Some Problems Of Secret Key Management

Key management plays a fundamental role in cryptography. In the symmetric cryp-tography, distribution of key requires tremendous communication and storage overheads. With the development of public key cryptography and the presentation of digital certifi-cates, the difficulties in key’s distribution and identity authentication had been solved. Such, the cryptography can provide the users with complete cryptographic services. Now, the cryptography has been applied to every aspects of our society and plays one of technical foundations for the rapid development of every fields, especially finance and e-commerce.However, there are still a lot of issues in key management need to be studied, mainly:(1) In some scenes, such as electronic vote and vehicular ad hoc networks, users’priva-cy is needed to be protected. However, the management of privacy enhancing certificates requires high cost of computation and communication, especially, the revocation of certifi-cates in anonymous setting has been the focus of recent research.(2) In identity-based cryptosystems, the key escrow problem enable a malicious PKG can sell the users’key, sign messages or decrypt ciphertexts on behalf of the users. Though lots of schemes have been proposed, no one can entirely prevent the malicious PKG from impersonating the users.(3) The public key infrastructure (PKI) is very successful in applications, however, in some special networks, such as mobile ad hoc networks and wireless sensor networks, the PKI is not practical because of expensive overhead. Thus, the key management schemes hitting the characters of these special networks are required.(4) When we introduce the key management scheme to other cryptographic systems as a module, how much risk is increased in the practical operating environment? Moreover, the key management schemes running in physical devices are also under the threat of side-channel attacks. Therefore, it is valuable to define a model to quantify the confidentiality information leakage. In this way, we can evaluate the real security of key management schemes in the physical execution environment.In this paper, we studied these problems mentioned above. The results of ours are stated in the following:1. Certificate revocation in anonymous setting. Revocation of anonymous credentials is an indispensable feature for privacy-enhancing cryptographic protocols, which are usually deployed in anonymous authentication protocols. In most revocation schemes, computation cost of proving and verifying increases linearly with respect to the num-ber of users. On the other hand, the revocation schemes with constant computation cost lack of flexibility when integrated with other existing systems. In this paper, we propose an efficient membership proof scheme and an extended framework of anony-mous certificate revocation. With the framework, we can easily apply the membership proof scheme for revocation in privacy-protecting authentication systems. We prove that the proof scheme is correctness, security and zero knowledge. Moreover, our scheme costs (O(1)) computation for membership proving and verifying.2. An identity-based signature scheme without key escrow. The key escrow problem seems to be inherent in identity-based signature. The scheme proposed in our paper is key escrow free and can prevent the malicious PKG from impersonating the users’ identity totally. As we known, few scheme could achieve this before. Our scheme doesn’t introduce high cost of pairing operation, it is particularly suitable for some networks with constrained computation resources.3. An identity-based key management scheme for mobile hoc networks. The scheme eliminates the burden of certificates management and can be high level tolerance to nodes compromise. It possesses complete security properties including existential unforgeability against adaptive chosen message attacks, forward security and proactive security. The scheme is also achieve high compatibility and can provide useful module for other security protocols since implementing over any cyclic group where the strong Diffie-Hellman problem is supposed to be hard.4. A method for quantifying confidentiality leakage. Here, confidentiality means the secret and privacy. When introducing the key management scheme to other cryp-tographic systems as a module, we need an efficient method to quantify the risk of confidentiality leakage facing the side-channel attacks. Based on information theory, an efficient approach named as confidential procedure model (CPM) is proposed. The advantages of our model are mainly:(1) Confidentiality loss is quantified in a relative manner;(2) The optimal attacks, which include all possible attacks, are taken into account when defining the concept of confidential degree. With the help of CPM, the confidentiality of perfectly-secure cryptographic systems, the information leakage in side-channel attacks and the anonymity of DC-net can be analyzed in a quantitative manner.