Multi-Algorithm Anomaly Detection With Reliability Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in communication operation and maintenance is the difficulty in manually monitoring and maintaining the large number of KPIs in communication networks, exacerbated by the introduction of 5G technology, and the lack of reliable anomaly detection models due to high annotation costs and the need for professional knowledge, making it difficult to rely on unsupervised or supervised learning with limited samples.
Innovation Solution
An anomaly detection method utilizing multiple anomaly detection algorithms, followed by reliability verification and user input for confirmation or cancellation, to improve the accuracy of anomaly detection results, and a system that integrates these methods into a computer-readable storage medium for implementation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple anomaly detection algorithms are used to improve detection accuracy, then the reliability of anomaly detection is improved, but the device complexity increases
Solution Approach 1:
The patent divides the anomaly detection task into multiple independent algorithm modules (e.g., statistical detection, machine learning-based detection, rule-based detection). Each algorithm processes the KPI data independently and produces separate detection results. This segmentation allows the system to leverage the strengths of different algorithms while maintaining modularity and manageability, thereby improving overall detection reliability without creating an unmanageable monolithic system.
Solution Approach 2:
The patent merges the results from multiple anomaly detection algorithms through a result integration module. The integration mechanism combines detection outcomes from different algorithms (such as voting, weighted aggregation, or consensus-based approaches) to produce a final anomaly detection result. This merging process synergistically combines the capabilities of individual algorithms, achieving higher reliability than any single algorithm could provide alone.
2Measurement precision
If supervised learning algorithms are trained with annotated samples to improve detection accuracy, then the measurement precision is improved, but the loss of time increases due to high annotation costs
Solution Approach 1:
The patent applies partial annotation strategy where only a subset of KPI data requires manual annotation rather than annotating all available data. The system uses this partially annotated data to train supervised learning models, while the remaining data is processed using unsupervised methods or transferred from similar scenarios. This partial action approach achieves acceptable detection precision while dramatically reducing the time and resource investment required for annotation.
Solution Approach 2:
The patent implements preliminary annotation of representative or typical KPI scenarios that capture the most important anomaly patterns. By annotating key representative cases in advance, the system establishes a foundation of labeled data that can be used to train initial models. This preliminary action provides a head start for supervised learning while avoiding the need to annotate all possible scenarios, thereby reducing overall annotation time while maintaining detection precision.
3Measurement precision
If the number of KPIs to be monitored is increased to cover more network aspects, then the measurement precision is improved, but the ease of operation deteriorates due to manual monitoring difficulty
Solution Approach 1:
The patent implements automated anomaly detection systems that self-monitor multiple KPIs without requiring manual intervention. The system automatically collects data from numerous KPIs, processes them through detection algorithms, and generates anomaly alerts autonomously. This self-service capability enables the system to handle a large number of KPIs effectively, maintaining high measurement precision across comprehensive network monitoring while eliminating the operational burden that would make manual monitoring of such extensive metrics impractical.
Data Source
Figure 1~2
AI summary
Disclosed in embodiments of the present disclosure are an anomaly detection method and apparatus. The method includes: detecting first feature data of collected first data by using each anomaly detection algorithm, respectively, where N is an integer greater than or equal to 1; in response to first detection results corresponding to one or more anomaly detection algorithms being abnormal, performing a reliability verification on the first feature data of the first data to obtain first verification results; and determining second detection results according to the first verification results and to whether a reliability of the one or more anomaly detection algorithms to which the first detection results being abnormal correspond is greater than or equal to a first preset threshold.