Cellular Network Root Cause Analysis Using Abnormality Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current networks face challenges in identifying and remediating equipment malfunctions due to high false alarm rates and large data processing requirements, especially with vast amounts of user equipment data.
Innovation Solution
A systematic hierarchical approach using user equipment data clusters and spatial temporal graphs of abnormalities is employed to reduce data processing needs and identify failures with high confidence, utilizing filters, clustering algorithms, and graph databases like Neo4j to analyze network behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If event reporting and signatures are used to identify equipment malfunctions, then detection capability is improved, but false alarm rate increases
Solution Approach 1:
The patent segments the detection process into multiple hierarchical levels: first filtering events by severity thresholds, then clustering similar events, and finally analyzing clustered patterns. This multi-stage segmentation allows the system to maintain high detection capability while reducing false alarms by progressively refining what constitutes a genuine anomaly versus normal variation.
Solution Approach 2:
The patent introduces event clustering as an intermediary step between raw event detection and final anomaly identification. By grouping similar events together and analyzing their collective patterns rather than individual events, the system acts as a mediator that filters out false alarms while preserving true anomalies, thus resolving the contradiction between detection sensitivity and false alarm rate.
2Measurement precision
If huge amounts of data are collected and analyzed to identify equipment malfunctions, then identification accuracy is improved, but data processing resources increase
Solution Approach 1:
The patent extracts and analyzes only the most relevant features from the massive dataset by applying severity thresholds and filtering criteria. Instead of processing all collected data equally, the system extracts only those events that meet specific severity conditions, significantly reducing processing resources while maintaining identification accuracy for critical anomalies.
Solution Approach 2:
The patent segments the data processing workload into hierarchical stages: initial filtering by severity, intermediate clustering of similar events, and final analysis of clustered patterns. This segmentation allows the system to process massive datasets efficiently by dividing the computational burden across multiple passes, each handling a subset of the data with different processing requirements.
3Productivity
If traditional filtering methods are applied to user equipment data, then data processing load is reduced, but anomaly detection accuracy decreases
Solution Approach 1:
The patent applies preliminary filtering actions based on severity thresholds before conducting detailed anomaly analysis. By pre-filtering events that clearly meet severity criteria and grouping them into clusters, the system reduces the data processing load for subsequent analysis while ensuring that potentially significant anomalies are not lost in the filtering process.
Solution Approach 2:
The patent implements feedback mechanisms where clustering results inform subsequent filtering and analysis stages. The system uses information from clustered event patterns to adjust filtering parameters and focus analysis on the most promising anomaly candidates, thereby maintaining detection accuracy even as data processing load is reduced through hierarchical filtering.
Data Source
AI summary
Concepts and technologies are disclosed herein for using user equipment data clusters and spatial temporal graphs of abnormalities for root cause analysis. User equipment data can be obtained from a cellular network. A filter having a threshold can be applied to the user equipment data to obtain records. A determination is made whether the threshold is to be adaptively adjusted. If a determination is made that the threshold is not to be adjusted, the records can be added to a record set. The records in the subset of records can be correlated based on a key to obtain a filtered and correlated version of the record set, a spatial temporal graph of abnormalities associated with the cellular network can be generated based on the filtered and correlated version of the record set, and a root cause of a failure can be determined based on the spatial temporal graph of abnormalities.