High accuracy indoor positioning system using Galois field-based cryptography and hybrid deep learning.

Area of Science:

• Computer Science
• Electrical Engineering
• Cybersecurity

Background:

• Global Positioning System (GPS) is ineffective in indoor environments.
• Traditional indoor positioning systems (IPS) face challenges in accuracy, resilience, and security due to environmental complexity and signal noise.
• Smart manufacturing and logistics require robust indoor location solutions.

Purpose of the Study:

• To develop an accurate, resilient, and secure indoor positioning system.
• To leverage deep learning and advanced cryptographic techniques for enhanced IPS performance.
• To address the limitations of traditional IPS in complex indoor settings.

Main Methods:

• Utilized a two-phase approach: offline data collection and online location classification.
• Employed signal processing for noise reduction and data augmentation, followed by DBSCAN clustering.
• Developed the Deep Spatial-Temporal Attention Network (Deep-STAN), a hybrid model combining CNNs, ViTs, and LSTMs with attention mechanisms.
• Integrated Elliptic Curve Cryptography (ECC) for data encryption, QR codes for location marking, and blockchain for immutable data storage.

Main Results:

• Achieved high performance metrics: accuracy of 0.9937, precision of 0.987, sensitivity of 0.9898, and specificity of 0.9878.
• Demonstrated sustained accuracy (0.9804) even with 80% data usage, indicating model stability.
• The integrated ECC, QR codes, and blockchain significantly enhanced data integrity, confidentiality, and system resilience.

Conclusions:

• The proposed deep learning-based IPS with advanced security features offers a stable and flexible solution for indoor positioning.
• The hybrid Deep-STAN model and cryptographic enhancements provide superior accuracy and security compared to traditional methods.
• The system is well-suited for real-world applications requiring low-latency, secure, and reliable indoor location services.