A Research On Techniques Of Private Information Retrieval

With the rapid development of big data and cloud computing,while data brings value,it also increases the risk of personal information leakage.With the disclosure of personal information,users may suffer from spam message harassment,and even cause serious economic losses.Protecting personal privacy is an essen-tial step when users get information from the server.Private information retrieval is a mean of protecting user information,in which users download bits mixed with some useless bits to achieve the purpose of confusing the server,so as to protect the sequence number of the bits that user really needs.While protecting the priva-cy of users,private information retrieval technology needs to consider the efficiency of the implementation scheme,that is,to obtain the demand messages of users with the least download cost.In the case of asymmetric interception of the server transmission channel,we derive the PIR capacity formula under arbitrary asymmetric interception mode and give the optimal retrieval scheme.The optimal retrieval scheme allocates the total downloads to the server according to the optimal solution of linear pro-gramming.The converse theorem is proved by the more extensive Han's inequality.Compared with symmet-ric eavesdropping model,the application of asymmetric eavesdropping model is more common in practice.Therefore,the conclusion of this study provides a design scheme and theoretical guidance for the retrieval of actual private information.To meet the demand of users with cache to retrieve multiple messages at the same time,we propose a MDS code-based multi-message private information retrieval method that makes full use of cache data,and deduce the upper bound and lower bound of retrieval capacity.The proposed method caches the single bit in the cache-free multi-message PIR scheme,and the proposed bound combines the inverse theorem method of cache-free multi-message PIR with the cached single-message PIR.The proposed upper and lower bounds are consistent in three cases:the number of retrieved messages is greater than half of the number of all messages,the number of all messages is integer multiple of the number of retrieved messages,and the cache capacity is small and the cache capacity is large.Therefore,in these three cases,we find the PIR capacity.When the cache capacity was₃¹,the number of servers and the number of requested files were 2,the downloads of cached multiple message PIR saved 58.33%/47.61%respectively compared with non-cached multiple message PIR and cached single message PIR.For servers with limited storage capacity without encoding and users with side information,we designed the scheme of private information retrieval with limited storage based on MDS code,and deduced the capacity formula for arbitrary number of servers and messages.Our results show that symmetric storage is optimal for servers with limited storage rate.The restricted storage model is more practical than a sufficiently large storage model.When both the number of messages and the number of databases are 3 and the storage rate is²3,the cached PIR model can save 25%in downloads compared to the non-cached PIR model.