Bridging network and business: a new approach redefining RAN automation

As competition intensifies in a market dominated by data, RAN automation has become key. To maintain an upper hand, CSPs need to develop faster issue resolution for new services, consistently higher network performance, and lower operational costs.

Previous attempts to address these RAN automation needs have fallen short of delivering the promised benefits. While SON specifications emerged back in 2008, their solutions have been inadequate: Typically, they focus on the network without considering broader business impact, while remaining largely vendor-specific. To compensate, CSPs largely developed their own in-house tools, resulting in a tangled mess of automation systems – and a host of automation challenges, such as:

• Fragmented automation systems & tools: Hundreds of automations spread across in-house systems, with SON and third-party platforms that are expensive to manage & maintain.
• Limitations of rules-based approaches: “if-then” style automation has hit a ceiling in the levels of automation achievable in complex radio environments.
• Network-centered: Current RAN automation remains primarily “network-driven” rather than taking a business/customer-driven approach, limiting its value to the organization.
   

We recognized that to achieve greater levels of automation, a new approach was necessary – providing a single home that could tap into the latest advances of AI/ML, and exposing an agent-ready interface to both the network and the business sides of the organization. The vision led us to develop three core elements:

• Platform rationalization: Migrate to a RAN automation platform that works across all vendors – not necessarily creating a single platform to replace existing systems, but one that oversees control and coordination of network automation, including existing SON systems.
• Apply systematic intelligence: Automate more through increased use of machine learning and digital twins. Rather than just making the models and digital twins available, we sought to streamline and simplify how automation accesses these capabilities.
• Business-ready agentic RAN: Create an open, agentic RAN that can support ever-changing business goals, with the objective of positioning RAN intelligence and automation as a tool that can be leveraged by both network and business AI agents.
   

Implementing the vision: RAN Automation 2.0

Building on these core elements and years of industry experience, we developed Amdocs Cognitive RAN Automation to put this next-generation approach into practice. By leveraging industrialized open-source software, we sought to bridge the gap between network operations and business results, delivering flexibility, vendor neutrality, and ML-powered intelligence capable of supporting both existing and emerging network architectures. 

Differentiators of our approach include:

• Control & flexibility across RAN vendors & technologies: Based on O-RAN Alliance NONRTRIC standards, the platform is vendor-neutral, with support for both traditional and cloud-based RAN environments.
• AI/ML framework: Provides pre-integrated, higher-level AI/ML building blocks with foundational models to support forecasting, anomaly detection, and KPI correlation – allowing app developers to deliver incremental value to their automations.
• Integrated, federated digital twins: A RAN digital twin system helps operators model and predict future network states, offer recommendations, and autonomously take actions. By leveraging multiple twin types – including network simulation, tower inventory, and vendor parameters – it allows targeting different levels of network abstraction for increased accuracy.
• Extensive rApp ecosystem: The platform includes pre-built rApps for energy savings and massive MIMO, complemented by a partner network for developing and certifying third-party rApps.
• Business-ready intelligence: By enabling intelligent network agents that allow natural language communication with the system and access to user data (alongside network data), automations are driven by business needs rather than purely network technology requirements.

By adopting this next-generation approach, CSPs can position their organizations to move beyond technical limitations and harness network intelligence for true business advantage – and finally realize the full potential of their RAN investments.