INTELLIGENT HOME IOT DEVICES: AN EXPLORATION OF MACHINE LEARNING-BASED NETWORKED TRAFFIC INVESTIGATION

Abstract

In the rapidly evolving landscape of smart homes powered by Internet of Things (IoT) devices, the twin specters of safety and privacy loom large, exacerbated by pervasive security vulnerabilities. Confronted with a heterogeneous array of devices each with unique Value of Service (QoS) requirements, devising a singular network management strategy proves untenable. To mitigate these risks, device categorization emerges as a promising avenue, wherein rogue or vulnerable devices are identified and network operations are automated based on device type or function. This novel approach not only fortifies IoT security but also streamlines network management, offering a multifaceted solution to the burgeoning challenges. Recognizing the burgeoning interest in leveraging machine learning for traffic analysis in IoT environments, this study delves deep into the potential and pitfalls of such techniques. Beginning with a comprehensive framework for categorizing IoT devices, the research meticulously examines methodologies and remedies across every stage of the workflow. Key focal points include the categorization of public datasets, nuanced analysis of IoT traffic data collection methodologies, and the exploration of feature extraction techniques. Through a rigorous evaluation of machine learning algorithms for IoT device classification, the study elucidates emerging trends and highlights promising avenues for future exploration. The culmination of this investigation manifests in meticulously crafted taxonomies, offering insights into prevailing patterns and informing future research trajectories. Moreover, the study identifies and advocates for uncharted territories within this burgeoning domain, propelling the discourse forward and catalyzing innovation in IoT security and management.