Physical Layer Authentication for Inter-Satellite Communication

Satellite communications will play a vital role in future networks as they enable extended coverage and bandwidth, which is important, particularly in remote areas. In fact, megaconstellations of LEO satellites provide solutions to connect the world with minimum delay. Satellites in constellations communicate with each other using interlinks, creating space networks to facilitate traffic routing between satellites while increasing throughput and bandwidth. Consequently, ensuring secure and reliable interconnection between these satellites is necessary to achieve the required connection and coverage. However, intersatellite links are vulnerable to passive and active attacks since their signals are transmitted in open space. Specifically, LEO satellites are vulnerable to spoofing attacks due to their relative proximity to earth and the increasing number of satellites in this region. LEO satellites are characterized by high doppler shift frequency due to their high-speed mobility. In this thesis, we propose a physical layer authentication algorithm that uses the doppler shift frequency feature to authenticate LEO satellites and detect spoofing attacks. We develop a deep learning model that predicts mobility information (position and speed) for LEO satellites. According to the predictions, the algorithm calculates the estimated doppler frequency shift and uses it for authentication. We create a dataset of 8 LEO satellites to train our model. We simulate and evaluate our algorithm with various scenarios to prove its ability to detect spoofing attacks with low false alarm rates. The simulation results show that our algorithm can accurately predict satellites’ mobility and successfully authenticate them accordingly (while detecting spoofing attackers).