Assurance leads the way in the move from a scattering of ML deployments across the telco towards a future of more coordinated automations using agentic systems and multiple AI/ML tools to provide closed-loop, intent-based autonomy. This is driven by the need for more than simple, rules-based automations and the deployment of increasingly data-rich environments such as 5G.
Progress to date
Currently, ML is increasingly well-used and those interviewed in recent research were happy with its contributions in areas such as anomaly detection. Its use in predictive algorithms is on the rise in use cases such as the creation of dynamic baselines for KPI monitoring; however, current solutions lack maturity, especially when data is insufficient or poor-quality, or when predicting is difficult due to the appearance of a rare event.
The figure below summarizes a piece of secondary research analysing recent vendor announcements of new AI/ML deployments. The pink text in bold shows where multiple vendors have made announcements, with pink text (non bold) showing less popular areas; black text then shows other places where AI/ML could be deployed but no announcements were found:
The table below presents a proposed agentic architecture to support a telco’s vision for autonomy in assurance and outlines the goals of each architecture layer. The final column identifies new technologies and processes required to achieve these goals:
