<img height="1" width="1" style="display:none;" alt="" src="https://px.ads.linkedin.com/collect/?pid=1448210&amp;fmt=gif">
Operators Have Already Built the Foundations for Autonomous RAN. So Why Start Again?

Operators Have Already Built the Foundations for Autonomous RAN. So Why Start Again?

November 17, 2026

The industry is progressing quickly to develop autonomous Radio Access Networks (RANs), with network operators facing pressure to reduce their costs while managing increasingly complex networks from multiple vendors. This is driving the need for Artificial Intelligence (AI)-based automation, intent-based operation, and network autonomy. Much of the automation required to support this transition is already deployed in the RANs of operators worldwide.

Many Mobile Network Operators (MNOs) have already invested heavily in network automation technologies, like Self-Organizing Networks (SONs). They have been collecting relevant data for years. So, they already have the foundation needed to start creating value with autonomous RAN and start evolving toward more complex and distributed automation applications. Service Management and Orchestration (SMO) can manage and coordinate existing automation alongside new AI-driven applications, allowing operators to build toward greater network autonomy without replacing everything already in place.

AI will be mandatory in operators' evolution toward autonomous RAN. However, simply introducing AI into existing network environments will not be enough. Operators will need to connect AI capabilities with network-specific data, existing automation functions, and operator-defined policies to enable more advanced decision-making.

 

The Problem with Automation Islands

The bulk of network automation already exists in operators' networks in the form of SON solutions, which have been optimizing coverage, capacity, and network performance for many years. However, these individual systems have largely been developed as standalone applications by different vendors, running on isolated data sources and operating through their own management interfaces. This has created a fragmented landscape of automation tools.

As networks evolve to support cloudified RAN and network slicing, the number of network functions and vendors involved continues to increase. This raises the potential for conflicting objectives between individual automation applications. For example, an application optimizing energy efficiency may make decisions that affect network capacity, while another is attempting to improve performance. These functions need to be coordinated to deliver broader network objectives.

Traditional automation can execute specific tasks and follow predefined rules, but autonomous RAN requires more complex decision-making based on network conditions, operational knowledge, and operator-defined policies. SON is not becoming obsolete. It has accumulated years of operational experience and optimization capabilities in live networks, which will remain valuable as operators introduce more advanced AI-driven applications.

There is also a significant commercial consideration. Replacing existing automation would mean writing off years of investment, potentially introducing considerable new costs and operational disruption. Much of the value needed for autonomous RAN already exists in the optimization algorithms, data, and experience embedded in existing SON systems.

 

SMO as the Foundation for Autonomous RAN

SMO provides the framework to address this fragmentation. It connects existing and new network management systems, optimization applications, data, and AI functions through standard interfaces and common workflows. This allows operators to coordinate automation applications from different vendors, manage how they interact, and oversee their lifecycle across the RAN.

Operators can therefore bring existing SON capabilities together with newer automation applications, including rApps, without requiring a wholesale replacement of their current systems. Open interfaces are particularly important in multi-vendor environments, allowing operators to integrate applications from different suppliers and avoid dependence on a single vendor's management platform.

As operators progress toward higher levels of autonomy, SMO will also become increasingly important for managing AI-driven and intent-based operations. AI can support pattern detection, network behavior prediction, and more complex decisions, but SMO provides the coordination and policy management needed to apply those decisions across the network. The objective is to make existing and new automation work together, while establishing the foundations for more advanced network autonomy.

 

Operator Scenarios

NTT DOCOMO provides an interesting example of how this transition can take place. The Japanese operator introduced its OREX SMO in 2023 to support automation in a multi-vendor environment, including energy efficiency, parameter tuning, and network optimization.

Nokia's MantaRay SON is among the solutions deployed in DOCOMO's multi-vendor 5G network, demonstrating how established SON capabilities can contribute to a broader SMO strategy. DOCOMO's longer-term direction builds on its existing Centralized SON (C-SON) capabilities, with open rApps and AI supporting more advanced automation.

Importantly, operator-managed SMO environments and vendor-provided SMO solutions can coexist through open and interoperable interfaces. This allows operators to retain their own operational tools while integrating vendor-specific SON, SMO, and energy optimization applications across the RAN. Interoperability, however, between the individual applications running on top of autonomous RAN from multiple vendors, will become a key challenge for operators. They will need to manage all these different systems in their live networks, which also requires managing the applications’ lifecycles.

 

The Next Investment Decision Is About Orchestration

Transitioning to an autonomous RAN is not going to be a short-term exercise and as this evolves, the operator’s existing automation is going to continue to play a critical role. AI will be mandatory and its value to network operators will be enhanced by its successful integration with network-specific data, operator-defined policy, and existing automation functions.

While there are cases where legacy systems need to be replaced by more modern automation tools because they lack necessary interfaces, integration capabilities or even sufficient lifecycle support, most automation systems already in place deliver highly positive operational effects.

Operators have already made much of the investment needed to move toward autonomous RAN. The question now becomes whether they can bring those investments together, or whether fragmented systems and competing vendor platforms will make the transition more expensive than it needs to be.

For a deeper analysis of how operators can build autonomous RAN on existing investments, read ABI Research's Executive Brief, “Building Autonomous RAN on Today's Foundations.”

 

Building Autonomous RAN On Todays Foundations Social Card

 

Tags: Telco AI, 5G, 6G & Open RAN


Dimitris Mavrakis

Written by Dimitris Mavrakis

Senior Research Director
Dimitris Mavrakis, Senior Research Director, manages ABI Research’s telco network and cloud computing coverage, including hybrid cloud platforms, digital transformation, and mobile network infrastructure. Research topics include AI and machine learning in telco networks, hybrid cloud deployments and technologies, telco software and applications, 5G, 6G, cloud-native networks, and both telco and cloud ecosystems.

FREE RESEARCH

Stay Ahead of Technology Trends With Free Research

You May Also Like

Key Insights Most operators will make progress toward Level 4 autonomy through individual use cases, but scaling across network domains requires shared data, orchestration, and controls for executing ...

Michael Moreno Michael Moreno

Key Insights: Telco infrastructure vendors are diversifying because traditional network growth is slowing and AI is pulling capital, attention, and product investment away from core telecom markets. ...

Dimitris Mavrakis Dimitris Mavrakis