loading page

Leveraging Atmospheric Effects with LSTM for Ensuring Cybersecurity of Satellite Links
  • Rajnish Kumar ,
  • Shlomi Arnon
Rajnish Kumar
Author Profile
Shlomi Arnon
Author Profile

Abstract

The progress in quantum computing would make the traditional encryption methods easier to hack and hence the next generation of communication networks should include various cross-layer solutions to ensure the cybersecurity of network. In this work, we propose a physical layer solution that rely on taking advantage from the inherent physical properties of the channel that marks its imprint on the propagation of the electromagnetic signal through the stochastic atmospheric channel. The proposed architecture would ensure the cybersecurity of a ground station against an MITM attack launched from an aerial platform (AP). As the signal traverses through the channel, various atmospheric effects including thermal noise and tropospheric scintillation fading causes rapid fluctuations in the received signal. The signal at the ground station from a satellite and an AP would traverse different spatial path over the dynamic atmosphere and hence will show different noise like rapid variations at the ground station. These noisy variations are extracted from the received signal using wavelet filtering technique and processed through a deep neural network based on long-short term memory network (LSTM) to identify whether the incoming signal belong to a legitimate satellite or a spoofing AP. We show that an accuracy of more than $98 \%$ is achieved in the considered scenario with the proposed network architecture as the altitude of AP changes for launching a machine-in-the-middle (MITM) attack in the line-of-sight direction.