Anomaly Detection Root Cause Analysis Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current anomaly detection methods in communication networks are insufficient for precise identification of causation without domain expert intervention, relying on high-level Key Performance Indicators (KPIs) that lack specificity and often result in false positives.
Innovation Solution
A method utilizing a network node that obtains KPIs, classifies multivariate data through a multiclass classification incorporated into an unsupervised self-learning neural network model, and provides anomaly classification with root cause analysis, thereby enhancing automated troubleshooting across multidimensional network data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If univariate anomaly detection is used with customized algorithms per counter, then detection accuracy for specific counters is improved, but device complexity and manual tuning effort increase significantly
Solution Approach 1:
The patent segments the anomaly detection process into distinct phases: data collection from multiple sources, feature extraction, anomaly detection, and root cause analysis. Each phase is handled by specialized components, allowing the system to achieve high detection accuracy without requiring manual tuning of a single complex algorithm across all counters.
Solution Approach 2:
The patent introduces intermediate processing layers including feature extraction modules and data normalization components that bridge raw counter data and the anomaly detection algorithm. These intermediaries automatically prepare and transform data, eliminating the need for manual algorithm customization per counter while maintaining detection precision.
2Ease of operation
If high-level KPIs are used for anomaly detection, then ease of operation is improved, but measurement precision and ability to identify root cause deteriorate
Solution Approach 1:
The patent adds a temporal dimension to KPI monitoring by analyzing trends and patterns over time rather than relying on single-point snapshots. It also introduces multiple dimensions of data collection (different counters, metrics, and time periods) that work together to provide both operational simplicity and precise root cause identification through multivariate analysis.
Solution Approach 2:
The patent implements a nested monitoring structure where high-level KPIs provide the overarching monitoring framework for ease of operation, while nested within this framework are detailed counter-level analyses and multiple data sources that enable precise root cause identification. The system automatically correlates anomalies across different nesting levels.
3Device complexity
If isolated monitoring metrics are used, then device complexity is reduced, but measurement precision and ability to identify true cause deteriorate
Solution Approach 1:
The patent merges multiple isolated monitoring metrics into a unified anomaly detection system that collects and correlates data from diverse sources including different counters, performance metrics, and network parameters. This consolidation maintains relatively simple device operation while achieving high causation identification accuracy through the integrated analysis of combined data streams.
Solution Approach 2:
The patent creates a universal monitoring framework that handles multiple types of metrics and data sources through a single standardized processing pipeline. This multi-functional system can detect anomalies across different network parameters using the same core algorithm, reducing device complexity while maintaining precision through the universal application of advanced detection methods.
Data Source
AI summary
Embodiments herein relate, in some examples, to a method performed by a network node for anomaly detection in a radio access network, RAN, in a communication network. The network node (11) obtains KPIs for predicting one or more characteristics of the RAN. The network node (11) further classifies multivariate data related to the obtained KPIs in a multiclass classification incorporated into an unsupervised self-learning neural network model; and provides anomaly classification with a root cause of the classified multivariate data from the unsupervised self-learning neural network model.