How to make networks more agile and robust with generative AI

Commercial context

As customer demands evolve, effective monetization of telecoms resources depends on being able to dynamically bundle product components and flexible contract terms. At the same time, the growing complexity and sophistication of telecoms networks means CSPs can face serious challenges in their brownfield network operations and ecosystems. Unresolved issues in a CSP’s network operations center (NOC) can lead to operational inefficiencies, revenue loss and support team overheads, ultimately impacting customer experience and investor confidence.

With the number of network events increasing rapidly, CSPs need an automated approach to address alarm noise, correlating alarms and other events, and to analyze root causes and thereby optimize assurance outcomes. To become more sustainable and lower operating costs, CSPs also need to optimize their energy consumption across various network domains.

The solution

The ReNOVATE AI: Rejuvenating network build and operations through AI Catalyst is using generative artificial intelligence (genAI) to help CSPs increase operational efficiency and pave the way to more sustainable autonomous networks that will enhance customer experience. In particular, the project team is using genAI to support intelligent and dynamic network design and planning. The goal is to build (and then standardize) a genAI module that can take the existing network designs, service designs, network security configurations and other information to generate designs dynamically, based on a specific order.

“Using variations of the large language models (LLMs), such as narrow language transformations or private LLMs, telcos can look to accelerate the process of network build and evolution,” explains Sreeraj Sivadasan, Head of Engineering, SRIMS, BT. “This Catalyst tries to explore and, in future, possibly standardize, various genAI modules which can then be reused in various areas of telecoms, such as the marketplace, plan and build of networks, and network assurance.”

To help automate network assurance, the Catalyst is developing APIs within an AI-ops framework (based on the TM Forum Open Digital Architecture) that can perform root-cause analysis. The solution will detect anomalies against pre-defined KPIs and implement corrective actions to enable closed-loop operation across network applications. Using the cloud infrastructure provided by the hyper-scalers, the solution will also deliver AI-driven assistance to NOC teams for expedited resolution of network incidents. By minimizing disruption, the framework will help to optimize operating costs and customer experience.

Applications and wider value

When the solution is complete, CSPs will be able to conduct real-time analysis and near real-time corrective action as events arise in their NOCs. The goal is to increase network operational efficiency by 15% and thereby improve customer experience. By harnessing genAI, the solution will also empower CSPs to build bespoke products, bundles, and networks quickly, strengthening their operational agility and automation, while ensuring seamless customer journeys. The Catalyst is aiming to reduce the time required for network and service design activities by 25%.

The solution could also employ AI to optimize the energy consumption of the radio access network (RAN). As it identifies when assets at the network edge are idle, it will trigger switching between these elements to minimize the energy usage based on the current state of that network asset. If the solution can lower operational expenditure by 15% through the optimization of RAN assets/elements, that could save a typical CSP millions of dollars a year.

The work of the Catalyst points to the potential of genAI to make telcos far more agile and responsive. “AI holds significant promise for the telecoms industry in the coming years,” concludes Sreeraj Sivadasan of BT. “Whilst traditional machine learning based algorithms and AI models have the potential to decipher correlations and patterns within the telco network data, genAI equips the operators to unlock the complexities of the networks through the right kind of engineered prompts, leading to the generation of various insights and artefacts for network build and creation.”