Multi-Sensor Data Reduction Using Cross-Sensor Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data management systems face challenges in efficiently reducing the amount of data to be stored from multiple sensors, particularly when the data from one sensor is accurately predictable based on data from another sensor, leading to unnecessary data storage and potential storage capacity issues.
Innovation Solution
A data management system that includes a data-amount reducing unit which deletes part of the measurement data from a target sensor based on prediction data generated using measurement data from another sensor, selecting the target sensor based on differences between actual and predicted data, and reducing data storage by either decimating samples over time or reducing data size, ensuring efficient data management and storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If measurement data from all sensors is stored completely, then data completeness is improved, but storage capacity is consumed excessively
Solution Approach 1:
The patent extracts and removes redundant measurement data that can be predicted from other sensor data. The data-amount reducing unit identifies and deletes predictable data points while preserving essential information, thereby reducing storage requirements without significantly compromising data completeness.
Solution Approach 2:
The patent changes the parameter of data representation by using prediction models to estimate sensor values. Instead of storing actual raw data from all sensors, the system stores prediction data or markers indicating predictable periods, transforming the data from complete raw measurements to compressed predictive representations.
2Quantity of substance
If measurement data is reduced by deleting predictable data, then storage efficiency is improved, but data accuracy may deteriorate
Solution Approach 1:
The patent creates prediction data as a copy or approximation of the original sensor data using prediction models. This prediction data serves as a substitute for the actual measured data, maintaining sufficient accuracy for most applications while significantly reducing storage requirements. The system can restore or retrieve original data when high precision is needed.
3Adaptability or versatility
If data from multiple sensors is stored, then measurement coverage is improved, but data processing complexity increases
Solution Approach 1:
The patent merges the functions of multiple sensors by using prediction models that combine data from available sensors to estimate values from absent or redundant sensors. This merging approach maintains comprehensive measurement coverage while reducing the need to process and store data from every individual sensor separately.
Data Source
AI summary
A data management system is provided, which includes a data acquiring unit that acquires measurement data generated by measuring a measurement target from each of a plurality of sensors; a data storage unit that stores the acquired measurement data; and a data-amount reducing unit that deletes part of the measurement data acquired from a target sensor based on the measurement data acquired from another sensor among the plurality of sensors to reduce an amount of data to be stored.