A Study on AI and Unsupervised Learning Approaches for Clustered Analysis of Attacker Activities in IoT Data

Abstract

The realm of Internet of Things (IoT) has expanded its influence across various applications, spanning from compact wearable devices to large-scale industrial systems, delivering substantial benefits to humanity. With an escalating number of devices equipped with sensing and processing units, the increased interconnectedness poses a heightened risk of data breaches. Security and privacy concerns loom large over diverse applications utilizing IoT technology. The susceptibility of IoT data to malicious activities by attackers, who may compromise the integrity of the data by hacking into IoT devices, is a significant issue.

This paper introduces distinct clustering techniques aimed at identifying potential attacker activities within IoT data. Leveraging AI and Machine Learning techniques, the proposed approach clusters both attacker-modified data and authentic data. The utilization of Bayesian algorithm, Chi-square algorithm, and convergence algorithm contributes to training and validating a model designed to recognize such attacks on IoT data.