High-Resolution Correlator Based Detection of GPS Spoofing Attacks Using the LASSO

This work proposes a novel sparsity-based decomposition method for the correlator output signals in GPS receivers capable of detecting spoofing attacks. We model complex correlator outputs of the received signal to form a dictionary of triangle-shaped replicas and employ a sparsity technique that selects potential matching triangle replicas from said dictionary. We formulate an optimization problem at the receiver correlator domain by using the Least Absolute Shrinkage and Selection Operator (LASSO) to find sparse code-phase peaks where such triangle-shaped delays are located. The optimal solution of this optimization technique discriminates two different code-phase values as authentic and spoofed peaks in a sparse vector output. We use a threshold to mitigate false alarms. Additionally, we present an expansion of the model by enhancing the dictionary to a collection of shifted triangles with higher resolution. Our experiments are able to discriminate authentic and spoofer peaks from synthetic GPS-like simulations. We also test our method on a real dataset, namely the Texas Spoofing Test Battery (TEXBAT). Our method achieves less than 1% detection error rate (DER) in nominal signal-to-noise ratio (SNR) conditions.