Hierarchical Network Event Aggregation for Performance Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network operation centers face difficulties in analyzing network performance data due to overlapping patterns and incomplete datasets, making it challenging to identify the root cause of network problems and predict future overloads, especially in real-time monitoring scenarios.
Innovation Solution
A computer-automated method and system that perform multi-level hierarchical analysis by aggregating network events into event groups and super-groups, allowing for visualization of these sequences in a way that each type is visually distinct, enabling domain experts to recognize patterns and timing sequences characteristic of specific network issues, even with incomplete data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex analytics are applied to network performance data to identify root causes and predict overloads, then the accuracy of problem detection improves, but the complexity of the analysis system increases
Solution Approach 1:
The patent segments the complex analysis system into multiple hierarchical levels: event-level analysis (individual network events), pattern-level analysis (sequences of events), and root-cause-level analysis (aggregated patterns). This segmentation allows the system to tackle complex problems step-by-step at each level rather than attempting to analyze all complexity simultaneously, thereby improving detection accuracy while managing system complexity through structured decomposition.
2Extent of automation
If fully automatic data-driven algorithms with machine learning are used to analyze usage data, then the automation extent increases, but the difficulty of detecting and measuring patterns increases
Solution Approach 1:
The patent introduces pattern templates as intermediary structures that bridge automatic machine learning algorithms and human expert analysis. These templates encode domain knowledge about typical network problem patterns, serving as a mediator that guides automated algorithms to focus on relevant patterns while remaining interpretable to human experts. This intermediary layer enhances automation while making pattern detection more manageable and less difficult by providing a structured framework for both machines and humans.
3Ease of operation
If hierarchical aggregation of events into groups and super-groups is performed, then the interpretability of network patterns improves, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-defining pattern templates and aggregation rules before actual network data analysis occurs. Event groups and super-groups are structured in advance based on domain knowledge of typical network problems. When real-time data arrives, the system matches events against these pre-established patterns rather than creating patterns from scratch, significantly reducing processing time while maintaining high interpretability through the use of pre-characterized pattern structures.
Data Source
AI summary
A computer system and computer-automated method for analyzing performance data in a telecommunications network. A data set is provided that contains a log of a first time sequence of network events which are classified into event types. A second time sequence is generated from the first time sequence by aggregating the events into event groups, and at least a third time sequence is generated by aggregating the event groups into event super-groups. A multi-level time sequence event hierarchy of at least three levels is thus created. The multiple time sequence levels are rendered into a visualization in which the different event types are visually distinct from each other. The visualization reveals to a domain expert patterns of behavior in the data set which can be used to detect current network problems and to predict future network loading, for example in a network operations center.