Carriers & OperatorsSingtelArtificial Intelligence5gNetwork Automation
Singtel Cuts Network Fault Resolution Times by 80% with AI
Singtel says AI has cut network problem resolution times by almost 80% in a year, detecting faults before customers notice. Engineers still make the final call.
Carriers & OperatorsWhy it matters
- Singtel cut network fault resolution times by nearly 80% over the past 12 months using AI.
- The operator's 1-5-60 regime targets awareness in one minute, isolation in five, restoration in 60.
- AI optimises radio resources in real time, extracting more capacity from limited spectrum.
- Cheong Hai Thoo: 'You cannot outsource the decision making and accountability.'
- AI workloads are driving 'tremendous' 5G traffic, opening enterprise robotics and XR opportunities.
The story
Singtel has cut network problem resolution times by almost 80% over the past 12 months, Cheong Hai Thoo, VP of mobile network engineering at the Singapore operator, revealed during a keynote at Mobile World Live's Unwrapped – The 5G Evolution event.
The gain comes from AI that detects abnormalities and optimises fault management across the operator's mobile network. Cheong said AI has shifted how Singtel runs operations, replacing predefined rules and manual intervention with systems that can "learn, predict, adapt, and are increasingly able to make decision autonomously".
How does the system actually work?
The AI monitors network key performance indicators and flags "small deviations" from normal operating behaviour. Catch those deviations early and incidents never become outages. Cheong said Singtel has on several occasions resolved problems before customers noticed anything was wrong.
The operator formalises this discipline in what it calls the 1-5-60 approach to fault management. Engineers should become aware of an issue within one minute, isolate it within five minutes, and restore service within 60 minutes. An 80% cut in resolution time suggests the framework now has serious machinery behind it.
The engineering job is changing too. Cheong said AI has freed engineers from spending hours correlating network alarms, pushing them towards "higher value engineering decisions" that improve the overall network experience. Fewer spreadsheets, more judgement.
What about the radio side?
The same technology optimises radio network resources in real time, improving uplink and downlink performance and squeezing more capacity from limited spectrum. That matters in a market where spectrum is scarce, and Cheong didn't understate it: spectrum is "a very precious resource in radio network".
Traffic is growing as well. Cheong pointed to "tremendous" 5G volumes driven by AI workloads, which he framed as an opportunity rather than a burden. Singtel aims to support enterprise applications including robotics and extended reality through differentiated connectivity, network slicing, and tuning uplink capacity and latency to meet emerging applications' requirements.
Will AI replace the engineers?
Not on Cheong's watch. He stressed that automation would not remove the need for human oversight, particularly when operating critical national infrastructure. He acknowledged concerns around AI replacing jobs, but argued the technology should motivate employees to "adapt and learn".
His dividing line is straightforward. The technology can "analyse and recommend corrective action", but engineers must keep responsibility for decisions. "You may outsource a lot of analytics work, detections and recommendation, but you cannot outsource the decision making and accountability," Cheong concluded.
It's a familiar bargain: the machine narrows the options, the human signs off. Singtel's numbers suggest the machine is already doing most of the narrowing. The operator will next be judged on whether its 1-5-60 targets hold as AI-driven 5G traffic keeps climbing.
Also reported
Source: Mobile World Live
More from Amara Osei
Show full bio
News editor covering technology and product news in Carrier business, 5G and 6G, fiber and broadband at Telecom Gazette.
553 articles