Radio Anomaly Detection via Cluster and Trend Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting radio communication anomalies in sensor networks, such as those using clusters generated from normal time-series data or threshold values, often fail to accurately identify anomalies due to inclusion of quality degradation values or limited waveform patterns, leading to incorrect detection.
Innovation Solution
A device capable of communicating with sensor nodes via radio waves, which gathers and stores parameters indicating radio communication quality, classifies parameter sets into clusters, and performs trend analysis to determine rapid and slow radio quality degradation, allowing for precise identification of anomalies without relying on normal time data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cluster-based anomaly detection is used with normal time-series data, then anomaly detection capability is provided, but detection accuracy deteriorates due to inclusion of quality degradation values in the cluster
Solution Approach 1:
The patent extracts and removes quality degradation values from the clustered normal time-series data before generating anomaly detection thresholds. By separating degraded data points from the normal operation cluster, the system ensures that thresholds are based purely on normal operations, thereby improving detection accuracy while maintaining anomaly detection capability.
2Ease of operation
If threshold-based anomaly detection is used, then simple detection is provided, but detection accuracy deteriorates due to limited waveform patterns
Solution Approach 1:
The patent transforms the detection approach by changing from fixed threshold values to dynamic thresholds generated through clustering analysis of historical data. This allows the system to adapt to varying operational conditions and waveform patterns, improving detection accuracy while maintaining relative operational simplicity through automated threshold generation.
3Speed
If rapid quality degradation is detected using cluster comparison, then rapid anomaly detection is provided, but slow degradation trends are missed
Solution Approach 1:
The patent segments the anomaly detection process into two distinct components: (1) cluster-based rapid detection for sudden quality degradations, and (2) trend analysis for gradual degradation patterns. This segmentation allows each method to specialize in detecting specific types of anomalies, improving overall detection reliability without sacrificing rapid response capability.
Solution Approach 2:
The system dynamically switches between or combines cluster comparison and trend analysis methods based on the detected pattern of degradation. For rapid changes, cluster comparison provides immediate detection, while for gradual changes, trend analysis captures the evolving pattern, making the overall system adaptable to different degradation scenarios.
Data Source
AI summary
A radio communication anomaly detecting method includes: gathering and storing different kinds of parameters indicating a radio quality with a sensor node in a storage unit; classifying parameter sets, each containing specific kinds of parameters gathered during a prescribed time period among the stored parameters, into clusters; estimating that a rapid radio quality degradation has occurred during the prescribed time period when there is a cluster of which all average values of specified kinds of parameters among the different kinds of parameters degrade more than those of another cluster; and performing a trend analysis for a time period during which it is not estimated that the rapid radio quality degradation has occurred to determine whether each of the different kinds of parameters gathered during the time period exhibits a trend of degradation, and estimating that a slow radio quality degradation has occurred based on a result of the trend analysis.