Hybrid KPI Node Profiling via ML Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods in wireless communication networks rely exclusively on Subject Matter Experts to identify and resolve issues using Key Performance Indicators (KPIs), which are difficult to analyze in real-time due to their complexity and dependence on multiple parameters, leading to potential missed PM counters and inadequate analysis.
Innovation Solution
A system and method that computes a hybrid KPI by combining Configuration Management (CM) parameters and Performance Management (PM) counters using machine learning or statistical techniques, enabling automatic profiling of nodes and real-time issue identification and resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods use Subject Matter Experts to analyze KPIs manually, then analysis accuracy may be maintained, but real-time issue identification cannot be achieved and productivity is reduced
Solution Approach 1:
The system enables automatic self-profiling of network nodes by computing hybrid KPIs through machine learning models that automatically correlate CM parameters and PM counters, eliminating the need for manual expert analysis while maintaining accurate issue identification
Solution Approach 2:
The patent replaces the mechanical manual analysis process with an automated computational system using machine learning algorithms to compute hybrid KPIs and identify performance issues in real-time
2Extent of automation
If hybrid KPI computation combines multiple CM parameters and PM counters using machine learning, then automatic node profiling and real-time monitoring are enabled, but device complexity increases
Solution Approach 1:
The patent introduces a hybrid KPI as an intermediary metric that bridges CM parameters and PM counters, using machine learning models as mediators to automatically compute and correlate multiple parameters into a unified performance indicator
Solution Approach 2:
The system transforms multiple static parameters (CM parameters and PM counters) into a dynamic hybrid KPI through machine learning computations, changing the parameter representation from individual metrics to a correlated composite indicator
3Reliability
If multiple CM parameters and PM counters are analyzed simultaneously by Subject Matter Experts, then comprehensive issue identification may be achieved, but it becomes impossible to accurately identify impact of multiple issues at the same time
Solution Approach 1:
The patent merges multiple CM parameters and PM counters into a unified hybrid KPI computation framework, where machine learning models simultaneously process all parameters together to identify their collective impact on node performance
Data Source
AI summary
Disclosed is a system for profiling one or more nodes based on a hybrid Key Performance Indicator (KPI). Initially, a flag indicating an issue with a KPI is received. A set of Configuration Management (CM) may be changed or identified by SME. Deviation in magnitude of each CM parameters from a predefined CM magnitude is computed to determine a changed CM parameter with deviation magnitude higher than deviation magnitude of remaining CM parameters. A set of Performance Management (PM) counters is identified by comparing magnitude of each PM with a predefined threshold value or using machine learning or statistical techniques. A hybrid KPI is created based on combination of the changed CM parameters and a subset of PM counters. One or more nodes are profiled by comparing the hybrid KPI associated to the node with hybrid KPI corresponding to each of the one or more nodes.