Online Anomaly Detection System for Mobile Networks.

Area of Science:

• Telecommunications Engineering
• Network Performance Monitoring
• Data Science

Background:

• The deployment of fifth-generation (5G) networks necessitates enhanced mobile network capacity.
• Smaller cell topologies increase frequency reuse, leading to a surge in data collection nodes and performance metrics.
• Effective management and analysis of these metrics are crucial for network operators.

Purpose of the Study:

• To propose a methodology for the online, real-time detection and tracking of anomalies in mobile network performance indicators.
• To provide network operators with tools for automated information extraction from vast metric datasets.
• To proactively prevent network degradation and minimize user complaints.

Main Methods:

• Development of a novel methodology for real-time anomaly detection in mobile network performance metrics.
• Evaluation of the system's feasibility using diverse performance metrics.
• Validation with a real-world LTE Advanced dataset.

Main Results:

• Demonstrated the system's capability to detect and track anomalies in real-time.
• Achieved effective performance in identifying issues within mobile network data.
• Comparative analysis showed competitive or superior performance against existing state-of-the-art anomaly detection systems.

Conclusions:

• The proposed methodology offers an effective solution for real-time anomaly detection in mobile networks.
• This system aids operators in maintaining network quality and user satisfaction.
• It represents a significant advancement in automated network performance management for 5G and beyond.