|
The collection, retention, and misuse of private information can pose threats to individuals, institutions, and collective values. Cryptography can be a useful tool to help address these concerns, enhancing our security and our ability to feel secure. Historically, however, this potential benefit has been largely unrealized, perhaps because most of the research in cryptography have gone to securing electronic commerce and to investigating foundational questions in the field. Cognizant of this, the PI's research aims to develop and popularize scientific work in cryptography that is more directly connected to securing people's communication. The work aims to be as scientifically well founded as the most rigorous theory-focused cryptography, but to be more directly oriented towards improving people's communication security and preserving for the individual some protected, private space.
The research will follow the traditions of practice-oriented provable security, which deals with definitions, practical protocols, and concretely analyzed proofs. The first aim is to treat constructs like onion routing and anonymous key-distribution within this framework. The goal is to have clear, quantitative, and self-contained definitions, protocols, and proofs for privacy problems that are well-known, particularly to practitioners, but that lack satisfactory foundations. Next, the research will investigate a new approach for secure messaging: communications-efficient messaging, in the star topology, with an untrusted server as the hub. Senders send their messages to a server, and receivers pluck their messages from it, but neither the server nor a global adversary will know who has sent what to whom. Finally, the research will study the privacy and authenticity of secret-sharing schemes, including deterministic secret-sharing of high-entropy secrets. This work will inform the development of a state-of-the-art secret-sharing tool for at-risk stakeholders. Some of the work will be facilitated by a new definitional paradigm, oracle silencing, in which a naive game-based cryptographic definition is automatically modified, using a notion's correctness condition, to withhold credit for trivial adversarial wins.
|
SaTC: CORE: Small: Crypto-for-Privacy