IACR News item: 12 December 2024
Elsie Mestl Fondevik, Kristian Gjøsteen
ePrint Report
In this paper we define the novel concept token-based key exchange (TBKE), which can be considered a cross between non-interactive key exchange (NIKE) and attribute-based encryption (ABE). TBKE is a scheme that allows users within an organization to generate shared keys for a subgroup of users through the use of personal tokens and secret key. The shared key generation is performed locally and no interaction between users or with a server is needed.
The personal tokens are derived from a set of universal tokens and a master secret key which are generated and stored on a trusted central server. Users are only required to interact with the server during setup or if new tokens are provided. To reduce key escrow issues the server can be erased after all users have received their secret keys. Alternatively, if the server is kept available TBKE can additionally provide token revocation, addition and update. We propose a very simple TBKE protocol using bilinear pairings. The protocol is secure against user coalitions based upon a novel hidden matrix problem. The problems requires an adversary to compute where the adversary must compute a matrix product in the exponent, where some components are given in the clear and others are hidden as unknown exponents. We argue that the hidden matrix problem is as hard as dLog in the bilinear group model.
The personal tokens are derived from a set of universal tokens and a master secret key which are generated and stored on a trusted central server. Users are only required to interact with the server during setup or if new tokens are provided. To reduce key escrow issues the server can be erased after all users have received their secret keys. Alternatively, if the server is kept available TBKE can additionally provide token revocation, addition and update. We propose a very simple TBKE protocol using bilinear pairings. The protocol is secure against user coalitions based upon a novel hidden matrix problem. The problems requires an adversary to compute where the adversary must compute a matrix product in the exponent, where some components are given in the clear and others are hidden as unknown exponents. We argue that the hidden matrix problem is as hard as dLog in the bilinear group model.
Additional news items may be found on the IACR news page.