Design of a Secure Biometric Network Traffic Packet Data Transmission using Deep Learning and Encryption Techniques

This study presents a secure and efficient system for processing and transmitting biometric network traffic packet data by integrating Convolutional Neural Networks (CNN) and Paillier Homomorphic Encryption (PHE) techniques. In the system, the CNN model developed was used for feature extraction and PHE for secure data transmission. The system addresses the critical need for privacy-preserving solutions in applications such as biometric authentication, secure access control and Internet of Things (IoT) security. The CNN model, trained on a biometric dataset using Google Colab, achieved a testing accuracy of 98.5% while the PHE Scheme was implemented in NS-3 to encrypt and transmit biometric data securely and the results of the implementation reported that it achieved an average encryption time of 12ms and decryption time of 8ms. Network simulations in NS-3 demonstrated the system's efficiency and reliability, with an end-to-end latency of 25ms under normal conditions and 99.8% packet delivery under 1% packet loss. The study's results highlighted that the system has good potential to revolutionize biometric data processing and network packet transmission, providing a robust solution for privacy-preserving applications. The study further recommends that future work should focus on optimizing the system for real-time use, supporting non-linear operations in homomorphic encryption, and testing in real-world scenarios.