Cross-Imputability Sensor Clustering for Bandwidth-Constrained Telemetry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional system bus architectures in enterprise computing systems provide limited I/O bandwidth, constraining telemetry sampling rates and impacting prognostic cybersecurity techniques, which rely on high sampling rates for effective pattern recognition.
Innovation Solution
A system that uses cross-imputability analysis to cluster sensors and implement a Round-Robin Staggered-Imputation (RRSI) technique, selectively transmitting sensor values based on their predictability from other sensors, allowing for higher sampling rates within bandwidth constraints without hardware modifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of physical sensors and telemetry metrics is increased to improve monitoring coverage, then measurement precision and reliability are improved, but I/O bandwidth requirements increase beyond what conventional system bus architectures can provide
Solution Approach 1:
The patent extracts and transmits only the most critical sensor values to the management processor, leaving less important data to be imputed from available information. This selective extraction reduces the quantity of data transmitted over the I/O bus while maintaining monitoring precision for essential parameters.
Solution Approach 2:
The patent creates imputed copies of sensor values that are not directly transmitted. The management processor generates these copies based on available sensor data and imputation algorithms, reducing the need to transmit every actual sensor value while maintaining data completeness for monitoring purposes.
2Adaptability or versatility
If I/O bandwidth is reduced to accommodate legacy system bus architectures, then adaptability to existing systems is improved, but telemetry sampling rates decrease impacting prognostic performance
Solution Approach 1:
The patent implements periodic imputation where the management processor periodically generates imputed sensor values at high sampling rates. This periodic generation of data compensates for the reduced transmission frequency, maintaining effective sampling rates for prognostic algorithms while adhering to legacy bandwidth constraints.
Solution Approach 2:
The system performs preliminary imputation of sensor values before they are needed for prognostic analysis. By pre-generating imputed data based on available information, the system ensures high sampling rates are achieved without requiring high bandwidth for real-time transmission.
3Reliability
If all sensor values are transmitted at high sampling rates to improve prognostic cybersecurity techniques, then pattern recognition accuracy is improved, but I/O bandwidth requirements exceed system bus architecture capabilities
Solution Approach 1:
The patent applies different quality levels to different sensor data. Critical sensor values are transmitted with high fidelity and frequency, while less critical values are imputed with lower priority. This local differentiation in data quality maintains prognostic reliability for essential parameters while reducing overall data transmission volume.
Solution Approach 2:
The system discards transmission of certain sensor values that can be reliably imputed, and recovers them through imputation algorithms. This selective discarding of transmittable data in favor of imputed data reduces transmission volume while maintaining the completeness needed for reliable pattern recognition.
Data Source
AI summary
The disclosed embodiments relate to a system that reduces bandwidth requirements for transmitting telemetry data from sensors in a computer system. During operation, the system obtains a cross-imputability value for each sensor in a set of sensors that are monitoring the computer system, wherein a cross-imputability value for a sensor indicates how well a sensor value obtained from the sensor can be predicted based on sensor values obtained from other sensors in the set. Next, the system clusters sensors in the set of sensors into two or more groups based on the determined cross-imputability values. Then, while transmitting sensor values from the set of sensors, for a group of sensors having cross-imputability values exceeding a threshold, the system selectively transmits sensor values from some but not all of the sensors in the group to reduce a number of sensor values transmitted.