AI and Blockchain-Based Secure Data Dissemination Architecture for IoT-Enabled Critical Infrastructure.

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Blockchain Technology

Background:

• The Internet of Things (IoT) is crucial for critical infrastructure, but faces security and privacy challenges due to heterogeneous data and sensor communication vulnerabilities.
• Attacks targeting IoT sensor communication can disrupt essential services in manufacturing, transportation, and agriculture.

Purpose of the Study:

• To propose a novel artificial intelligence (AI) and blockchain-driven architecture for secure data dissemination in IoT-based critical infrastructure.
• To address security and privacy concerns arising from heterogeneous data handling in IoT environments.

Main Methods:

• Dimensionality reduction using Principal Component Analysis (PCA) and Explainable AI (XAI).
• Classification of data as malicious or non-malicious using AI classifiers: Random Forest (RF), Decision Tree (DT), Support Vector Machine (SVM), Perceptron, and Gaussian Naive Bayes (GaussianNB).
• Secure data storage using an Interplanetary File System (IPFS)-driven blockchain network and anomaly detection to counter data poisoning attacks.

Main Results:

• The Random Forest (RF) classifier achieved the highest accuracy at 98.46%, outperforming other evaluated AI classifiers.
• The proposed architecture demonstrated effectiveness in identifying malicious data instances and removing poisoned data.
• The IPFS-driven blockchain network provided enhanced security for non-malicious data.

Conclusions:

• The integrated AI and blockchain architecture offers a robust solution for securing IoT data in critical infrastructure.
• The study highlights the efficacy of advanced AI techniques, particularly RF, in detecting and mitigating cyber threats within IoT ecosystems.
• Anomaly detection is crucial for maintaining the integrity of AI classifiers against data poisoning attacks in IoT security.