Control Plane Node Anomaly Detection for Voice Access Failures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies are unable to accurately determine where within the radio access network or core network a voice call failure occurs and why it fails, leading to inefficiencies in resolving such failures.
Innovation Solution
A system utilizing machine learning models to analyze key performance indicators (KPIs) from control plane nodes, such as AMF, UPF, PCF, and SMF, to identify anomalous nodes causing voice call failures, enabling rapid detection and resolution of issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If machine learning models are used to analyze KPIs from control plane nodes, then the speed of detecting voice access failures is improved, but the device complexity increases
Solution Approach 1:
A centralized anomaly detection system acts as an intermediary between network nodes and operators. The system collects KPI data from multiple control plane nodes (AMF, UPF, PCF, SMF), processes it through machine learning models, and provides actionable insights about voice access failures without requiring complex distributed intelligence across the network.
Solution Approach 2:
The anomaly detection system is segmented into modular components: data collection modules for different network nodes, machine learning analysis modules, and notification modules. This segmentation allows the complex system to be built, deployed, and maintained through independent, manageable units that can be activated based on specific detection needs.
2Productivity
If machine learning models are used to analyze KPIs from control plane nodes, then the speed of resolving voice access failures is improved, but the device complexity increases
Solution Approach 1:
The system establishes a feedback loop where KPI data from network nodes continuously feeds into machine learning models, which generate predictions about voice access failures. These predictions trigger notifications to operators, creating a closed-loop system that enables rapid response and resolution through informed decision-making.
Solution Approach 2:
The anomaly detection system performs self-service by automatically collecting data from network nodes, processing it through trained machine learning models, and generating actionable alerts without requiring manual intervention. The system serves itself by maintaining its own operational awareness and autonomously identifying issues that need operator attention.
3Measurement precision
If network node data is collected and analyzed to identify anomalous nodes, then the precision of failure location is improved, but the loss of information increases
Solution Approach 1:
The system extracts only the critical KPI data needed for anomaly detection from the vast amount of network node data. By focusing on specific indicators (such as call establishment success rates, signal strength, and network congestion metrics), the system achieves precise failure location without processing unnecessary information, thereby minimizing data loss and processing overhead.
Solution Approach 2:
The anomaly detection system applies local quality analysis by examining KPIs specific to each network node's function and location. Different nodes are monitored with node-specific KPIs relevant to their role in voice access, allowing precise localized failure identification without requiring uniform analysis of all network data, thus reducing overall information processing requirements.
Data Source
AI summary
At a high level, the technology disclosed herein relates to network node anomaly detection using one or more network node anomaly detection machine learning models. In embodiments, Key Performance Indicator associated with voice call establishment (e.g., for Voice over New Radio, Evolved Packet System Fallback, etc.) may be received. Key Performance Indicator of particular network node data associated with the voice call establishment may be provided to the one or more network node anomaly detection machine learning models (e.g., a density function machine learning model) for anomaly detection. In embodiments, the particular network node data may correspond to control plane nodes, such as an Access and Mobility Management Function (AMF), User Plane Function (UPF), Policy Control Function (PCF), Session Management Function (SMF), etc. An indication of the control plane node identified based on time and location correlation via the anomaly detection may be provided.