ENHANCING PRIVACY-PRESERVING DATA ANALYTICS THROUGH HOMOMORPHIC ENCRYPTION: TECHNIQUES AND APPLICATIONS

Abstract

In the era of data-driven decision-making, the protection of sensitive information has become a critical challenge, particularly as organizations increasingly rely on data analytics for insights. This study explores the potential of Homomorphic Encryption (HE) in enhancing privacy-preserving data analytics, a promising cryptographic technique that enables computation on encrypted data without exposing sensitive information. We evaluate the performance of three different HE schemes in terms of encryption and decryption times, computational overhead, analytical accuracy, and scalability. Our findings reveal that while HE incurs significant computational overhead, particularly for larger datasets, the accuracy of analytical results remains comparable to that of plaintext data, demonstrating its potential for privacy-preserving analytics. A performance decrease occurred when data set sizes grew which led to encryption along with decryption taking much longer than regular encryption approaches. The accuracy levels of data processing functions including categorization and regression did not change during periods when unencrypted data was utilized. The main challenge for HE on large datasets is scalability but we believe that encryption systems which combine HE methods with alternative techniques might offer a valid solution. HE demands perfectly integrated hardware acceleration technology combined with algorithm developments if it aims to achieve practical usability. Our work establishes fundamental elements for scalability and efficiency growth which will enhance confidential analytics uses through supporting heightened HE awareness.