Research On Network Coding Based Distributed Storage

Distributed storage is a technology that synchronously distributes files of data over computersand storage devices connected with internet. Comparing with centric scenario, distributed storageadopts scalable architecture and utilizes storage servers to share task, increasing availability,reliability and storage efficiency. In current computer networks, storage node’s failure and erroroften results in data loss. Replication based strategy is not applicable in massive distributed storagebecause of its high cost. In order to improve availability and reliability, network coding wasintroduced into distributed storage system. However, in network coding based storage, dataoperation suffers complexity and high computational cost. To address this problem, we proposeddata operation and confidentiality scheme of network coding based distributed storage as follows:1) Aiming at frequent data operation resulting from data dynamics, a high efficient dataoperation scheme, including modification, deletion and insertion, was proposed in this paper. Thisscheme only operates data blocks which are directly related data operation of original data objectwithout re-encoding the whole data object. By analysis, it shows that, comparing with traditionalmethod, our proposal reduces computational complexity in distributed data operation with harmingdata reliability.2) To address the requirement of confidentiality in distributed storage, we proposed a lightweight scheme which encoded data object and then encrypted encoded vectors. Analytical resultsprove that, comparing with traditional methods, the confidential scheme proposed in this paper thatcombined network coding and data encryption greatly reduces data size needed to encrypt andincreases performance efficiency while ensuring certain security levels.Both the data operation scheme and confidentiality scheme proposed in this paper can beapplied in cloud computing, internet of things and P2P networks.