| Along with the rapid development of computer network technology and the application, the meaning and the way of calculation have changed tremendously. People are facing distributed, increasingly powerful computing environment such as high performance computing, grid, cloud computing and so on. In this kind of environment, on the one hand user can make use of the powerful computing resource to complete their task through the network. On the other hand, all kinds of information is stored in the network, transmited on the network, and processed through the network. How to protect the sensitive and privacy information becomes an important problem. Secure multi-party computation (SMPC) is just causing the attention of people in this background.The connotation of secure multi-party computation is very extensive. We can study the security model of SMPC, or the realizablity and safe protocol of general mathematical function or (random) process. We can also study the specific secure multi-party computation protocols and their application. Most cryptography task can be regarded as a special case of SMPC such as coin flipping jointly, bit commitment, oblivious transmission, zero knowledge proof, and even secure communication, digital signature, key agreement, and so on. And it can be used broadly in all kinds of area, such as in the authentication protocol, auction protocol, voting protocol, privacy protecting data query or data mining.After more than30years of research and development, the basis theory of SMPC has been relatively mature. General protocols for arbitrary functionality have been designed in the information theory model and cryptography model along with the security proof. However there are serious problems in this kind of protocol such as the cost of computation time, space, communication, and interaction round. The protocols rarely can be used to solve practical problems directly for their high cost. The gap between the theory and application forced people to design practical protocols for specific secure multi-party computation problems. Most of the specific SMPC protocol is based on some of the more basic cryptographic protocols, this kind of relationship between the protocols is similar to the relationship between the program and sub-program. So we hope to design SMPC protocols through hierarchical and modular method. This kind of designing method can reduce the repetitive work of protocol designing, improve the efficiency and security of protocol implementation, and promote the standardization of security protocol.This paper mainly focused on the research of primary secure multi-party computation protocols and their application. In the primary protocol area, we have studied the non-commiting protocol, oblivious transfer protocol, and oblivious pseudorandom function computation protocol and their application in high level secure multiparty computation protocol. In the area of the application of SMPC, we have studied the biometric authentication applications. From point of view of the adversary model, we have mainly studied protocol design under the semi-honest adversary and adaptive adversary model.Adaptive adversary model is a kind of strong security model. Under the adaptive adversary model, the invaded party collection through the SMPC protocol executing process is not fixed, i.e., choosing when and which participants to invade is according to the information view of the adversary in the protocol executing process. Adaptive adversary model can be devided into Erasing type and Non-erasing type. In Erasing type, the parties will delete the internal state information partly or fully in the process of protocol execution. Whereas in Non-erasing type the parties will not delete any internal state information. Under Non-erasing type adaptive adversary model it is required that in security proof, it can’t distinguish the real view in protocol execution and the simulated view when the invasion is happening. This is more difficult than the situation under the corresponding static adversary model. Non-committing encryption is almost always needed when constructing protocol under adaptive adversary model. But this tool encrypt message bit by bit and the efficiency is very poor. So how to improve the efficiency of non-commiting encryption, or how to construct an protocol under adaptive adversary model without non-commiting encryption become a important problem.Oblivious transmission (OT) protocol is one of the fundamental cryptographic protocols. There are two parties in OT protocol:the sender and the receiver. The sender holds two (or more) secret message and the receiver hope to obtain a message with specific index. The security of OT protocol require the receiver can’t get the other messages while the sender can’t get the choice of the sender. After the proposal of OT it was used to construct general secure two-party computation protocol and secure multi-party computation protocol for arbitrary functionality. Because of the important station of OT protocol in general protocol construction, the study of its various security features and its efficiency has attracted many attention in the field of secure multi-party computation.We have obtained two results in the research of the cryptography scheme under the adaptive adversary model:1. A non-committing encryption schemes named LCC scheme is analyzed. We pointed out the error in the scheme and the security proof. We argued that although the scheme has good efficiency, but it does not meet the security requirement of non-committing encryption and can’t resist the attack of adaptive adversary.2. We put forward a UC secure OT1n framework that can resist adaptive adversary. We use a tool named multi-branch public key encryption scheme which transformed from dual mode public key encryption scheme. The novel protocol get higher efficiency on communication and computation than the existing OT1n protocol that can resist adaptive adversary.The pseudo-random function (pseudorandom function, PRF) is a kind of function with key that can be effectively computed. The values of the function can’t be distinguished with the random values in the range of the function. A oblivious pseudo-random function (OPRF) computation protocol is a protocol that can securely compute the PRF. There are server S and client C in the protocol. S holds the key K while C holds the input x. The parties compute the function fk(x) through interaction, and the client obtaion the function value ultimately. The computation process of OPRF is through a oblivious way,i.e, in the process of computation, the server can’t get any useful information of x while the client can’t get any useful information of k which is the security requirement of secure multi-party computation. In this article we obtain1results in the research of OPRF secure computing protocol:According to a pseudo-random function named Naor-Reingold, We have put forward a more efficient OPRF secure computation protocol by homomorphic encryption mechanism. Based on the new OPRF protocol, we designed the corresponding secure pattern matching protocol and secure set intersection protocol. We obtained higher efficiency in the OPRF protocol and security pattern matching protocol than previous protocols under the standard security model, and we get better usability in the secure set intersection protocol compare than the most efficient secure set intersection protocol.In the architecture of secure multi-party computation, complex protocols can be constructed based on the basic protocols. The protocols can be constructed directly using general protocol. However, more efficient and safe method is to utilize the basic protocol that have been proved to be secure as black box. In this paper we studied secure sorting protocol and bidirectional electronic auction protocol and obtain2results:1. Based on the secret sharing protocol and security protocol of matrix product we put forward efficient safe permutating protocol. And based on safe permutating protocol a novel efficient secure sorting protocol is proposed.2. We improved the protocol in BCD+09based on secret sharing and constant round secure comparing protocol. In the protocol we reduced the condition of honest majority in the old protocol. We improved the security of the scheme while maintain the high efficiency. The protocol can be applied in situation which require higher security and lower robustness.In the FIPS specification of advanced authentication, biometric authentication is the most senior authentication scheme and has the best security characteristics in many aspects relative to other authentication scheme. For example, in the password based authentication schemes password may be forgotten or leaked, while in hardware token based authentication schemes the token may be stolen or lost. However fingerprint and iris biometric information based authentication schemes have not the problem such as lost or forgotten, and leakage or theft of biological information is also relatively more difficult. And biological information is the user’s inherent attribute and it’s easy to carry. Thus biological authentication method bring greater security and convenience for user authentication. In this paper we studied the application of secure multi-party computation in biometric authentication and obtained1results:This paper proposed a new multi-factor based remote authentication protocol which improved multi factor authentication protocol in the biological features matching process and the interactive authentication process. The security of the new multi-factor based authentication protocol depended only on the long-term key of the authentication server and the collector. The protocol will be safe even the password and the smart card of the user is stolen. The protocol also conquered the problem of the previous protocol that using the hash function and causing failure in feature matching. This made the protocol obtain better security and usability. |
PDF Full Text Request