Potential Problem Analysis and RF Based Improvement of RPM on Android Devices

Global Navigation Satellite System (GNSS) spoofing attack, a serious threat to GNSS services across various sectors, occurs intermittently. Receiver Power Monitoring (RPM) is an effective algorithm to detect such attack by taking into account both AGC and C/N0 fluctuations. However, RPM has two notable limitations: it is primarily effective only during the initial stage of overpowered spoofing attack, and establishing an effective detection threshold proves challenging. In this study, we propose improvements to RPM by introducing 4 types of data features: carrier-to-noise ratio (C/N0) correlation coefficients, C/N0 fluctuations and their first-order differences, the number of visible satellites and satellites used for positioning within each constellation/frequency band, and the percentage of nulls in the sliding window of AGC and C/N0. The Random Forest (RF) algorithm was employed as the foundation for a data-driven approach to train an anti-spoofing detection model. To evaluate the performance of the improved algorithm, we collected AGC and C/N0 data from both authentic and spoofing signals under varying motion states at 4 representative sites. The extracted data features were used to build a detection model utilizing the RF algorithm. The model evaluation results demonstrate that the RF-RPM algorithm effectively addresses more complex spoofing scenarios, with an out-of-bag (OOB) error rate of less than 1.64% when the decision tree size exceeds 15. This study lays the groundwork for future engineering implementations and scientific research in GNSS anti-spoofing field.