Smart Service Analyzer for ML-Based Mobile Network Capacity Planning

The patent replaces manual mechanical analysis methods with automated machine learning-based processing. The system automatically collects CDR data, processes it through ML models to generate KPIs, CEI estimates, and SQI estimates, eliminating the need for time-consuming manual capacity planning and performance estimation while keeping pace with exponential traffic growth

The system enables self-service automated performance estimation by continuously collecting network data, processing it through embedded machine learning models, and generating performance estimates without requiring manual intervention. The automated pipeline processes CDR data to produce actionable intelligence independently, freeing network planners from repetitive manual analysis tasks

The patent segments the complex data processing task into distinct modular stages: data collection from probing devices, CDR data processing to generate KPIs, CEI estimation using ML models, and SQI estimation using ML models. Each stage processes specific aspects of the data independently, reducing overall processing complexity while maintaining comprehensive analysis and high estimation accuracy

The patent introduces Key Performance Indicators (KPIs) as intermediary metrics between raw CDR data and final service performance estimates. The ML models first process CDR data to generate KPIs, which then serve as inputs for CEI and SQI estimation. This intermediary layer simplifies the processing of large-volume raw data by transforming it into meaningful intermediate metrics that are easier to analyze

The patent applies preliminary action by using machine learning models to predict future network performance and identify capacity requirements before actual capacity expansion is needed. The system continuously estimates service performance metrics (CEI and SQI) and projects future network states, allowing planners to prepare capacity expansion plans in advance based on accurate predictions rather than waiting for performance degradation to occur

The patent implements a feedback mechanism where the system continuously monitors actual network performance through CDR data collection and compares it against predicted performance. The ML models are trained on historical data and continuously refined based on actual outcomes, creating a closed-loop system that improves prediction accuracy over time and enables more precise capacity planning that reduces unnecessary expansion cycles

The patent extracts only the most relevant and actionable information from the comprehensive CDR data through the KPI generation process. Instead of processing and analyzing all raw data, the system extracts key performance indicators that capture essential service performance characteristics, then uses these extracted metrics as inputs for ML-based CEI and SQI estimation, significantly reducing data volume while maintaining complete performance visibility

The patent transforms raw CDR data into different parameter representations at each processing stage. The system changes data parameters from raw network events to aggregated KPIs, then to standardized CEI and SQI indices. This parameter transformation reduces data complexity and volume while preserving the essential performance information needed for comprehensive multi-service monitoring