Date of Award

December 2022

Document Type


Degree Name

Master of Science (MS)


Computer Science

First Advisor

Naima Kaabouch


Unmanned Aerial Vehicle systems (UAVs) are widely used in civil and military applications. These systems rely on trustworthy connections with various nodes in their network to conduct their safe operations and return-to-home. These entities consist of other aircrafts, ground control facilities, air traffic control facilities, and satellite navigation systems. Global positioning systems (GPS) play a significant role in UAV's communication with different nodes, navigation, and positioning tasks. However, due to the unencrypted nature of the GPS signals, these vehicles are prone to several cyberattacks, including GPS meaconing, GPS spoofing, and jamming. Therefore, this thesis aims at conducting a detailed comparison of two widely used machine learning techniques, namely weak and strong learners, to investigate their performance in detecting GPS spoofing attacks that target UAVs. Real data are used to generate training datasets and test the effectiveness of machine learning techniques. Various features are derived from this data. To evaluate the performance of the models, seven different evaluation metrics, including accuracy, probabilities of detection and misdetection, probability of false alarm, processing time, prediction time per sample, and memory size, are implemented. The results show that both types of machine learning algorithms provide high detection and low false alarm probabilities. In addition, despite being structurally weaker than strong learners, weak learner classifiers also, achieve a good detection rate. However, the strong learners slightly outperform the weak learner classifiers in terms of multiple evaluation metrics, including accuracy, probabilities of misdetection and false alarm, while weak learner classifiers outperform in terms of time performance metrics.