Research On Secure Multi-Party Computation Protocols For Private Set Intersection

With the rapid development of Internet technology and the wide application of cloud computing technology,various applications of big data have penetrated into every aspect of people’s life.These applications make people’s life more convenient,but at the same time,a large amount of valuable and sensitive private information is constantly mined,and people’s privacy is threatened.Secure Multi-Party Computation(SMPC)can solve the problem of how data owners who do not trust each other can collaborative computing securely in distributed scenarios.Using SMPC technology,people not only realize the joint computation of data,but also ensure the privacy of data.In many data application scenarios,private data held by different participants can usually be represented by sets.Therefore,Private Set Intersection(PSI)is an important SMPC technology.Using PSI technology,two or more parties can calculate the intersection of their private sets without giving away any other information.Most existing PSI protocols focus on the intersection itself,but in many practical application scenarios,the associated information of the intersection is equally important,such as the size of the intersection and various function values of the associated data of the intersection elements.In this paper,secure computation protocols are proposed to calculate the intersection statistical function without disclosing the intersection cardinality is proposed for two-party and multi-party scenarios respectively:(1)For the two-party scenario,we designed a set of secure computation protocols to calculate the data statistical functions related to the intersection of the two parties(including cardinality,sum,mean,square difference,maximum value and minimum),without disclosing any information about private data except the results.To realize these functions,we design the Arithmetic Shared Private Membership Test(ASPMT)protocol based on Arithmetic secret sharing.Based on ASPMT protocol,many kinds of statistics of intersection associated data can be calculated securely and efficiently.All basic computation is built on secret sharing and oblivious transfer.Due to the use of Online/ Offline technology for precomputation,all protocols have high computing efficiency in the online stage.In addition,we prove the security of the protocol under the semi-honest model.(2)For multi-party scenario,we designed a set of secure computation protocols for computing multi-party intersection and its associated data statistics functions by introducing a pair of external auxiliary servers.Through secret sharing,the protocols share the privacy data of multiple clients to two non-collusive servers,and delegate PSI computation to two servers,so as to transform the inefficient secure multi-party computation protocol into efficient secure twoparty computation protocol.Based on arithmetic secret sharing,we design a new construction of Zero Test(ASZT)protocol,so as to realize secure multi-party intersection related data statistical function calculation without disclosing any information.In our protocols,the clients only need to delegate their private data to the external auxiliary servers in secret share to go offline,so as to transform the complex multi-party computing task into efficient two-party computing task.Therefore,our protocols has high computation and communication efficiency.