Toward End-to-End Intent-Based Networking for 6G: Architectures, Challenges, Recent Advances, and Future Directions
Intent-Based Networking (IBN) has emerged as a promising paradigm for simplifying network management by allowing operators and applications to specify high-level service objectives rather than low-level device configurations. Early IBN research was mainly developed in Software-Defined Networking (SDN), Network Function Virtualization (NFV), transport networks, core networks, and data-center environments, where programmability, virtualization, and relatively stable infrastructure models enabled intent translation, orchestration, and assurance. However, realizing IBN in end-to-end mobile networks is more challenging because the Radio Access Network (RAN) is highly dynamic, wireless-channeldependent, mobility-sensitive, and governed by multiple control timescales. The emergence of Open RAN (O-RAN) changes this landscape by making the RAN programmable, disaggregated, data-driven, and control-lable through non-real-time and near-real-time intelligent control loops. This survey reviews the evolution of IBN from SDN/NFV-enabled automation toward O-RAN-driven end-to-end intent-based networking for 5G-Advanced and 6G. We discuss architectural mechanisms, key challenges, recent advances in AI-driven and agentic IBN, and future research directions including Large Language Model (LLM)-based intent translation, contractbased O-RAN slicing, digital twin-assisted validation, and trustworthy closed-loop orchestration.