demonstrate that the proposed deep learning model achieves prediction accuracy of jamming in the order of 99.93% at low SNR. It successfully works well at low jamming power levels and hence it can be considered for real-time monitoring of the jamming signals.