Abstract
Database watermarking is one of the most effective methods to protect the copyright of databases. However, traditional database watermarking has a potential drawback: watermark embedding will change the distribution of data, which may affect the use and analysis of databases. Considering that most analyses are based on the statistical characteristics of the target database, keeping the consistency of the statistical characteristics is the key to ensuring analyzability. Since statistical characteristics analysis is performed in groups, compared with traditional relational databases, time series databases (TSDBs) have obvious time-grouping characteristics and are more valuable for analysis. Therefore, this paper proposes a robust watermarking algorithm for time series databases, effectively ensuring the consistency of statistical characteristics. Based on the time-group characteristics of TSDBs, we propose a three-step watermarking method, which is based on linear regression, error compensation, and watermark verification, named RCV. According to the properties of the linear regression model and error compensation, the proposed watermark method generates a series of data that have the same statistical characteristics. Then, the verification mechanism is performed to validate the generated data until it conveys the target watermark message. Compared with the existing methods, our method achieves superior robustness and preserves constant statistical properties better.
Introduction
In the modern digital economy, time series databases (TSDBs)—which track data points like stock prices or sensor readings over time—are invaluable assets. To protect their copyright, owners often employ digital watermarking to embed hidden signals. However, traditional watermarking methods often introduce "noise" that shifts the data's distribution. For analysts who rely on statistical integrity (e.g., calculating means or correlations), these shifts can lead to erroneous conclusions. The challenge lies in embedding a robust watermark that is "statistically invisible," ensuring the data remains useful for analysis post-watermarking.
The RCV Framework
The researchers developed the RCV framework (Regression, Compensation, Verification) to address this challenge by leveraging the inherent "time-grouping" nature of TSDBs.
The embedding process follows three primary steps:
- Regression: The algorithm identifies "confidential" data points for watermarking and "non-confidential" reference points. A linear regression model predicts confidential values based on these references.
- Error Compensation: This core innovation applies a specific "compensation parameter" to generated noise residuals. This mathematical adjustment ensures the watermarked data maintains the exact mean, variance, and covariance of the original set.
- Verification: Instead of direct modification, the system iteratively generates statistically-preserved data groups and checks them against a hash-based verification function. If a group conveys the desired watermark bit, it is accepted.
For extraction, the scheme employs Majority Voting. By embedding the same watermark bit across multiple data groups, the system can recover the signal even if significant portions of the database are tampered with or deleted.
Experimental Evaluation
The scheme was evaluated using the Circuit Load Data of Singapore, a public dataset containing 340,080 data points (2003–2022). Attributes included timestamps, electricity load (reference), and real-time electricity price (confidential).
Key Results
- Perfect Statistical Fidelity: Experimental results confirmed that the mean, variance, and correlation coefficients of the watermarked database were identical to the original data (e.g., Mean: 5137.4943; Variance: 894129.9816).
- Superior Robustness:
- Alteration Attacks: Even with 90% of data groups altered, RCV maintained an extraction accuracy of over 84%, outperforming existing methods like DEW and GAHSW.
- Deletion/Insertion: The scheme showed near-perfect resilience to insertion attacks and remained stable even when 50% of the database was deleted.
Conclusion and Future Work
RCV treats statistical preservation as a hard constraint rather than an afterthought. By decoupling data modification from bit-embedding through a "search and verify" approach, it ensures high resilience without compromising analytical utility.
Future research will focus on extending this model to high-dimensional data and multi-column relationships. Additionally, while RCV preserves group statistics, individual data points are modified; exploring "lossless" versions for scenarios requiring absolute point-to-point accuracy remains a priority.
Summary updated