GNSS Spoofing Discovery by Using a Type-Based Detector

With the increase in the number of devices relying on Global Navigation Satellite Sustems (GNSS) and the wide use of such technology for critical applications, such as in aviation and autonomous vehicles, security is a main concern. In particular GNSS spoofing attacks and their mitigation attracted much attention in the research community. Availability of efficient methods for combating spoofing are a significant factor in the use of GNSS technology. This paper considers GNSS spoofing detection without the need for despreading the received signal. As a specific application of our method we consider the most commonly used civilian GPS C/A L1 signal. The proposed technique consists of two stages: signal preprocessing stage and the detection stage. We show theoretically that signal samples at the output of the preprocessing stage follow a mixture Gaussian distribution, rather than a Gaussian distribution as infered from other publications based on measurement results. Using computer simulations we demonstrate the adequacy of our mixture Gaussian model. The consequence of our models is that detection methods for such application that are based on the Gaussian model may not be adequate. Then the second stage consists of a type-based technique for spoofing detection. Type detection is a trainable data oriented technique with foundations in Information Theory, and hence it can be considered as a learning algorithm. In our specific application such detector requires training only on a signal without spoofing. In order to apply the type-based detector in our system, a linear quantizer is used and the effect of the number of quantization levels is studied. Simulation results for Receiver Operating Characteritics (ROC) indicate that reliable performance can be achieved when using a type-based detector following the preprocessing stage for GNSS spoofing detection. Furthermore, the shortcomings of energy detection in such applications are also studied. Both theoretical and simulation results in our paper indicate the advantages of type-based detection over energy detection in GNSS spoofing discovery.