AI assessment tools for autonomous networks are key to customer experience

Commercial context

Many CSPs are struggling to increase profits, amidst slow revenue growth and pressure on average revenue per user – at the same time, user experiences are too often marred by service congestion, network failures, and long waiting times for service provision or changes. The aspiration of some CSPs to deliver a ‘zero-X’ network experiences (i.e. zero-wait, zero-touch, and zero-trouble), unfortunately, faces numerous barriers which require a holistic solution. These include high workloads for complaint handling, how to deal with traffic surges in RAN networks, how to achieve end-to-end fault detection and quality monitoring, how to ensure real-time transport resource assurance and network experience in dynamic IP networks, and how to coordinate IP and optical networks effectively.

CSPs are therefore turning increasingly to autonomous network (AN) technologies, both to increase efficiency of provision and offer customers more reliable and personalized propositions in the process, by implementing AN level 3 (conditional automation) and level 4 (high automation) solutions in their networks. Telkomsel Indonesia, for instance, is aiming to achieve AN level 4 by the end of 2025. It’s increasingly clear that deployment of AN is recognized as the holistic solution needed here – according to TM Forum’s 2023 AN Report, 71% of CSPs cite customer experience as the top driver of investment in network automation.

The solution and its applications

The AN empowers distinctive experience for cognitive transformation Catalyst is therefore showing how CSPs can improve efficiency in operations and maintenance (O&M) across IP networks, transmission networks, core networks, and service domains by developing a practical set of AN assessment methods – while also enhancing user experience and achieving higher customer satisfaction with guaranteed SLAs. The participating CSPs include AIS, China Unicom, Telkomsel Indonesia, MTN and Zain.

The Catalyst has helped develop and deploy TM Forum standards-compliant O&M tools, such as an intent-based, end-to-end autonomous system to streamline service provisioning – and AI tools to proactively detect potential customer complaints about the radio network and cell outages, then initiate repair processes. AI-based tools can also enable customers to self-resolve issues arising from spikes in demand. To improve the performance of optical transport networks, the project has also employed AI to help predict demand, while using digital twins to enable differential path computation and traffic optimization, along with fault demarcation in multi-layer transport networks. The Catalyst is also harnessing AI to proactively detect abnormal traffic patterns.

After less than nine months of development and deployment, Telkomsel has reached autonomous network layer 4 (ANL 4.0) in some scenarios. As of August 2023, Telkomsel had used the new O&M tools to resolve 620,000 tasks, with an 87% improvement in the average time to repair a fault, following a reduction in the time it takes to locate a fault from three hours to just two minutes. As a result, the CSP estimates it has reduced complaints by 5,309, retaining US$3.37 million of services revenue. Telkomsel also reports a 3-5% improvement in payload, an increase in resource check efficiency of 90%, and a 6% increase in the downlink speeds enjoyed by customers. These performance enhancements are boosting customer satisfaction - the CSP says the new O&M tools have helped it achieve an NPS 29% above the industry average.

Wider value

The Catalyst has proven operators can achieve high AN levels, even with existing limited AI technology. “This quick win, with great business value achievement, underlines the fact that AN is available for the wider industry and society,” notes Ari Sondang, General Manager for Network Quality Digitalization, at Telkomsel Indonesia. “The most important impact of this project is how TM Forum’s AN assets have guided us to achieve ANL 4.0 by precise re-engineering of a few siloed AI-powered applications and platform interoperability. The underlying benefits of not requiring large upfront investment into the resource layer will benefit the AN community as more champion executives take the leap of faith.”