Virtual Machine State Detection with Adaptive Density Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional virtual machine anomaly detection methods rely on fixed threshold values and manual intervention, lacking flexibility and efficiency, and are prone to errors due to high-dimensional data complexity and imbalanced positive and negative samples.
Innovation Solution
Employing an unsupervised kernel density estimation algorithm to build probability density models for virtual machine monitoring data, using CPU utilization, network traffic, and correlation analysis to determine a health score through function distribution and abnormal value ratio measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If fixed threshold values are used for anomaly detection, then the detection method is simple to implement, but it lacks flexibility to change with time and business conditions
Solution Approach 1:
The patent transforms static fixed threshold values into dynamic adaptive thresholds by using kernel density estimation to model the probability distribution of monitoring data. The thresholds automatically adjust according to the learned distribution characteristics, enabling the system to adapt to changing business conditions and time patterns without manual intervention.
Solution Approach 2:
The patent changes the parameter representation from fixed threshold values to probability density functions. By estimating the kernel density of monitoring data and comparing actual values against this probabilistic model, the system achieves both automation and adaptability while maintaining computational feasibility.
2Measurement precision
If manual checking of each indicator is performed by professional personnel, then the detection accuracy may be improved, but the operation efficiency becomes low and unreliable
Solution Approach 1:
The patent enables the system to perform self-diagnosis and self-assessment of virtual machine health status. The kernel density estimation model automatically learns normal behavior patterns and identifies anomalies without requiring professional personnel to manually check each indicator, achieving both high accuracy and efficiency through automated intelligent detection.
Solution Approach 2:
The patent replaces the mechanical manual checking process with an automated computational system. Instead of professionals manually examining each indicator, the system uses kernel density estimation algorithms to automatically analyze monitoring data, compare it against learned distributions, and generate health assessments, thereby eliminating human labor while maintaining or improving detection quality.
3Ease of manufacture
If conventional anomaly detection methods are used, then the implementation is straightforward, but the detection accuracy is insufficient due to high-dimensional data complexity and imbalanced samples
Solution Approach 1:
The patent addresses high-dimensional data complexity by transforming the detection approach from fixed-threshold comparison in original feature space to probability density estimation in a transformed statistical space. This dimensional transformation allows the system to handle complex multivariate relationships and imbalanced samples more effectively while maintaining implementation feasibility through standardized kernel density estimation procedures.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a method of detecting a running state of a virtual machine based on a kernel density estimation, including: acquiring (S210) a training data set of at least one feature used to characterize at least one running state of the virtual machine; building (S220) a probability density model for the training data set by using a kernel density estimation algorithm, so as to obtain a probability density distribution curve; determining (S230) a probability density threshold value according to the probability density model; acquiring (S240) a test data set of the at least one feature; acquiring (S250) a probability density distribution curve according to the built probability density model; comparing (S260) the probability density distribution curve of the training data set with the probability density distribution curve of the test data set to determine a probability density distribution measurement total score; comparing (S270) a plurality of probability density values of the test data set with the probability density threshold value to determine an abnormal value ratio measurement total score; determining (S280) a virtual machine state total score according to the probability density distribution measurement total score and the abnormal value ratio measurement total score; and determining (S290) the running state of the virtual machine according to the virtual machine state total score.