GPS Threat Detection Using CG-Trained ANN Framework

Abstract

The widespread dependence on Global Navigation Satellite Systems (GNSS), particularly GPS, has significantly increased vulnerabilities to intentional threats such as spoofing and jamming attacks. While various machine learning techniques—including Convolutional Neural Networks (CNNs), ensemble methods, and deep learning architectures—have demonstrated promising results in threat detection, the exploration of advanced optimization algorithms for training Artificial Neural Networks (ANNs) remains limited. This paper provides a comprehensive review of GNSS security threats and existing machine learning-based detection approaches. It proposes a novel research framework that employs the Conjugate Gradient (CG) optimization algorithm for efficient ANN training. Furthermore, it details the integration of CG-trained ANN (CG-ANN) models to achieve accurate, low-latency, and real-time detection of spoofing and jamming in GPS signals, potentially enhancing GNSS resilience in critical applications.