Enhancing personal data de-identification using des-based transformation techniques

Abstract

Operational data sharing often requires stable record linkage while reducing identity disclosure risk. This paper outlines a compact two-layer de-identification design: DES-family deterministic pseudonymization (preferably Triple-DES) to transform direct identifiers into reversible tokens, and k-anonymity-guided generalization/suppression to limit linkage via quasi-identifiers. We show how DES-style permutations can be adapted to format-constrained fields through rank-then-encipher methods, and we summarize key security and utility trade-offs.