Continuous GPS Authentication with Chimera using Stochastic Reachability Analysis

Abstract

Due to the fact that civilian GPS signals are currently unencrypted and have publicly available signal structures, all civilian users are vulnerable to spoofing attacks. Seeking to address this vulnerability, the U.S. Air Force Research Lab has proposed the Chips-Message Robust Authentication (Chimera) signal enhancement for the future GPS L1C signal, which allows users to authenticate the received GPS signal every 6 seconds using Chimera’s fast watermark channel. Correspondingly, if a user sought to only utilize the Chimera-authenticated GPS measurements, this would result in a 6-second latency. However, for many moving receiver applications (e.g., autonomous vehicles, UAVs), the receiver often relies on a much faster rate of GPS measurements, at 5-20 Hz. In this work, we propose a method to provide continuous GPS signal verification between Chimera authentication times by using stochastic reachability analysis to validate the received GPS measurement against any self-contained sensor measurement for state propagation (e.g., IMU, wheel encoders, LiDAR, direct control inputs, etc.). Furthermore, we derive our stochastic reachability-based spoofing detector to satisfy a user-defined false alarm requirement in the presence of bounded biases in the error distributions of the self-contained sensor and GPS receiver, during authentic conditions. We demonstrate via Monte Carlo simulations that our spoofing detector probabilistically satisfies a user-defined false alarm requirement throughout the trajectory during nominal conditions, while demonstrating its ability to successfully detect spoofing during a simulated attack. We further demonstrate the ability of our stochastic reachability state estimation filter to successfully bound the true receiver state, during both authentic and spoofed conditions.