Blind Anti-spoofing through Satellite Constellation by Model Predictive Filtering based NNs

Global Positioning System (GPS) receivers are widely utilized across various domains, including everyday life, technological applications, medicine, and military operations. However, ensuring the reliability of GPS remains a significant challenge, particularly due to spoofing interference as one of the most critical threats to system integrity. Over recent decades, numerous spoofing techniques have been developed with the intent to mislead GPS receivers from their intended routes without triggering false alarms. This paper introduces an efficient method for approximating a novel GDOP-based parameter by incorporating model predictive filtering into the training process of neural networks. The proposed approach enables the network to learn the relationship between GDOP and satellite azimuth and elevation, thereby addressing limitations inherent in traditional backpropagation learning. Receiver performance and computational cost associated with neural network–based GDOP approximation and classification are analyzed to identify counterfeit signals under suboptimal satellite constellations. The algorithm demonstrates the capability to compensate for spoofing errors with an accuracy exceeding ۸۵%.

کلیدواژه ها:

GPS NN ، Model Predictive Filtering ، Satellite Constellation ، GDOP