Secured UAV Navigation: A Novel Intrusion Detection System Based on PWM Signal Analysis

Unmanned Aerial Vehicles (UAVs) have seen exponential growth in applications, from surveillance to logistics. Ensuring their security, especially in a Global Position System (GPS)-compromised environment, is critical. This paper introduces an Intrusion Detection System (IDS) as a countermeasure against sensor’s attacks that leverages Pulse Width Modulation (PWM) signals, generated by the UAV Board. The IDS, positioned at the Electronic Speed Controllers (ESCs), can potentially detect anomalies even when conventional ground station detections fail. Multiple unsupervised machine learning algorithms were evaluated for this purpose. The Local Factor Outlier (LOF) emerged as the most accurate with 75.87%, closely followed by One Class SVM (OCSVM) at 74.17%, and Isolation Forest (IF) at 71.59%. These findings suggest that monitoring PWM signals could provide a novel layer of security for UAVs, making them resilient against certain forms of cyber-attacks.