|
Towards zero-touch autonomy: A vision and agentic GenAI architecture for intent-based self-configuration
|
Authors: Yosra Njah, Wael Jaafar, Rami Langar, Bassant Selim, Mohamed Cheriet
Status: Final
Date of publication: 30 June 2026
Published in: ITU Journal on Future and Evolving Technologies, Volume 7 (2026), Issue 2, Pages 133-144
Article DOI : https://doi.org/10.52953/XBJF7355
|
Abstract:
Next-generation networks are evolving toward zero-touch autonomy to manage increasing architectural complexity and dynamic service demands. Intent-Based Networking (IBN) has emerged as a key paradigm in this transition, enabling high-level business objectives, or intents, to be automatically orchestrated across heterogeneous resources. This paper presents a vision, reference architecture, and research roadmap for agentic Generative Artificial Intelligence (GenAI)-driven intent-based self-configuration in autonomous networks, positioning IBN as the core of self-configuration upon which higher-order self-X capabilities, such as self-optimization, self-healing, and self-protection, are built. Building on an analysis of existing IBN frameworks across the intent lifecycle, we identify persistent challenges in intent-policy translation, conflict-free activation, and closed-loop assurance across multi-domain, multi-stakeholder environments, and argue that current IBN implementations lack the dynamic cognitive reasoning required to cope with the scale and heterogeneity of the 6G era. To bridge this gap, we introduce an end-to-end agentic GenAI-driven IBN reference architecture leveraging specialized Large Language Model (LLM) agents that orchestrate intent fulfillment through Knowledge Graphs (KGs) and sufficiency-aware Retrieval-Augmented Generation (RAG-S). The architecture is instantiated within a 6G Open Radio Access Network (O-RAN) ecosystem and illustrated through an Industry 5.0 use case, advancing IBN toward practical, reasoning-aware autonomy and outlining key research challenges and future directions toward trustworthy zero-touch autonomous networks. |
Keywords: Autonomous networks, generative artificial intelligence, intent-based networking, large language models, zero-touch networks
Rights: © International Telecommunication Union, available under the CC BY-NC-ND 3.0 IGO license.
|
|
|
|
| |