Customers have high expectations for fully connected, real-time experiences with 5G. Meeting these exacting requirements is challenging without new tools to transform network operations and empower fast, data-driven decisions for network and service experience management.

Network operations (NetOps) are at an inflection point as engineering teams explore new ways to dynamically manage and optimize network performance and services and gain deeper insights into customer needs and behaviors. New generative AI (GenAI) technologies can augment data analytics and revamp processes, such as customer care and support, network optimization, troubleshooting, etc., to ultimately create a new generation of network assurance.

GenAI can interpret and analyze differing forms of content and media — including text, image, code, unstructured data, etc. — to correlate information across domains. These capabilities represent a breakthrough for analytics and, in turn, NetOps.

This blog highlights findings from Heavy Reading’s 2024 5G AIOps Operator Survey and focuses on the impact of GenAI on analytics and 5G network operations (NetOps). The new survey intends to help the industry better understand the status of network analytics, AI and automation and provide insights into operators' strategies. (To download a copy, click here.)

5G end-to-end troubleshooting

Traditional network monitoring methods often present domain-specific historical and live network data feeds via dashboard consoles for reactive management. Solutions must evolve to proactive approaches to assure the network and meet the stricter key performance measurements (KPIs) of 5G standalone services.

In the figure below, a 2024 5G AIOps Survey question seeks to understand the importance of 5G end-to-end network analytics for troubleshooting. 94% of operators acknowledge the importance of 5G end-to-end network analytics for troubleshooting (based on a group combining "very important" [45%], "important" [35%] and "somewhat important" [14%]). The split of votes between groups "very important," "important" and "somewhat important" may indicate that some respondents currently focus on specific domain solutions (e.g., the RAN) or that they are still determining their long-term strategy.

Only 2% of operators believe 5G end-to-end network analytics for troubleshooting is "not important," recognizing the prominence of these solutions to assure real-time 5G services. 5G troubleshooting is vastly more complex than previous generations, requiring analysis across multiple layers, such as cloud infrastructure, orchestration/ containerized environments, network domains and services, to maintain performance and pinpoint problems. Operators also need more granular insights to manage customer experience (e.g., user application layer latency, service and end-to-end jitter, service operation location, network slice visibility, etc.), which will assist in troubleshooting and assurance.

Many operators will operate complex hybrid environments that include multiple network generations, radio technologies, non-3GPP access, etc., making troubleshooting challenging without different tools and techniques. For operators, these environments will elevate the focus and impact of analytics within their network operations.

How important is 5G end-to-end (i.e., RAN to core) network analytics for troubleshooting?

Process impact

Operators must drive down opex and increase customer satisfaction to maintain a sustainable business. Central to this conundrum is how operators correlate and extract new value from their network and subscriber data. Operators must try innovative approaches, such as GenAI natural language analysis, to extract previously untapped data and increase understanding.

Another question from Heavy Reading's 2024 5G AIOps Operator Survey explores which 5G network GenAI process will have the highest impact, as shown in the figure below.

Operators are keen to use GenAI in technical network operational processes, with over a third (38%) believing GenAI will have the biggest impact on network optimization, followed by network service assurance (32%) and technical customer care and support (17%). Results may reflect confidence from early trials and development involving technologies such as GenAI tools built on telco-specific large language models. These tools can consume massive datasets from existing knowledge bases and correlate them against live network data to assist processes like network optimization, service assurance, customer care, etc., to address telco-specific problems.

Network planning (14%) scores lowest, indicating AI may currently be under-prepared for the complexity of this task. For example, planning can require site surveys, testing, cost analysis and vendor management to ensure capacity, coverage, and capabilities for the latest services and applications.

Early telco GenAI use cases currently focus on assisted scenarios, including locating resources and information, assisted fault diagnosis and resolution, assisted agents, etc. Operators will wish to transform standalone GenAI use cases to fully autonomous operations, but this will take time and vary across domains to prove accuracy, data governance and regulation.

Which generative AI process will have the highest impact on the 5G network?

Conclusion

To deliver real-time services and new capabilities, operators must use groundbreaking technologies to assure the end-to-end network. It is early days, but moving to incorporate new techniques like GenAI alongside analytics offers the ability for faster insights and network performance excellence to forge an unparalleled customer experience.

For more information on this topic and the survey, check out this report: Heavy Reading's 2024 5G AIOps Operator Survey.

This blog is sponsored by RADCOM.