Time-Series Sensor Data Formulization for Compact Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing volume of time-series sensor data from critical systems poses challenges in storage space management and raises concerns about inadvertently disclosing personally identifiable information (PII), making it difficult to efficiently store and analyze this data while ensuring privacy.
Innovation Solution
The system formulizes original time-series sensor signals into a set of equations using the telemetry parameter synthesis system (TPSS) technique, generating synthetic signals with the same correlation structure and stochastic properties, which are then stored instead of the original data, thereby reducing storage needs and eliminating PII concerns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If original time-series sensor data is stored in time-series databases, then machine-learning researchers can access the data for analysis, but storage space requirements become excessive and personally identifiable information may be disclosed
Solution Approach 1:
The patent creates synthetic time-series signals that copy the statistical properties, correlation structure, and stochastic characteristics of the original sensor data without storing the actual data. These synthetic signals are generated from compressed mathematical models (autoregressive moving average equations) that capture the essential patterns, enabling researchers to analyze data characteristics while reducing storage requirements by eliminating the need to store the original data itself.
Solution Approach 2:
The patent transforms the data from its original form into a mathematical model characterized by parameters such as autoregressive coefficients, moving average coefficients, and noise variance. This parameter transformation allows the data to be represented compactly through a small set of mathematical parameters rather than storing the entire time-series data, achieving significant storage reduction while preserving data analytical value.
2Loss of information
If original time-series sensor data is stored, then complete data information is available for analysis, but personally identifiable information may be inadvertently disclosed
Solution Approach 1:
The patent generates synthetic time-series signals that replicate the statistical and structural properties of the original data without containing any actual sensor readings or identifiable information. These synthetic copies preserve correlation structures, stochastic properties, and data patterns needed for analysis while completely eliminating personally identifiable information, thus enabling safe data sharing and research without privacy risks.
Solution Approach 2:
The patent extracts only the essential mathematical characteristics and statistical properties from the original sensor data, separating these from the actual data values that may contain PII. By taking out and storing only the mathematical model parameters (autoregressive coefficients, moving average coefficients) rather than the raw data, the system preserves analytical information while removing all potentially identifiable content.
3Productivity
If time-series sensor data is stored in databases, then data can be accessed for machine-learning evaluations, but the finite storage space becomes insufficient for large volumes of data
Solution Approach 1:
Instead of storing the original time-series data, the system stores compressed mathematical models that can generate synthetic data copies on demand. These models include autoregressive moving average equations with coefficients that capture the data's statistical properties, allowing unlimited synthetic data generation from a single compact model storage, effectively eliminating storage capacity constraints while maintaining full data accessibility for machine-learning evaluations.
Solution Approach 2:
The patent compresses the data representation from raw time-series values to a small set of mathematical parameters (autoregressive coefficients, moving average coefficients, noise variance). This parameter compression reduces the storage footprint dramatically while enabling complete data reconstruction through the mathematical model, thus expanding effective storage capacity without physical storage limitations.
Data Source
AI summary
The disclosed embodiments relate to a system that compactly stores time-series sensor signals. During operation, the system receives original time-series signals comprising sequences of observations obtained from sensors in a monitored system. Next, the system formulizes the original time-series sensor signals to produce a set of equations, which can be used to generate synthetic time-series signals having the same correlation structure and the same stochastic properties as the original time-series signals. Finally, the system stores the formulized time-series sensor signals in place of the original time-series sensor signals.