Vodafone, Google Cloud and TM Forum have published a technical white paper (see below) that sets out a practical framework to help the telecoms industry shift from manual, reactive network operations to proactive, intent-based network autonomy.
Autonomous networks will think and act independently, self-heal and anticipate demand, delivering enhanced customer experiences and greater agility for operators. Rather than manually managing time-consuming tasks, operators use an intent-based model, where they set the desired outcome and the network’s cognitive system determines and executes the actions needed to achieve it.
The paper titled, ‘Self Optimising Autonomous Networks: An Implementation Guide’, marks an important step in Vodafone’s autonomous network strategy. It draws on Vodafone’s engineering network prowess, Google Cloud’s AI capabilities, and TM Forum’s leadership in industry standards for autonomous networks.
Closed-loop automation
As telecoms networks grow more complex, traditional manual tuning and imperative automation (a step-by-step process to programming) are no longer enough. The paper explains how the use of AI, data analytics and closed-loop automation can create self-optimising, autonomous networks that adjust in real time to improve performance, resilience, and efficiency.
The paper describes how operators can use intent-based closed loops to define target network outcomes, such as latency and throughput, instead of relying on repeated manual checks. The closed loop then observes, analyses, decides and acts autonomously to maintain that outcome as conditions change across business, service and resource layers of a network.
Another central theme of the paper is the integration of network controllers with intelligent, cross-domain reasoning. This approach combines the complementary strengths of AI running directly within the network infrastructure with centralised AI agents. For example, while local automation handles immediate tasks like optimising congestion inside a mobile base station, centralised AI agents interpret high-level business goals and predict network needs by analysing diverse data sources, including weather reports and social media feeds.
Translating business goals
AI agents, together with knowledge graphs, network data lakes and digital twins (to test changes before making them), are key enablers of autonomous networks. Collectively, they support enhanced network planning, simulation and orchestration, with security and governance. Human oversight must also be built in to define policies, goals and explicit guardrails for AI agents, preventing unauthorised actions and ensuring that major network changes require a ‘human-in-the-loop’ approval workflow.
One example is how an operator can set strict availability and low-latency requirements for a 5G service which can then be translated into coordinated automated actions across the entire network, eliminating issues such as traffic bottlenecks.
To implement this vision, the paper recommends using a combination of both hybrid and public clouds. In this scenario, ultra-low-latency resource loops running close to the network or within domain controllers are housed in a hybrid cloud, while broader business-level reasoning and AI capabilities are supported by a centralised cloud environment such as Google Cloud.
Unlocking value
Autonomous networks offer major opportunities to improve the customer experience and create value for operators, the paper concludes. According to STL Partners, the estimated potential upside per operator is circa $800 million annually. Realising that value, however, will depend on mastering the orchestration of hierarchical closed loops across different network layers, while maintaining cross-domain coordination and applying the right tools in the right place. This, in turn, will enable the proactive execution needed to unlock greater network autonomy and ultimately superior customer experience.
