Synthetic Time Series Anonymization With Temporal Pattern Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anonymization methods for time series data, such as k-anonymization and L-diversity, fail to account for the unique temporal characteristics of time series data, leading to information loss and inefficiencies in preserving anonymity.
Innovation Solution
A method and system that generates synthetic time series 'avatars' by identifying nearest neighboring time series in the frequency domain, applying random weight coefficients, and modifying temporal characteristics to ensure effective anonymization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If k-anonymization or L-diversity methods are applied to time series data, then anonymity is improved, but information loss increases and temporal characteristics are not preserved
Solution Approach 1:
The patent transforms time series data from time domain to frequency domain using Fourier transform, then applies k-anonymization on frequency coefficients. This parameter transformation allows anonymity to be applied without directly modifying temporal characteristics in the time domain, resolving the contradiction between anonymity and temporal characteristic preservation.
Solution Approach 2:
The patent replaces traditional time-domain anonymization methods with a frequency-domain approach. By substituting the mechanical process of direct time-series manipulation with spectral analysis and frequency coefficient modification, the method achieves anonymity while preserving the underlying temporal patterns through inverse Fourier transform.
2Reliability
If data generalization is applied to satisfy k-anonymization queries, then anonymity is improved, but data integrity deteriorates
Solution Approach 1:
Instead of generalizing time series data in the time domain, the patent changes to frequency domain parameters and applies anonymization on Fourier coefficients. This allows maintaining the original time-domain data integrity while achieving anonymity through frequency domain manipulation, avoiding the information loss associated with time-domain generalization.
3Adaptability or versatility
If synthetic data generation is used to preserve anonymity, then data utility is improved, but computational complexity increases
Solution Approach 1:
The patent generates synthetic time series by copying and combining k nearest neighboring time series in the frequency domain. This copying approach creates anonymized synthetic data that preserves statistical properties and utility, while the use of k-NN and frequency domain operations keeps the computational complexity manageable compared to other synthetic data generation methods.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The method delivers synthetic time series as an anonymised version of time series (x1 to xE) and comprises an identification (Fb3, Fb4) of K nearest neighbours with a distance calculation law (Eq3, Eq4), and generation (Fb5) of a first version of synthetic time series (xiA(t)) corresponding to a time series (xi) by a combination (Eq6) of the K nearest neighbours. In accordance with the invention, the method comprises an additional anonymisation procedure (Fb6) targeting temporal characteristics of phase (Eq10), of number of measurements (n, ni, nj) and/or of measurement steps (PL, PLA), the procedure carrying out, on a first version of synthetic time series (xiA(t)), a modification of at least one temporal characteristic from at least one temporal characteristic of the same type of one of the K identified nearest neighbour time series, which is selected using a predetermined selection law (Eq9).