Research On ORAM Technique To Protect Data Access Pattern

Cloud storage system is a technology developed rapidly, it is based on the cloud computation and combines the technology of networked and virtualization, acquiring high performance of computation and storage But the security problem imposes restrictions on the development of cloud computation. When the client stores data on the cloud storage, they lose the control power of the data. The simple method is to encrypt the data, which protects the privacy and confidential. The encrypted data is stored on the cloud storage system. However, there are some securities on the progress of searching for the encrypted data. For example, the data access pattern may leak the sensitive data.ORAM is a useful primitive that allows a client to hide its data access pattern and ORAM technique as a wide range of applications nowadays. The general idea is to use the storage of the third party, without revealing anything about the data being stored. It has been extensively used for software protection, secure Multi-party computation (SMPC), secure processors, and secure cloud storage system, especially for Searchable EncryptionORAM is first proposed to resolve the problem of the reversed engineer of software by Ostrovsky in 1996, in which Square root ORAM and Hierarchical ORAM is proposed.the performance of the two ORAM is inefficient and they are so complicated to be completed. With the high requirement for security, ORAM is developed rapidly. Afterwards, inspired by the work of Ostrovsky, Partition ORAM, Tree-ORAM, Path-ORAM and Ring ORAM are constructed. These ORAMs are efficient and easy to be completed. Moreover, ORAM based on homomorphic encryption scheme is proposed, it will be developed widely in the future. But it will appear the problem of inefficiency, so enhancing the performance of ORAM is the important aspect in the future.This paper provides an efficient Tree-based ORAM scheme that achieves perfect performance.We mainly make the following contributions. (1)Free of data overflow. We modify the method of reselecting the leaf node in the algorithm of read operation. Specifically, every selection we get an unoccupied lead node and fill it with new data. (2)No eviction operation. We put forward a new principle to write data back to the tree in the algorithm of add operation, and thus avoid complicated evict operation. (3)Less data access overhead. (4)Satisfy the requirement of security. We propose a modified way to reselect leaf node and write data back to the tree, which reduces the bucket capacity to a constant level, thus reducing server storage and overhead of data access. In his paper, we discuss the advantage and disadvantage of other ORAM schemes, and propose an efficient Tree-based ORAM scheme, making it widely used and more practical in the future.