An efficiency improvement in an anonymous credential system for CNF formulas on attributes with constant-size proofs

An anonymous credential system allows a user to convince a service provider anonymously that he/she owns certified attributes. Previously, we proposed a paring-based anonymous credential system with constant size of proofs, where the combinations of logical AND and OR relations on user attributes can be proved as CNF formulas. However, this system has a problem of requiring large online computation time during authentication, which depends on the number of AND relations in the proved formula. In this paper, we propose an efficiency improvement of the computational overhead based on online/offline precomputation technique. In our improvement, all exponentiations that can be used for the accumulator and witness computations are executed in advance in the precomputation algorithm. Thus, exponentiations in the online accumulator and witness computations are excluded, and only multiplications are needed. We implemented the system using a fast pairing library, and measured the processing times, while changing the size of the proved CNF formula. The experimental result shows that the computational costs of the proof generation in the case of using lots of AND relations are greatly reduced than the previous system. Hence, it is practical for mobile users.