Conflict management in intent-based networks

The theme for TM Forum Digital Transformation World (DTW) Ignite 2026 in Copenhagen is "The Future. Faster." As we move further into the decade, connectivity has evolved from a utility into the central nervous system of modern society. Telecommunications (telco) networks now support the most critical infrastructure—from remote healthcare and smart utility grids to national defense and emergency response. The transition from "telco to techco" means that they’re delivering the intelligent foundations that power a global digital economy.

However, this increased reliance on the network brings a new level of complexity. As networks transition toward autonomous network Level 4 autonomy (ANL4), the software entities managing them must act with speed and precision that humans cannot match. The challenge is autonomous entities often have individual and misaligned objectives. When they compete for the same pool of network resources, the resulting conflicts can threaten the stability of the entire network.

This is the focus of the Catalyst project: "Conflict management in intent-based networks."

The hidden friction of automated networks

An autonomous intelligent network is operated by a vast array of software agents. These entities pursue their specific goals—such as maximizing energy efficiency or optimizing signal quality—with individual decisions and actions. Without coordination, these actions inevitably lead to conflicts.  

Traditional automation is often reactive and prescriptive. It relies on static scripts that struggle to handle the dynamic nature of 5G and AI-native environments. If a conflict occurs, human intervention is typically required to mediate. This manual checkpoint acts as a ceiling that prevents service providers from reaching true ANL4 maturity. In phase 1 of this Catalyst project, a hierarchical precedent was used to address the issue of conflict management.

To break through this ceiling, a system is needed to detect and resolve conflicts automatically. Phase 2 of this Catalyst project demonstrates how to protect the network from the effects of conflicts using situational adaptive techniques that are fully automated and aware of business value.

A real-world test: First responders and the Stockholm RAN

In a modern O-RAN environment, the biggest hurdle to autonomy is the "black box" problem. A typical service provider uses specialized RAN applications (rApps) from various vendors—one for power saving, another for traffic steering, and a third for load balancing. Because these apps often operate with proprietary logic in isolation, they can easily issue conflicting commands to the same network resources. One app might attempt to power down a cell to meet a sustainability target while another is trying to boost capacity for a high-value 4K broadcast.

The stakes of conflict resolution are best understood through a real-world scenario. Consider the city of Stockholm, where a shared RAN infrastructure serves 2 very different groups: emergency responders and commercial broadcasters.

In this scenario, a video broadcasting customer requests a 4K live production slice over a network shared with an active slice for first responders. This creates an immediate resource conflict. The broadcaster needs high bandwidth, while the emergency responders require absolute priority and guaranteed low latency.

This Catalyst project solves this conflict by moving away from isolated scripts toward a collaborative intent management loop. By introducing new, open application programming interfaces (APIs), we enable third-party components to participate in a shared autonomous domain. This means rApps no longer act in a vacuum. Instead, they function as intelligent agents that must negotiate and validate their proposed actions against a centralized coordination agent.

The technical backbone of this coordination is a set of knowledge graphs. These graphs provide a unified view of end-to-end network context, aligning disparate intent requirements into a single source of truth. When an rApp proposes an optimization, the system evaluates the action against active service commitments and predefined business guardrails.

If a conflict is detected, such as an optimization that would compromise a first responder's latency, the action is blocked at the O-RAN SMO or non-RT RIC level before it can impact the network. This helps ensure that only one entity has the authority to act on a specific parameter at any given time, transforming the network from a collection of competing disparate environments into a self-governing ecosystem.

Figure 1: Conflict management in intent-based networking

Building the foundation for Level 4 autonomy

The transition to an autonomous intelligent network is a multiyear journey. Most service providers are currently at Level 2 or Level 3, where humans still handle many complex tasks. Moving to Level 4 requires closed-loop control in each domain and coordination between those domains through a healthy knowledge system.

With Red Hat OpenShift, Red Hat provides the common, cloud-native foundation for this transformation, using an open source model to provide transparency, independence from vendor lock-in, and the software integrity essential for national security. At the heart of this disaggregated architecture is open RAN, where Red Hat delivers the O-RAN compliant O-Cloud components necessary to host network functions and applications. By integrating with the Service Management and Orchestration (SMO) framework, the O-Cloud manager provides critical insights into resource availability, allowing for autonomous configuration that eliminates manual conflicts in settings like the Basic Input/Output System (BIOS).

A common cloud-native platform enables operational consistency across on-premise datacenters, edge locations, and hybrid clouds. This consistency is vital for an autonomous intelligent network. It allows automation policies to be applied uniformly, regardless of where the workload is running.

A collaborative effort for industry-wide impact 

This progress is made possible through the collaboration of a diverse ecosystem of industry leaders. The partners driving the technical development of this Catalyst include Aira, AWS, Ericsson, and Red Hat.

The project is championed by a global group of service providers committed to autonomous excellence: Axiata, e&, LG U+, Odido, Rogers, STC, T-Mobile, Telenor Group, Telstra, and Vodafone.

Summary of technical innovation

This Catalyst project marks a significant step forward in the realization of autonomous operations. The core innovations presented this year include:

• Business-aware intent loops: Resolving rApp conflicts by making decisions based on optimized business value.
• Resource availability shared northbound: The availability and configuration of resources in the O-Cloud are shared with the SMO and can be used to help resolve conflicts.
• Autonomous domain authorization: Introducing enforcement points in the SMO and non-real time radio access network intelligent controller (RIC) to protect the network from conflicting actions.
• Knowledge-driven feasibility: Using knowledge graphs to translate service intent into solutions and generate close, feasible alternatives when a conflict occurs.
• Collaborative agentic optimization: A multivendor environment enabled by new APIs that allow different vendors to participate in the same intent management loop.

In combination, these techniques provide a fully autonomous environment that detects and mitigates conflicts without human intervention. This isn’t just about technical efficiency—it’s about ensuring the resilience and stability of the digital infrastructure that society depends on.

Visit us at our booth at DTW Ignite 2026 in Copenhagen to meet our experts and see how we’re transforming the future of the autonomous intelligent network.