Anomaly Location Identification via Contribution Degree Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection methods in communication systems face challenges in uniquely identifying anomalous devices and precision, as multiple devices can be affected, leading to failed alerts and prolonged anomaly location identification times.
Innovation Solution
An anomaly location identification device that determines anomaly presence using a detection algorithm, calculates contribution degrees for each information item, and applies a causal model analysis to identify the anomalous device, improving precision and calculation speed by focusing on high-contribution devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If anomaly detection is applied to each device independently, then anomaly detection speed is improved, but anomaly location identification precision deteriorates due to multiple affected devices
Solution Approach 1:
The system segments the anomaly analysis process into two distinct phases: (1) system-level anomaly detection that determines whether an anomaly exists in the entire system, and (2) device-level contribution degree calculation that identifies which specific device caused the anomaly. This segmentation allows fast system-wide detection while maintaining precise device-level identification through contribution analysis.
Solution Approach 2:
The patent introduces contribution degree as an intermediary metric that bridges system-level anomaly detection and device-level identification. The contribution degree quantifies each device's responsibility for the detected anomaly, enabling precise location identification without requiring independent detection on each device. This intermediary measure resolves the contradiction by providing a computational bridge between the two levels.
2Measurement precision
If system-wide anomaly detection is applied considering correlation among observation information items, then anomaly detection precision is improved, but anomaly location identification becomes difficult
Solution Approach 1:
The patent extracts the location identification function from the anomaly detection process itself. Instead of attempting to identify the anomalous device within the general anomaly detection framework, the system separately calculates contribution degrees for each device after detecting a system-wide anomaly. This extraction simplifies the location identification task by focusing computational effort on quantifying device contributions rather than searching for anomalies across all devices.
Solution Approach 2:
The patent applies local quality by calculating contribution degrees specifically for devices that are candidates for causing the anomaly, rather than uniformly analyzing all devices. The contribution degree calculation focuses on local device characteristics and their specific impact on the observed anomaly, enabling precise identification without the complexity of system-wide device analysis.
3Productivity
If threshold-based anomaly detection is used for each device, then calculation speed is improved, but false negatives increase when outlierness does not reach threshold
Solution Approach 1:
The patent merges individual device observation information items into a unified system-wide observation dataset. By combining data from multiple devices and analyzing correlations among observation items across the entire system, the system achieves reliable anomaly detection without requiring each individual device's outlierness to exceed a threshold. This merging approach maintains calculation efficiency while improving reliability through aggregate analysis.
Data Source
AI summary
An anomaly location identification device includes a determination unit configured to determine presence or absence of an anomaly by inputting part or all of information items output from a plurality of devices into an anomaly detection algorithm; a calculation unit configured to calculate, in response to a determination made by the determination unit that an anomaly is present, with respect to one of the information items, an index indicating a degree of contribution to the anomaly; and an identification unit configured to perform calculation by an analysis algorithm using a causal model receiving the index as input, to identify an anomalous device, to improve the precision and calculation speed related to identification of an anomaly location.