Edge Artificial Intelligence: A Systematic Review of Evolution, Taxonomic Frameworks, and Future Horizons

TL;DR

Edge AI shifts intelligent processing from centralized clouds to the network edge, enabling real-time, privacy-preserving inference with lower latency. The authors employ PRISMA-guided systematic review and introduce a four-dimensional taxonomy (D1–D4) to map deployment location, processing capability, application domain, and hardware architecture across 79 primary studies. Key contributions include a historical synthesis from CDNs to on-device AI, an integrated framework that reveals interdependencies and trade-offs, and a comprehensive analysis of systemic challenges and future research directions, including next-generation hardware, adaptive algorithms, edge-cloud collaboration, and trustworthy AI. The work highlights the practical impact of Edge AI across smart cities, IIoT, autonomous systems, healthcare, and retail, while outlining the pathways and barriers to scalable, secure, and energy-efficient edge deployments in a heterogeneous ecosystem.

Abstract

Edge Artificial Intelligence (Edge AI) embeds intelligence directly into devices at the network edge, enabling real-time processing with improved privacy and reduced latency by processing data close to its source. This review systematically examines the evolution, current landscape, and future directions of Edge AI through a multi-dimensional taxonomy including deployment location, processing capabilities such as TinyML and federated learning, application domains, and hardware types. Following PRISMA guidelines, the analysis traces the field from early content delivery networks and fog computing to modern on-device intelligence. Core enabling technologies such as specialized hardware accelerators, optimized software, and communication protocols are explored. Challenges including resource limitations, security, model management, power consumption, and connectivity are critically assessed. Emerging opportunities in neuromorphic hardware, continual learning algorithms, edge-cloud collaboration, and trustworthiness integration are highlighted, providing a comprehensive framework for researchers and practitioners.

Figures (8)

• Figure 1: Multi-dimensional analytical framework for Edge AI systems, integrating deployment locations, processing capabilities, application domains, and hardware architectures.
• Figure 2: PRISMA 2020 flow diagram of the systematic literature identification, screening, and inclusion process.
• Figure 3: Evolution from centralized cloud computing to distributed Edge AI, mapping key technological milestones to the dimensions of our analytical framework.
• Figure 4: The Edge AI technology stack, illustrating the synergistic relationship between hardware accelerators, software frameworks, and communication protocols that enable intelligent processing across the deployment continuum.
• Figure 5: The spectrum of Edge AI paradigms, classified by Processing Capability (D2), ranging from ultra-constrained TinyML to collaborative Federated Learning.