Scalable and Secure Edge AI: Foundations, Applications, and Open Research Issues

Abstract

Edge Artificial Intelligence (Edge AI) represents a paradigm shift in intelligent computing by relocating model inference and training to the edge of the network. This transformation enables real-time decision-making, reduces data transmission, enhances user privacy, and supports context-aware applications. This paper presents a comprehensive survey of Edge AI, examining its technological foundations, system architectures, deployment strategies, and applications across sectors such as healthcare, transportation, manufacturing, and environmental monitoring. We analyze core challenges including model optimization under hardware constraints, secure deployment, privacy-preserving learning, and ethical concerns. Furthermore, we outline open research problems and discuss future trends including 6G-enabled edge intelligence, the adaptation of foundation models for embedded devices, and collaborative edge learning. The survey aims to provide researchers, engineers, and policymakers with an integrative understanding of Edge AI, guiding the development of scalable, secure, and sustainable intelligent systems.