CASE STUDY
Reducing alarm noise by 97% and accelerating the resolution of complex network failures
Key Facts
- 01 Customer
-
Tier-1 CSP,
Sub-Saharan Africa - 02 Product
-
Totogi Ontology
- 03 Use case
-
Cross-domain alarm management and accelerated resolution of complex network failures
- 04 Systems
-
NOC
OSS / NMS
RAN
Power
Transmission monitoring - 05 Time frame
-
One week
In brief
A Tier-1 mobile network operator serving more than 10 million subscribers in Sub-Saharan Africa was losing ground on network availability. Its network operations center received more than 1.2 million alarms a week across power, transmission, and RAN domains, most of them redundant or cascading from a single underlying fault.
Working with Totogi, the operator built a cross-domain alarm management capability powered by the Totogi Ontology. It compressed the weekly alarm flood into three root causes it could act on, a 97% reduction in alarm noise, all as an overlay on existing NOC and OSS tools with no rip-and-replace.
Impact
97% reduction in alarm noise across power, transmission, and RAN domains
>75% improvement in mean time to resolution (MTTR)
3 root causes surfaced from more than 1.2 million weekly alarms
Zero existing NOC and OSS tools replaced
We have 47 alerts in one region. Everyone is paged. Everyone is stressed. No one knows where to start.
Challenge
Network availability was slipping below target. The operator’s network operations center was taking in more than 1.2 million alarms a week across power, transmission, and RAN domains, and many were redundant or cascaded from a single underlying issue. Each domain had its own monitoring tool and its own data model, so teams could not correlate events quickly or trace an alarm storm back to its cause.
The result was manual triage across fragmented fault-management tools, delayed resolution, and rising call center volume as customers felt the impact. A regional incident might page every team at once with dozens of alerts and no clear starting point. The operator needed a cross-domain way to reduce the alarm noise, detect failures proactively, and restore visibility and control, without ripping out the tooling its teams already relied on.
Solution
The operator partnered with Totogi to build a cross-domain alarm management capability powered by the Totogi Ontology. Aligned with TM Forum standards, the Totogi Ontology provides a single semantic model for ingesting and interpreting real-time alarms from power, transmission, and RAN systems. It mapped the operator’s data structures into that shared model, with AI handling most of the mapping and computed entities filling the gaps. Custom extensions captured the operator’s processes and terminology.
On top of that shared model, an alert engine ran two correlation techniques: temporal graph networks to detect cascading failures from timing and topology, and machine-learning clustering to surface regional and pattern-based anomalies. Grounded in one semantic layer, the engine compressed the raw alarm flood into a handful of root causes, without replacing NOC tools or re-architecting OSS systems.
Business
impact
The Totogi Ontology cut alarm noise by 97%. In one week, the alert engine took the operator’s live feed of more than 1.2 million alarms, correlated them into roughly 50,000 groups, and resolved them to three root causes affecting 241 sites. The network operations center went from triaging dozens of dashboards to seeing three problems it could act on, each with root-cause context, impact analysis, and a recommended next step such as dispatching a single team.
Mean time to resolution improved by more than 75% as teams shifted from reactive firefighting to real-time, root-cause-driven response. Because the capability was an overlay, none of this required replacing the operator’s existing OSS or network management systems. The same semantic foundation is now reusable for broader AI use cases across customer experience, service assurance, and topology-aware automation.
About the customer
A Tier-1 mobile network operator serving more than 10 million subscribers in Sub-Saharan Africa, holding a dominant position in its national market. Facing mobile penetration above 100%, ARPU pressure, and new MVNO entrants, it has been modernizing how it manages and operates its network at scale. The customer is anonymized, so no name or logo is used.