Core Concepts
Enabling secure and verifiable computations in homomorphically encrypted data through innovative encoding techniques.
Abstract
I. Introduction
Homomorphic encryption allows operations on encrypted data without decryption.
Lattice-based schemes like BFV and BGV are widely used for privacy-preserving applications.
Lack of verification in existing schemes poses security risks in sensitive computations.
II. Problem Statement and Solution Overview
System & Threat Model: Considerations for HE-based computation scenarios.
Objectives: Privacy preservation and correctness assurance in computations.
III. Preliminaries
Introduction to key components like homomorphic encryption, encoders, and authenticators.
IV. Replication Encoding
Design of an error-detecting encoding scheme based on replication.
Construction of a homomorphic authenticator using the replication-based encoding.
V. Polynomial Encoding
Development of a compact encoding scheme based on polynomials.
Implementation of a homomorphic authenticator using the polynomial-based encoding.
VI. VERITAS
A. Implementation and Hardware
Introduction to VERITAS library facilitating secure computation verification.
B. Benchmarking BFV Operations
Comparison of operation timings between REP and PE schemes.
C. Experimental Case Studies
Ride-Hailing Services: Evaluation of VERITAS performance in ride-hailing location matching services.
Stats
Homomorphic encryption enables operations on ciphertexts directly without decryption.
Lattice-based schemes like BFV and BGV are commonly used for privacy-preserving applications.
Existing schemes lack verification capabilities, posing security risks in sensitive computations.