GNSS Interference Identification Driven by Eye Pattern Features: ICOA-CNN-ResNet-BiLSTM Optimized Deep Learning...

Area of Science:

• Signal Processing
• Artificial Intelligence
• Cybersecurity

Background:

• Global Navigation Satellite Systems (GNSSs) face significant security challenges from signal interference.
• Existing methods for interference detection and classification often lack efficiency and accuracy.

Purpose of the Study:

• To propose a novel deep learning framework for intelligent classification of GNSS interference types.
• To enhance the security and reliability of GNSS operations through advanced signal analysis.

Main Methods:

• Transforming GNSS signals into 2D eye diagrams for visual representation.
• Utilizing entropy-centric feature analysis for interference discrimination.
• Designing a hybrid deep learning architecture (CNN, ResNet, BiLSTM) for signal analysis.
• Employing an improved coati optimization algorithm (ICOA) for hyperparameter tuning.

Main Results:

• The proposed method achieved high performance metrics: 98.02% accuracy, 97.09% precision, 97.24% recall, 97.14% F1 score, and 99.65% specificity.
• Demonstrated significant improvements over existing models on diverse jamming datasets.
• The ICOA algorithm showed over 30% improvement in convergence accuracy compared to mainstream methods.

Conclusions:

• The developed eye diagram-based deep learning framework offers an efficient and accurate solution for GNSS interference identification.
• The entropy-aware feature extraction and hybrid network architecture are key to the method's success.
• This research provides a practically feasible approach to bolster GNSS security against various interference threats.