03 Mar
15:00 Doctoral defense Fully distance
Cryptographic Engineering of Privacy-Preserving Algorithms
Pedro Geraldo Morelli Rodrigues Alves
Advisor / Teacher
Supervisor: Diego de Freitas Aranha / Co-supervisor: Edson Borin
Brief summary
Data collected from users is the gold of the modern age, and the ability to collect it is just as crucial as the ability to securely store and manipulate it. This is a thesis in the format of a collection, composed of articles published or in the process of being reviewed, that explore different aspects of computing that preserves privacy, such as the efficient implementation of primitives, protocols and applications. Our work offers a framework for an always encrypted database, which can store cryptograms and respond to encrypted queries without the need for decryption. In the same direction, we also study the case of large-scale data collection from smart meters. In this context, an entity such as the electricity supplier collects user data that can be used in statistical methods such as machine learning, and performs multipart computing over a network without revealing user information to nodes. On the other hand, we also present works that explore the efficient implementation of the arithmetic used by modern completely homomorphic cryptography schemes, such as BFV and CKKS. We experiment with different methods targeting the CUDA architecture and show how cryptosystems can be accelerated through proper choice of data structure, locale and algorithm used in polynomial multiplication. Four works are presented dealing with these topics, as well as a discussion that connects the work.
Examination Board
Diego de Freitas Aranha IC / UNICAMP
Guido Costa Souza de Araújo IC / UNICAMP
Julio Cesar Lopez Hernandez IC / UNICAMP
Hilder Vitor Lima Pereira KU Leuven/Belgium
Fábio Borges de Oliveira LNCC
Ricardo Dahab IC / UNICAMP
Marcos Antonio Simplicio Junior POLI / USP
Rafael Misoczki Google