Implementation and Performance Measurement of the SUHDSA (Secure, Useful, and High-performance Data-Stream Anonymization)+ Algorithm
Ji‐Yeon Lee, Soon Seok Kim · 2024
This paper focuses on tables in microdata format, where each tuple contains specific individual information collected in real-time and stored at the tuple level. The collected information within each tuple, being personal, necessitates privacy protection. Therefore, we need to anonymize this information, which is the main theme of this paper. In this context, we have proposed the SUHDSA algorithm, an improvement over Sopaoglu and Abul's UBDSA algorithm, and have further enhanced its security with the proposed SUHDSA+ algorithm. This paper aims to validate the results of our previously proposed SUHDSA+ algorithm through empirical experiments. The experiments show that while the existing SUHDSA+ takes between 3.94 to 7.98 seconds longer in execution time compared to the SUHDSA algorithm, it performs about 45.6% faster than the original UBDSA. In terms of information loss, there was an increase of up to 18% compared to the original SUHDSA; however, when the k-value increased from 3 to 10, the loss rate was up to 19% less, indicating a significant impact of the increased k-value.