Year: 2023
Type: dissertation
Abstract: Privacy is a fundamental human right that ensures individuals have control over their personal information and are protected from unwarranted surveillance. While headlines such as the Snowden revelations and the Cambridge Analytica scandal showed the world the extent of government and corporate surveillance, the privacy situation today is still a significant concern. This is a multifaceted problem characterised by imbalance of power, inadequate regulation, lack of awareness, technological limitations and so on. In this thesis, we focus on the technological aspect with the hope that better technologies would motivate their use and improve individual privacy. The focus of this thesis is secure computation. It is a tool that allows two or more parties to jointly compute a function on their private inputs. This technology opens the door to many unconventional possibilities. For example, sequencing an individual's genomic data while ensuring the protection of their privacy and sensitive genetic information, or browsing a webpage without revealing to the service provider the specific page being accessed by the individual. This thesis evolves around two secure computation techniques—homomorphic encryption and garbled circuits. On the homomorphic encryption side, we built a private set intersection protocol and a privacy-preserving decision tree algorithm. Although we do not improve the asymptotic complexity compared to prior work, our novel way of using homomorphic encryption improved the concrete performance by a few orders of magnitude. On the garbled circuits side, our work on multi-party garbled circuits improved the asymptotic performance, rendering it much more practical than prior work in scenarios involving a large number of parties. We also optimised registration based encryption, a garbled circuit-based advanced encryption scheme, reducing the ciphertext size by more than a half.
Source: Lirias
Author: Cong, Kelong
Institution: KU Leuven
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7070515650
