Community Research and Development Information Service - CORDIS

ERC

CryptoCloud Report Summary

Project ID: 339563
Funded under: FP7-IDEAS-ERC
Country: France

Mid-Term Report Summary - CRYPTOCLOUD (Cryptography for the Cloud)

The goal of the project is to provide fully-safe access to the Cloud for everybody.
This includes secure storage, private outsourced computations, and anonymous data sharing.

Fully Homomorphic Encryption is a magic, but still impractical, primitive for private outsourced computations in the Cloud. On the one hand, we addressed the privacy of the evaluated function.
On the other hand, we looked for interactive alternatives with compatible additive and multiplicative encryption schemes, and encryption switching protocols. We thus obtained a generic approach for secure two-party evaluation, quite well-suited for algebraic functions.

We extended the Smooth Projective Hash Functions (SPHFs) to cover more complex languages and with additional properties. As initially remarked, they could be used for interactive and non-interactive proofs. They more concretely found applications
- in Password-Authenticated Key Exchange (PAKE) protocols;
- as Quasi-Adaptative NIZK, for an efficient structure-preserving IND-CCA encryption scheme;
- as implicit zero-knowledge proofs, for efficiently preventing malicious behaviors in two-party protocols;
- as programmable proofs, for removing the need of erasures in adaptively-secure protocols (PAKE and Oblivious-Transfer) in the Universal Composability framework (i.e. the highest security level against the strongest adversaries).
More generally, we obtained various results on interactive proofs (with weaker computational assumptions) and on non-interactive proofs (with new security models), as well as compact signatures. Theses proofs are essential to prevent malicious behaviors and to get verifiable computations.

In order to query encrypted databases, we studied two approaches:
- symmetric searchable encryption, that allows to do keyword-search in an encrypted inverted index, without leaking too much information about the searched keywords. We focused on a dynamic index and verifiable answers;
- functional encryption, that allows to get aggregated informations in clear. But while general techniques exist in theory, no concrete constructions existed before. We proposed the first application to the inner product, that allows to compute weighted means.

Actually, functional encryption allows to obtain any function on the plaintext, which can be all-or-nothing according to an access policy. This has been defined under predicate and attribute-based encryption. We provided new constructions for theses primitives that allow to share some encrypted data in the Cloud to specific target users. This reduces the trust needed in the service provider.
In the same vein, we also studied various techniques without any PKI (since a PKI requires the trust in a central authority, we want to avoid), just using passwords.

Eventually, since privacy could be abused by adversaries, we designed new tools for anonymous credentials and a new methodology for anonymous (and traceable) e-cash. An authority (possibly distributed) can revoke anonymity in case of abuse.

Reported by

CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE
France
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top