EDGE ARTIFICIAL INTELLIGENCE FOR SECURE AND INTELLIGENT INTERNET OF THINGS SYSTEMS: A COMPREHENSIVE REVIEW

Abstract

The rapid expansion of the Internet of Things (IoT) has resulted in the deployment of billions of connected devices across healthcare, transportation, agriculture, manufacturing, smart cities, and domestic environments. These devices continuously generate large volumes of data that traditionally require transmission to centralized cloud platforms for processing and analysis. Although cloud computing provides substantial computational resources, dependence on remote servers can introduce latency, bandwidth consumption, privacy concerns, and security risks. Edge Artificial Intelligence (Edge AI) addresses these challenges by moving machine learning and intelligent decision-making closer to the location where data are generated. This paper presents a comprehensive review of Edge AI for secure and intelligent IoT systems, focusing on architectural principles, machine learning techniques, security mechanisms, privacy preservation, resource optimization, and practical applications. The paper discusses how lightweight models, model compression, federated learning, and distributed intelligence can improve the efficiency of resource-constrained IoT devices. It also examines challenges related to computational limitations, energy consumption, adversarial attacks, data privacy, model management, and interoperability. Finally, future research directions are discussed, including trustworthy Edge AI, collaborative edge-cloud architectures, efficient learning techniques, explainable models, and sustainable intelligent computing. The study concludes that the integration of Edge AI and IoT can provide responsive, privacy-aware, and scalable intelligent systems when security and resource constraints are considered throughout the system lifecycle.