Research On ZigZag Decoding In Fault Node Repair And Multi-file Private Information Retrieval

With the rapid development of Internet technology,which has made data storage and the query of information integrated into people's daily lives.At the same time,It is important timely repairing fault data in storage systems and protecting users' privacy in the process of file retrieval.While most of the above operations are performed in a high-advanced field at present,and the computational complexity is high.Binary ZigZag Decoding(BZD)technology based on binary domain only needs simple XOR and backhaul operations during decoding process,they have the advantages on short decoding time and low computational complexity.Therefore,the research based on Binary ZigZag Decoding technology in fault node repairing and multi-file private information retrieval is of great significance.1)Research on repairing a fault node.By studying a single CP-BZD code node stores multiple data blocks,once the node fails,all the data blocks in the k surviving nodes need to be downloaded to be successfully repaired,For the problem that over-repairing the bandwidth,this is improves its coding structure.The basic idea is that in the parity node,one original data blocks is added to the locally check data block to avoid downloading data unrelated to the faulty system node during the repair process.This solution reduces the repair bandwidth to twice the data storage capacity of the node;Based on the coding framework of Piggybacking-MDS(Piggybacking-Maximum distance separable)code,this is creatively combined the BZD technology with the Piggybacking-MDS coding framework,proposed a new coding scheme,and designed a universal coding shift matrix.Meanwhile,we also propose a detailed mathematical proof of the combination properties of the code.Performance analysis shows that the scheme storage overhead is always smaller than other CP-BZD codes in the same situation.2)Multi-file private information retrieval.In the storage system based on the replicate database,the encoding and decoding operations of the multi-file private information retrieval are performed in a large finite field,which leads to large calculation cost.Aiming at this problem,this paper applies the ZigZag Decoding in the process of query downloading,reduces the computational difficulty in the user retrieval process while ensuring the performance of the system's private information retrieval capability.In addition,when the multi-file private information retrieval is performed based on(7)n,k(8)CP-BZD storage system,it is found that servers need to repeatedly transmit multiple sets of redundant encoded files to the user.In order to solve this drawback,This article redesigned the search query scheme,first of all,the parity nodes of distributed storage are divided into groups,and next sending the random vectors to the nodes in each groups with shift coding operation.Which can effectively reduce the communication cost between the client and server.