GPS Spoofing Detection of Small Delay Signals Using Neural Networks an‎d Dimensionality Reduction Algorithms

10/13/2026 12:00:00 AM

GPS-based positioning has become an integral part of daily life. A GPS receiver calculates its distance from a satellite by measuring the signal reception delay. After determining its position relative to four satellites, the receiver obtains its precise three-dimensional location. However, a major drawback of this positioning system is that satellite signals are very weak at the Earthʹs surface an‎d are vulnerable to interference in the ban‎d. As a result, even low-power interference can disrupt the GPS receiver. This dissertation employs spoofing detection based on the Cross Ambiguity Function (CAF). Initially, the introductory chapters of the dissertation provide an overview of GPS, spoofing, an‎d spoofing detection, an‎d review examples of modern an‎d classical methods for spoofing detection. Subsequently, in Chapter 4, the fundamentals of signal processing an‎d artificial intelligence methods required for simulating spoofing detection are discussed. Another challenge in spoofing detection using CAF is the high dimensionality of CAF data, which necessitates powerful hardware an‎d significantly slows down detection time. Additionally, detecting spoofing signals with delays of less than 0.5 chips compared to the main signal becomes difficult. To address this issue, Chapter 4 introduces the Peak Mapping Reduction (PMR) algorithm, based on peak positions, an‎d the SVD_Var algorithm, based on the dispersion of CAF data using singular value decomposition (SVD), to improve the speed an‎d performance of the detection process. These reduced-dimension images are then trained using a base Convolutional Neural Network (CNN). Furthermore, a modified CNN model (TCNN) an‎d a Convolutional Neural Network based on the GRU Recurrent Neural Network, referred to as FCG, are proposed in this dissertation to enhance accuracy. The subsequent chapters analyze simulation results, demonstrating a 98% to 99% improvement in network training speed with reduced dimensions compared to images with original dimensions, a 95% to 100% improvement in detection accuracy compared to the original dimensions, an‎d a 0.2% to 6% improvement in detection accuracy compared to the baseline model an‎d similar works using the proposed algorithms an‎d models.

سامانه موقعيت يابي جهاني , تشخيص فريب , تابع ابهام متقابل , شبكه TCNN , شبكه FCG , الگوريتمهاي كاهش ابعاد

GPS , Spoofing Detection , CAF , TCNN , FCG , Dimensionality Reduction Algorithms.

Mohammad Javad Jahantab

Dr. Mohammad Reza Mosavi