Detection of GPS Spoofing Using Post-Correlation Frequency Component Estimation

Abstract

This dissertation presents a new GPS spoofing detection method, Correlator Residual Frequency (CRF) detection, which is based on frequency-domain maximum likelihood estimation of differences between the receiver's correlator outputs. The use of correlator outputs for spoofing detection does not require additional hardware, such as multi-antennas or inertial measurement units, and can be implemented in receiver firmware for easier integration and reduced cost. Previously researched correlator-based methods use correlator magnitude or phase measurements. CRF spoofing detection uses a maximum likelihood estimator to identify misleading signals in the frequency domain.

While GPS spoofing and jamming takes many forms, a spoofer that can accurately recreate the RF signal environment of a target receiver poses a very difficult detection problem. An advanced spoofer can control time, frequency, and signal power to capture a targeted receiver with little or no interruption to the receiver’s operation. Detection of this type of advanced spoofing early in the processing stream can help prevent undesirable effects from propagating through the system. CRF detection is focused on using the correlator outputs to detect individual spoofed GPS signals to allow the receiver to remove them from computation or take other appropriate action.

An analytical model of the residual spoofing component in the correlator difference is derived. Statistical analysis and sensitivity analysis of the correlator difference are presented, and the detection algorithm is described in a step-by-step procedure. Results of Monte Carlo simulations of static and dynamic scenarios generated with commercial software showed a 74% detection rate of a spoofed signal, compared to 59% for the Early-Late Phase metric and 54% for the Signal Quality Monitoring Delta metric. Additional testing with Texas Spoofing Test Battery scenarios and live-sky signals helps validate the simulation results.