Dynamic Threshold Learning for Network Element Alarm Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional network monitoring systems rely on hardcoded or default threshold values for Performance Monitoring (PM) and Key Performance Indicators (KPIs, which are ineffective as they fail to adapt to changing environmental conditions, leading to inaccurate alarms and inefficient operations.
Innovation Solution
A system and method for network elements to self-learn and dynamically adjust threshold values based on actual operating conditions, using statistical analysis to determine 'fence' values for parameters like temperature and CPU load, enabling proactive and corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If default or hardcoded threshold values are used for Performance Monitoring parameters, then deployment is simplified and initial setup is faster, but the thresholds become ineffective or too noisy as they cannot adapt to changing environmental conditions
Solution Approach 1:
The patent implements dynamic threshold adjustment by enabling network elements to automatically learn and adapt threshold values based on actual operating conditions. The system transitions from static hardcoded thresholds to dynamic thresholds that evolve with environmental changes, traffic patterns, and device states, thereby maintaining threshold accuracy without requiring manual redeployment
Solution Approach 2:
The patent enables network elements to self-configure threshold values through automated learning mechanisms. Each network element independently monitors its own performance parameters, analyzes operational patterns, and determines appropriate thresholds without external intervention, eliminating the need for manual threshold configuration while ensuring reliable, context-appropriate values
2Measurement precision
If manual configuration of thresholds for different KPIs is performed, then threshold accuracy can be improved, but it becomes very cumbersome for network operators and requires thorough investigations
Solution Approach 1:
The patent automates the threshold configuration process by enabling network elements to independently determine appropriate thresholds through automated analysis of performance data. This eliminates the need for operators to manually investigate and configure thresholds for each KPI, reducing operational complexity while maintaining or improving threshold accuracy through data-driven decisions
3Stability of the object's composition
If absolute configured threshold values are used, then they provide stable reference points for comparison, but the thresholds become outdated as operating conditions change over time
Solution Approach 1:
The patent implements dynamic threshold adaptation where thresholds automatically adjust to reflect changing operating conditions. The system continuously learns from performance data and environmental changes, enabling thresholds to evolve over time while maintaining stability through systematic learning processes, thus resolving the conflict between stability and adaptability
4Adaptability or versatility
If thresholds are set too low or too high to account for variations, then coverage is improved, but the feature becomes ineffective or generates excessive noise
Solution Approach 1:
The patent applies local quality by determining thresholds specific to each network element's operating context rather than using universal thresholds. Each element learns its own characteristic performance patterns and sets thresholds appropriate to its local conditions, ensuring neither excessive noise nor insufficient coverage by matching threshold sensitivity to local operational requirements
Data Source
AI summary
Systems and methods include obtaining measured data from a plurality of network elements in a network, wherein the measured data is associated with Operations, Administration, and Maintenance (OAM) functions for each of the plurality of network elements, analyzing the measured data to statistically determine thresholds for any of the OAM functions, and configuring the plurality of network elements with the determined thresholds, wherein the plurality of network element utilize the determined thresholds to compare ongoing measured data for threshold crossings for any of alarms and actions.