Association Rule Mining for Mobile Network KPI Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of mobile networks generates vast amounts of performance data, which is difficult to analyze due to the large number of performance counters, leading to underutilization of available data and challenges in network performance monitoring.
Innovation Solution
The method involves converting performance counters into key performance indicators (KPIs), quantizing these KPIs into binary representations, and applying association rule mining and self-organizing map (SOM) techniques to analyze and visualize network performance data, enabling hierarchical association rule mining and data visualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the number of performance counters is increased to capture comprehensive network data, then the completeness of network performance monitoring is improved, but the complexity of data analysis increases significantly
Solution Approach 1:
The patent extracts and isolates the most critical performance indicators from the large set of performance counters by applying association rule mining algorithms. This extraction process identifies a minimal subset of KPIs that capture the essential network behavior, thereby maintaining monitoring completeness while reducing analysis complexity.
Solution Approach 2:
The patent introduces association rule mining and self-organizing maps as intermediary analytical layers between the raw performance counters and the final diagnosis. These intermediaries automatically process the complex counter data, transforming it into meaningful patterns and relationships, thus bridging the gap between comprehensive data collection and simplified analysis.
2Loss of information
If association rule mining is applied to all performance counters, then the comprehensiveness of relationship discovery is improved, but the computational time and resources increase
Solution Approach 1:
The patent segments the performance counters into hierarchical groups based on their functional relationships and importance. Association rule mining is then applied selectively to different segments at varying levels of detail, allowing comprehensive relationship discovery in critical areas while reducing computational effort in less critical areas.
Solution Approach 2:
The patent applies association rule mining to a strategically selected subset of performance counters that are most likely to contain valuable diagnostic information, rather than processing all counters equally. This partial action approach maintains relationship discovery effectiveness while significantly reducing computational overhead.
3Measurement precision
If multiple KPIs are used to monitor network performance, then the accuracy of performance assessment is improved, but the difficulty of data interpretation increases
Solution Approach 1:
The patent merges multiple related KPIs into unified diagnostic patterns through association rule mining. By combining correlated KPIs into single interpretive units based on their relationships, the system maintains the precision benefits of multiple measurements while simplifying interpretation through consolidated patterns.
Solution Approach 2:
The patent employs visual representation techniques that use color coding and graphical displays to represent complex KPI relationships and diagnostic patterns. This visual transformation converts difficult-to-interpret multi-KPI data into intuitive color-coded displays, making it easier for operators to quickly assess network performance and identify issues.
Data Source
AI summary
Embodiments are provided for using association rule mining to analyze performance counters of a mobile network, including converting a plurality of performance counters of a mobile network into a plurality of key performance indicators (KPIs), quantizing each KPI into one value of a set of values associated with that KPI, creating a set of items having multiple subsets corresponding to respective KPIs, where each item of a subset corresponds to a respective value of a particular set of values associated with a particular KPI, and generating association rules based, at least in part, on the set of items. In further embodiments, the quantizing the plurality of KPIs includes quantizing a first KPI into a first value of a set of three or more values associated with the first KPI.