Research On Performance Optimization Of Oblivious Random Access Machine

Oblivious Random Access Machine(ORAM)is a cryptographic primitive that can eliminate information leakage in the trajectory when a user accesses memory,hiding the user's access mode in an untrusted environment.It protects the privacy of the data by making the observable communication between the client and the server computationally indistinguishable for any two logical access sequences of the same length.However,some of the more efficient ORAM solutions currently have O(log N)bandwidth amplification,or rely on expensive computational overhead to achieve the so-called O(1)bandwidth breakthrough,which makes it difficult to meet the actual deployment requirements in terms of performance.In order to solve the problem of random random access machine performance,this paper designs and implements a tree ORAM solution SRORAM(Shift and Root eviction ORAM)with better performance for small client storage.SRORAM consists of three modules: data reading,shift root expulsion,and early mixing.While protecting the access mode,SRORAM reduces the overhead compared with previous ORAM schemes.Through eviction algorithm that changes the location of data blocks by shifting on the server side,and then reads and writes back to the root bucket,it reduces the cost of deportation process,realizes the O(1)the bandwidth of the breakthrough and O(1)the client-side storage,and does not need to use complex homomorphic encryption.And it reduces the computational overhead and the communication overhead between the server and the client while ensuring security,and optimizes performance.In this paper,the performance of SRORAM is evaluated from the aspects of access delay and bandwidth amplification by using the common experimental evaluation method in the industry.The experimental results show that the end-to-end latency of SRORAM is 1.6 times to 2 times lower than that of the typical Ring ORAM scheme used for small client storage in the existing technology.Second,in terms of bandwidth amplification,SRORAM is 1.4 to 1.8 times better than the comparison scheme,and the server computation and storage consumption as the cost is almost negligible.