Kalman filter design for spoofing attack detection and anti-spoofing

With the widespread use of global navigation satellite systems (GNSS) measurements in positioning solutions and especially forming the basis of autonomous vehicle navigation, the importance of measurement accuracy from these systems has increased. On the other hand, applications have also started to become widespread in recent years, for spoofing attacks, ensuring that the GNSS receivers obtain incorrect positioning solutions. Therefore, adding detection and countermeasure algorithms, within GNSS receivers, against possible spoofing attacks is critically important. For spoofing detection and countermeasure, there are many different methods developed within GNSS receivers, and these methods are recommended for use in different parts of the receiver structure. Within the scope of this study, in order to detect and prevent the spoofing attack a Kalman filter design is presented for use in a GNSS receiver mounted on an autonomous moving vehicle, which provides a positioning solution by using GPS, Galileo, GLONASS, and BeiDou systems, that are actively providing global positioning services. This proposed filter is designed to be used in the navigation block, just before presenting the final position solution, after generating a positioning solution with the signals received by the GNSS receiver from different positioning systems. The algorithm designed in the first part of the studies detects spoofing, based on the method of comparing the residual calculations of different GNSS measurements, and then continues to produce a positioning solution using non-spoofed system measurements. In addition to this algorithm, in the second part of the studies, generating positioning solutions with an accelerometer to be integrated into the GNSS receiver is modeled. The comparison of the residuals calculated by using these solutions with the GNSS residuals is included in the algorithm. By integrating the accelerometer into the system, the number of possible spoofing scenarios is increased. The performance of the designed filter has been tested with Monte-Carlo simulations. By creating different spoofing trajectories, the performance of the filter is examined in terms of percentage spoofing detection performance and detection time under all possible scenarios in which spoofing attacks can be applied in these trajectories. For the created spoofing trajectories, the relationship between the dominant and recessive GNSS position accuracies and detection times is explained in the applied spoofing scenarios. Finally, by testing the threshold value used by the filter to detect spoofing for different levels, receiver operating characteristics curves were drawn, and the performances of the filter in different spoofing trajectories were given by calculating the areas under these curves.