Time-Series Dataset Alignment and Deduplication for Network Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Monitoring computer networks faces challenges in accurately identifying anomalies due to misaligned and redundant time-series datasets from various sources, leading to overestimation of abnormal behavior and incorrect anomaly determination.
Innovation Solution
The solution involves aligning and deduplicating time-series datasets to identify unique metrics, using correlation thresholds to remove redundant data, and determining anomalies based on the proportion of non-deduplicated datasets exhibiting breaches, thereby improving the efficiency and accuracy of anomaly detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If time-series datasets from multiple sources are used for anomaly detection, then the coverage and comprehensiveness of monitoring is improved, but false positives increase due to redundant and misaligned data
Solution Approach 1:
The patent extracts and removes redundant time-series datasets from the collection before anomaly detection. By identifying and eliminating duplicate metrics and misaligned data through correlation analysis and temporal alignment, the system retains only unique, high-quality datasets for anomaly detection, thereby reducing false positives while maintaining monitoring coverage.
Solution Approach 2:
The patent changes the temporal parameters of time-series datasets by aligning them to a common time reference and resampling to consistent intervals. This parameter transformation enables accurate comparison and deduplication of datasets, resolving the issue of misaligned data causing false anomalies while preserving the multi-source monitoring advantage.
2Reliability
If multiple time-series datasets are analyzed simultaneously, then comprehensive system monitoring is achieved, but data processing complexity increases
Solution Approach 1:
The patent extracts and removes redundant datasets before analysis. By identifying duplicate metrics through correlation analysis and eliminating them, the system reduces the number of datasets requiring simultaneous processing while maintaining comprehensive monitoring coverage, thereby simplifying data processing complexity.
Solution Approach 2:
The patent performs preliminary temporal alignment and deduplication of time-series datasets before anomaly detection. By pre-processing the data to establish consistent time references and remove redundancies, the system reduces the complexity of subsequent analysis while preserving comprehensive monitoring capabilities.
3Productivity
If misaligned time-series datasets are used directly for anomaly detection, then data collection efficiency is maintained, but anomaly identification accuracy deteriorates
Solution Approach 1:
The patent changes the temporal parameters of misaligned datasets by aligning them to a common time reference and resampling to consistent intervals. This transformation maintains data collection efficiency while significantly improving anomaly identification accuracy, as the aligned data enables reliable comparison and pattern recognition across multiple sources.
Solution Approach 2:
The patent performs preliminary temporal alignment of datasets before anomaly detection. By establishing consistent time references and intervals in advance, the system enables accurate anomaly identification without sacrificing data collection efficiency, as the alignment process is automated and integrated into the data pipeline.
Data Source
AI summary
In some examples, time-series datasets received from a system may be temporally aligned. In some examples, one of the time-series datasets may be deduplicated. In some examples, whether an anomaly has occurred in the system may be determined based on a non-deduplicated time-series dataset of the time-series datasets.