Telemetry Anomaly Detection for Early Infrastructure Failure Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Organizations face challenges in detecting potential system infrastructure malfunctions before they occur, as the issues are not readily apparent, making it difficult to prevent disruptions and outages.
Innovation Solution
A method using machine learning techniques, implemented by a processor, which obtains telemetry data, compares it with historical data, detects anomalies, and determines potential system infrastructure problems, applying machine learning algorithms to analyze patterns and assign scores to predict future issues, and provides notifications or fixes accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional monitoring approaches are used to detect system infrastructure problems, then the detection process is simple and straightforward, but the ability to detect potential problems before they occur is insufficient
Solution Approach 1:
The system performs preliminary analysis by comparing current telemetry data with historical data patterns before problems actually occur. Machine learning models are trained on historical data to establish baseline patterns, enabling the system to detect deviations that indicate potential future failures before they manifest as actual system problems.
Solution Approach 2:
The patent replaces traditional rule-based or threshold-based monitoring mechanisms with machine learning-based anomaly detection. The ML models automatically learn complex patterns and relationships in telemetry data, substituting manual configuration and simple threshold comparisons with adaptive, data-driven detection that can identify subtle precursors to system failures.
2Measurement precision
If machine learning techniques are applied to analyze telemetry data patterns, then the precision of anomaly detection is improved, but the computational resources and processing time required increase
Solution Approach 1:
The system performs preliminary training of machine learning models using historical telemetry data during off-peak periods or when computational resources are more abundant. Once trained, the models are deployed for ongoing anomaly detection, reducing the computational burden during real-time operation while maintaining high detection precision through pre-learned patterns.
Solution Approach 2:
The system applies machine learning techniques selectively to the most critical or informative telemetry data points rather than processing all available data with equal computational intensity. This partial application of ML analysis focuses computational resources on high-value indicators while maintaining detection precision for the most significant anomalies.
3Reliability
If comprehensive telemetry data collection is performed across the entire system infrastructure, then the coverage and completeness of anomaly detection is improved, but the volume of data to be processed increases
Solution Approach 1:
The system extracts and focuses on the most relevant and informative telemetry data points from the comprehensive data set. Machine learning models identify and prioritize key indicators that are most predictive of system failures, extracting only the essential information needed for effective anomaly detection while discarding or de-prioritizing less relevant data.
Solution Approach 2:
The telemetry data collection and analysis process is segmented into multiple hierarchical levels. Rather than processing all data uniformly, the system divides monitoring into critical and non-critical components, applying machine learning techniques selectively to segmented portions of the data that provide the highest value for anomaly detection while maintaining comprehensive coverage.
Data Source
AI summary
A method and a system for detecting anomalies that indicate potential system infrastructure problems is provided. The method includes: obtaining a first set of telemetry data that relates to a system infrastructure events, logs, and metrics that occur during a predetermined time interval; retrieving, from a memory, historical data that relates to system infrastructure telemetry; comparing the obtained first set of telemetry data with the retrieved historical data; detecting an anomaly based on a result of the comparing; and determining a potential system infrastructure problem based on the detected anomaly. The method may include using a machine learning algorithm to analyze the obtained first set of telemetry data; identify a pattern in the data; assign scores to the events included in the data; and use the pattern and/or the scores to detect the anomaly using an artificial intelligence (AI) model that is trained using the retrieved historical data.