Efficient Hierarchical Identity-Based Encryption Scheme from Learning with Errors

Hierarchical identity-based encryption (HIBE) in fixed dimension has drawn wide attention because its lattice dimension keeps unchanged upon delegation, but there is a common defect of high complexity in trapdoor delegation stage of these schemes. Aiming at this problem, we propose two improved HIBE schemes under random oracle model and standard model respectively. We first use the MP12 trapdoor function to construct an optimized Z\-q-invertible matrix sample algorithm. Based on this optimized algorithm, combined with trapdoor delegation algorithm in fixed dimension and MP12 trapdoor function, we design system setup and trapdoor delegation stages. And we complete the HIBE scheme under random oracle model in conjunction with Dual-Regev algorithm. And then, we remove the random oracle by employing binary tree encryption system. The security of both proposed schemes strictly reduce to the hardness of learning with errors (LWE) problem, in which the scheme under random oracle model satisfies the adaptive security while the scheme under standard model satisfies selective security. Comparative analysis shows that, under the same security level, the overhead of trapdoor delegation in our scheme under random oracle model is reduced significantly compared with the relevant schemes, while the overhead of our scheme under standard model is reduced nearly 6 times compared with the relevant optimal schemes. Furthermore, the parameters such as lattice dimension, trapdoor size and ciphertext expansion rate etc., all decrease in some degree, and the computational cost is reduced obviously.