PhD defense Duy Nguyen: Decentralized Cryptography through Functional Encryption
Friday 20 Nov., 2026, at 10:00 (Paris time) at Télécom Paris
Télécom Paris, 19 place Marguerite Perey F-91120 Palaiseau [getting there], amphi 7 and in videoconferencing
Jury
- Shweta AGRAWAL, Professor, Indian Institute of Technology Madras, India (Reviewer)
- Geoffroy COUTEAU, Research Director, Institut de Recherche en Informatique Fondamentale (IRIF), Université Paris Cité, France (Reviewer)
- Dario FIORE, Professor, IMDEA Software Institute, Spain (Examiner)
- Thomas PETERS, Professor, Université catholique de Louvain, Belgium (Examiner)
- Melek ÖNEN, Professor, EURECOM, France (Examiner)
- Duong Hieu PHAN, Professor, Télécom Paris, Institut Polytechnique de Paris, France (Thesis supervisor)
- David POINTCHEVAL, Research Director, Centre national de la recherche scientifique (CNRS), École normale supérieure (Paris), France (Thesis co-supervisor)
Abstract
This thesis advances decentralized functional encryption through verifiable security, stronger and more modular theoretical foundations, and broader applicability to decentralized cryptographic protocols.
Functional Encryption (FE) enables controlled access to encrypted data through functional decryption keys, revealing only the result of an authorized computation. Multi-Client Functional Encryption (MCFE) extends this paradigm to joint computations over data from multiple clients, while Decentralized Multi-Client Functional Encryption (DMCFE) removes the need for a trusted authority by distributing key generation among the clients. Despite significant progress, important challenges remain in strengthening the security and theoretical foundations of decentralized functional encryption and extending its applicability beyond direct joint computation.
Learn more
First, we address verifiability against malicious clients. We develop a verifiable pseudorandom zero-sharing technique based on a new Combine-then-Descend approach using class groups of imaginary quadratic fields. This approach enables the construction of a range-verifiable DMCFE scheme for inner products from pairings through a cross-group construction combining class groups and bilinear groups. The resulting scheme allows malformed contributions to be publicly detected and malicious clients to be identified, while relying only on black-box access to the underlying group operations.
Second, we strengthen the theoretical foundations of Dynamic Decentralized Functional Encryption (DDFE). We introduce Updatable Pseudorandom Zero Sharing (UZS) in the standard model, which enables the removal of the Random Oracle Model from DDFE constructions for expressive functionalities such as function-hiding inner products and attribute-weighted sums. Furthermore, in the process of obtaining standard-model security, we establish a generic transformation preserving adaptive security when decentralizing function-hiding inner-product FE. These results yield an adaptively secure DDFE scheme for function-hiding inner products in the standard model, relying only on black-box access to the underlying FE scheme.
Finally, we extend the applicability of decentralized functional encryption to electronic voting. We introduce OMIX, an offline mixing paradigm that moves the trustees’ computation before the voting phase, enabling a mix-net-based self-tallying voting system whose tallying phase no longer requires trustee participation.