In large-scale telco networks, anomalies can quickly propagate and cause widespread service disruptions. By combining machine learning with graph based analysis, this telco group implemented an early warning system that identifies anomalies before they escalate, enabling timely corrective actions.

The client is a leading Italian and international telecommunications group managing a large, complex network that serves millions of users and requires continuous monitoring to ensure service quality.

Ensuring reliable service in a vast and complex network is increasingly challenging.

Key issues included:

  • Difficulty detecting anomalies early in highly interconnected network environments
  • Limited visibility into how local issues could propagate and impact the network
  • The need to analyze large volumes of heterogeneous data in near real time
  • Reactive interventions, often occurring only after users experienced service degradation

Improving early anomaly detection was critical to reducing service disruptions.

The solution is built on Machine Learning and graph based models designed to analyze complex network behavior.

Machine learning algorithms are trained to identify the early stages of anomalies, while graph representations of the network make it possible to understand relationships, dependencies, and potential propagation paths.

Live graph based visualizations allow domain experts to:

  • Inspect anomalies in context
  • Add operational knowledge
  • Accelerate decision making

The system integrates seamlessly with the existing technological stack.

BitBang designed a custom anomaly detection platform that:

  • Continuously analyzes diverse network data in near real time
  • Detects anomalies at their earliest stages
  • Represents the network as a graph to evaluate impact and propagation
  • Enriches algorithmic insights with domain expertise through interactive visual tools

Operations teams receive early alerts together with contextual information needed to act quickly.

The project delivered significant operational benefits:

  • Anomalies are detected before end users experience service failures
  • Operations teams can intervene earlier and more effectively
  • Full scale network failures are often prevented
  • At least 300,000 users per year avoid service disruptions thanks to the system

Overall network reliability and customer experience have improved substantially.