BACHAAV: machine learning-augmented human-AI and cryptographic architecture for threat detection in IoT-enabled oil and gas industrial networks

BACHAAV: machine learning-augmented human-AI and cryptographic architecture for threat detection in IoT-enabled oil and gas industrial networks

연구 분야: Safety

학회: International Journal of Information Technology

The implementation of Internet of Things (IoT) related technologies in the Oil and Gas Industrial sectors has drastically changed and elevated the industrial network’s critical infrastructure, exposing it to multiple cyber-attacks and threats. To handle such complex threats, it is very important to have an innovative, secure, and effective solution for real time threat response. The proposed novel hybrid architecture implements elliptical curves cryptography and block chain inspired zero knowledge proof in the edge layer that preserves the integrity and confidentiality of the IoT devices layer by preventing unauthorized access and data manipulation. At the fog layer, the architecture’s machine learning algorithms process received information from several IoT devices installed on different industrial zones of oil and gas industrial networks, which creates opportunities for monitoring and predictive analysis of cyber threat anomalies. The architecture employs deep learning techniques through the application of Convolutional Neural Networks (CNN) and long short-term memory (LSTM) to detect specific indications of an intrusion or malfunction of the system. The combined Edge and Fog Layer with Elliptic Curve Cryptography (ECC), Zero-Knowledge Proof (ZKP), CNN, LSTM provided 97.1% of accuracy with 335 ms processing latency which resulted in efficiently processed anomalies in IoT devices (sensor) data. Evaluation metrics, including 95% Confidence Interval (CI), revealed the performance of the system was steady with the rates of false positive being 2.35% and negative rates at 3.2% proving real-time anomaly detection and data integrity.