By: Kenan Jarah

Challenges in Driving Autonomy 
in Telecom Networks

Telecommunication networks are undergoing a fundamental transformation driven by cloud-native architectures, virtualization, distributed infrastructure, and increasingly complex multi-vendor ecosystems. Modern environments comprise highly distributed network functions, microservices, and radio elements that continuously generate large volumes of telemetry across multiple operational domains.

Ensuring reliable service delivery requires Communications Service Providers (CSPs) to continuously monitor and correlate performance metrics, fault events, and service indicators across domains, including the Radio Access Network (RAN), the core network, and cloud infrastructure. While the availability of data has significantly increased, the ability to consistently interpret and act on that data across domains remains a challenge.

Despite advances in analytics and automation, operational workflows remain constrained by fragmented toolchains, domain-specific silos, and reliance on manual investigation and escalation processes. Root-cause identification across domains remains slow, visibility across multi-vendor environments is often limited, and automation is typically confined within domain boundaries. These structural constraints directly affect operational efficiency and service reliability.

At the same time, the rapid advancement of Machine Learning and Generative AI introduces new capabilities. Machine Learning enables deterministic analysis of large-scale telemetry, including anomaly detection and forecasting, while Agentic AI introduces the ability to coordinate actions, reason across systems, and enable goal-driven execution.

Rethinking Operational Architecture: Toward Distributed Intelligence

A key observation emerging from modern telecom environments is that operational complexity is not driven solely by data volume, but by the distribution of intelligence across domains and systems. Traditional approaches have often relied on centralized data aggregation to enable analysis; however, in distributed and multi-vendor environments, such approaches introduce challenges related to scalability, governance, and latency.

The architectural approach explored in this article shifts the focus from centralizing data to coordinating intelligence. The architecture minimizes reliance on full data centralization by combining a structured data foundation with localized processing across systems. This allows intelligence to remain within each domain or vendor environment, while coordination is achieved through structured interaction.

This model enables each system to contribute its domain-specific capabilities within its own context, while higher-level orchestration combines these contributions into coherent operational outcomes. The result is a more scalable and adaptable approach to managing distributed telecom environments.

Way forward with Multi-Agent Operations 

In response to these challenges, an Agentic AI-based operational framework can be designed to enable coordinated network assurance across domains and vendors.

At the core of this approach is the ability to correlate and analyze events across multiple domains, including RAN, core, and cloud infrastructure, without relying on isolated tools. This supports a unified interpretation of performance, fault, and service data, improving the ability to identify and resolve issues efficiently.

Multi-vendor collaboration is treated as a fundamental requirement. The architecture enables interoperability between independent systems while preserving their autonomy. Vendor-specific intelligence remains within each system but can be accessed through structured interaction, reducing reliance on manual escalation and improving resolution cycles.