Using GenAI and knowledge graphs to transform network operations and maintenance

Commercial context

Traditional network operations models present a range of challenges that are resource-intensive and costly for CSPs to remedy. Complex networks, large data volumes, diverse problem types, rapidly changing demands, difficult manual maintenance, and high flexibility requirements are just a few of the issues with which CSPs typically must grapple.

Moreover, even discovering network issues often involves numerous manual processes such as passive alarms, active inspections, and in some cases simply user complaints. CSPs must then use network management tools and expert knowledge to locate problems and perform root-cause analysis, leading ultimately to manual resolution by a technician. All-in-all, traditional models mean long troubleshooting times, low automation levels, and high levels of expertise required to resolve operational issues.

The solution

The GenAI and knowledge driven 5G operations Catalyst breaks with traditional network operations models that rely primarily on expert knowledge and manual work. The first of its kind in telecoms, the Catalyst is developing a solution to implement 5G intelligent operations using generative AI to construct large communication models and a communication knowledge graph to perceive the quality and condition of the network and its services. This system can classify and perform root-cause analysis of problems, and then execute automated resolutions.

The project uses intent-driven techniques to translate solutions into network commands, greatly saving time on the complexity of traditional network operations. Building on TM Forum’s Open APIs, the solution enables intelligent closed-loop automation to create and sustain a network capable of advanced self-cognition. As well as greatly advancing the intelligence and efficiency of network operations, the solution can make a variety of other savings such as reducing 5G network operations and maintenance (O&M) costs, physical labor, and resource consumption.

This Catalyst is the world’s first example of how to build a 5G network that uses intelligent operations based on communication models and communication knowledge. With training data in the scale of terabytes and a model with billions of parameters, this approach achieves a unified perception of networks and services, unified classification, root-cause analysis of issues, and automatic generation of solutions.

Applications

The project has already been successfully launched in two provinces of China Unicom, improving O&M efficiency by over 10 times. VoNR voice service quality improved significantly, enhancing the user experience for hundreds of millions of users. This not only reduced O&M costs, but also improved network quality.

Jianjian Yang, Director at China Unicom, explains how “through this project, we have accumulated transferable large models and datasets in the telecom field, providing a solid technical foundation for developing similar AI applications in the future. Collaboration with industry partners has also given us a leading position in empowering networks with AI. Overall, this project promoted our transformation towards intelligent and autonomous network O&M. It also provided critical support for building high-quality 5G networks, enabling us to provide better digital services to our customers and society. This is of great significance for our digital transformation.”

Wider value

By using intent-driven techniques, these processes take the form of smart network commands, effectively shielding CSPs from much of the complexity of network operations. By implementing this AI-based 5G network capable of intelligent O&M, the wider telecoms industry is a major step closer to greatly improving network O&M efficiency and ultimately digital transformation. To find out more about the project join us at DTW Ignite in Copenhagen between 19-21 September 2023, or get in touch with the Catalyst team directly here.