From Intent to Implementation: Using Generative AI for Code Modules in Autonomous Networks
Supervisor: Dr Paul Harvey
Industry Partner: Rakuten Mobile, Japan
School: Computing Science
Description:
Telecommunication networks are the connectivity fabric that allows us to work, play, study, and call for help when we need it. This means that their high-quality continuous operation is essential for all of us.
In the last 100 years, the way that we use networks has changed a lot – to the point where our toothbrushes, toilets, cars, tvs, and home assistants use them as much as people. All this added pressure means that running these networks is now beyond the human capability to handle thousands of events requiring sub-second response times. Also, given that telco’s annual labour budget is ~$273 billion, hiring more people is not a practical option without increasing subscription costs.
In light of this, network operators want their networks to become autonomous networks: networks which can optimise, configure, heal, and adapt by themselves with little or no human involvement.
The ITU-T Autonomous Network (AN) architecture provides a standards-based framework for self-managing networks, capable of configuring, observing, optimizing, and healing themselves through closed-loop automation [ITU-T Y.3061, 2023]. A key aspect of this architecture is the use of modular controllers composed of independently developed modules, such as workflows or closed-loop functions, which can be flexibly combined to implement network autonomy.
This project investigates the use of Generative AI—specifically, Large Language Models (LLMs)— to automatically generate reusable code modules that align with the ITU-T AN reference framework. By providing high-level functional intents or module specifications, an LLM can synthesize prototype code (e.g., in Python or YAML) for components such as monitoring agents, orchestration stubs, or intent translation functions. This aligns with recent work on LLM-aided design, where generative models are used for structured system prototyping and code generation.
The goals are twofold: (1) to evaluate the feasibility of using generative AI for rapid prototyping of modules within a standards-based network automation framework, and (2) to assess how such AI-generated modules can be validated for conformity with the ITU-T AN architecture. This exploration highlights how AI can accelerate the development and adaptability of next-generation autonomous networks.
Concrete Outcomes are:
• Proof-of-concept implementation of LLM-generated control loops
• Performance evaluation on simulated networks using NS3, or SimpleCDNSim
• Evidence-based documentation of the feasibility of the approach and open-source models