AI-Driven Edge Intelligence for Secure and Energy-Efficient Internet of Things Networks: A Comprehensive Review

Abstract

The rapid expansion of the Internet of Things (IoT) has catalyzed a shift toward decentralized computing, processing data closer to sensors. However, traditional cloud computing induces severe network latency, excessive transmission energy expenditure, and heightened privacy risks. Edge Intelligence (EI)—integrating Artificial Intelligence (AI) directly into edge nodes—provides an effective solution. This review paper presents a detailed examination of AI-driven edge intelligence for secure and energy-efficient IoT networks. We analyze model compression techniques (quantization, structural pruning, knowledge distillation), privacy-preserving collaborative learning (Federated Learning), and lightweight intrusion detection systems. By synthesizing empirical evaluations, analytical trade-offs, and simulation paradigms, this paper highlights research gaps, presents a multi-tier architectural benchmark, and outlines future avenues for self-sustaining, resilient Edge-IoT ecosystems.

KEYWORDS: Edge Intelligence, Internet of Things (IoT), Federated Learning, Model Compression, Energy Efficiency, Cyber-Physical Security, Edge Computing.