Blockchain-Based Distributed Data Security for The Internet ofThings Based on Artificial Intelligence

The Internet of Things (IoT) is a highly prevalent technology in various sectors including manufacturing,automation, transportation, robotics, and agriculture, making use of the sensing capabilities of IoT sensors. Itplays a crucial role in the digital transformation and smart revolution of critical infrastructure environments.However, the security and privacy challenges associated with handling diverse data from different IoT devicesare considerable. Adversaries often target the communication between two IoT devices' sensors with theintention of disrupting the regular operations of IoT-based critical infrastructure. In this paper, we propose anarchitecture that combines artificial intelligence (AI) and blockchain to ensure secure data dissemination andaddress the security and privacy concerns of critical infrastructure. Initially, we employ principal componentanalysis (PCA) and explainable AI (XAI) techniques to reduce dimensionality. Subsequently, we utilize variousAI classifiers, including random forest (RF), decision tree (DT), support vector machine (SVM), perceptron, andGaussian Naive Bayes (GaussianNB), to classify the data as either malicious or non-malicious. Moreover, weleverage a blockchain network driven by the interplanetary file system (IPFS) to secure the non-malicious data.Additionally, to enhance the security of the AI classifiers, we investigate data poisoning attacks on the dataset,which involve manipulating sensitive data to deceive the classifier and yield inaccurate results. To mitigate thisissue, we introduce an anomaly detection approach that identifies malicious instances and eliminates thepoisoned data from the dataset. We evaluate the proposed architecture using performance evaluation metricssuch as accuracy, precision, recall, F۱ score, and receiver operating characteristic (ROC) curve. The resultsdemonstrate that the RF classifier outperforms other AI classifiers with an accuracy of ۹۸.۴۶%.