Generic Construction of Certificate-Based Encryption from Certificateless Encryption Revisited

Certificateless public key encryption (CLE) and certificate based encryption (CBE) are two novel public key cryptographic primitives requiring no authenticity verification of the recipient’s public key. Both of them are motivated to simultaneously solve the heavy certificate management problem inherent in the traditional public key encryption (PKE) and the key escrow problem inherent in the identity-based encryption (IBE). It is an attractive cryptographic task to formally explore the relation between CBE and CLE. In 2005, Al-Riyami and Paterson proposed one general conversion from CLE to CBE. Shortly later, Kang and Park pointed out a flaw in the security proof of AlRiyami-Paterson conversion. In 2012, Wu et al. proposed another generic conversion from CLE to CBE. Compared with Al-Riyami-Paterson conversion, Wu et al.’s method can be proved secure, but it has to additionally involve collision resistant hash functions. It remains an open problem whether the generic conversion due to Al-Riyami and Paterson, which is very neat, is provably secure. We aim to solve this open problem. First, we formalize CLE’s new security model, featured by introducing a new security property overlooked by previous security models. With this new security model as the basic technique, we succeed in proving that the Al-Riyami-Paterson generic Email addresses: sdgaowei@gmail.com (Wei Gao), guilin@uow.edu.au (Guilin Wang), kfchen@sjtu.edu.cn (Kefei Chen), wangxuyuyan@gmail.com (Xueli Wang ) This work is completed when the first author visited University of Wollongong (20112012) Preprint submitted to **** February 27, 2012 conversion from CLE to CBE is secure, if the CLE scheme is secure in our new security model. A concrete provably secure CBE scheme is presented to demonstrate the application of our result.