A Generalized Framework for Preserving Both Privacy and Utility in Data Outsourcing.

Area of Science:

• Cryptography and Data Security
• Applied Computer Science

Background:

• Property preserving encryption (PPE) enhances encrypted data utility in outsourcing settings.
• Existing PPE schemes are limited to specific data types and vulnerable to inference attacks.

Purpose of the Study:

• To generalize prefix preserving encryption for broader data types and enhance security against inference attacks.
• To introduce a multi-view outsourcing framework for secure and accurate data analysis.

Main Methods:

• Developed a novel prefix-aware encoding technique for generalized prefix preserving encryption.
• Proposed a multi-view outsourcing framework generating indistinguishable data views.
• Minimized computation and communication overheads based on privacy leakage bounds.

Main Results:

• The prefix-aware encoding is applicable to diverse data types including geo-locations, DNA sequences, and numerical data.
• The multi-view outsourcing framework achieves 100% analysis accuracy while preserving privacy.
• Empirical evaluations on real datasets demonstrate effective defense against common inference attacks.

Conclusions:

• The proposed framework successfully generalizes prefix preserving encryption, offering enhanced security and applicability.
• This approach balances data utility and privacy, making encrypted data analysis more robust in outsourcing environments.