Integration of 5G/6G Communication Networks and Artificial Intelligence for Intelligent Connected Systems: A Structured Review of Architectures, Enabling Techniques, and Open Challenges

Ananya R. Sen, Rohit K. Das

Abstract


ABSTRACT
The convergence of fifth-generation (5G) and emerging sixth-generation (6G) communication networks with artificial intelligence (AI) is reshaping how connected systems sense, communicate, and act. Network densification, millimetre-wave spectrum, massive antenna arrays, and heterogeneous service demands have made manual, rule-based network management increasingly unsustainable, while AI offers data-driven adaptation across the radio, edge, and core domains. This review presents a structured synthesis of the literature on 5G/6G–AI integration, organised around an AI-native layered architecture, five families of AI techniques (deep learning, reinforcement learning, federated learning, graph and attention models, and generative or semantic AI), and six network functions. Findings are compared against International Telecommunication Union (ITU) performance targets and mapped to four application domains: healthcare, autonomous mobility, industrial automation, and agriculture. The synthesis shows that deep learning and reinforcement learning are the most mature enablers for physical-layer and resource-management tasks, that federated learning is the most credible route to privacy-preserving intelligence, and that generative and semantic AI remain largely exploratory. Persistent barriers include scarce operator data, poor model generalisation across deployments, limited explainability, the energy cost of AI itself, and the security of learning pipelines. A research agenda centred on trustworthy, energy-aware, and standards-aligned AI-native networks is proposed.

KEYWORDS: 5G, 6G, Artificial intelligence, Machine learning, Intelligent connected systems, Open RAN, Edge intelligence, Federated learning, Network slicing, Semantic communication

Full Text:

PDF 1-15