| With the development of the Internet and the popularization of the broadband, the amount of data is growing explosively. Distributed data storage using extensible system structure, using more than one storage server for data storage, reduces the storage load of each storage server, and the storage information is located by location server. Distributed storage system not only improves the reliability of the system and its scalable system structure satisfies the requirement of mass data storage.Distributed data storage is a key technology in the data collection wireless sensor networks. But the available storage scheme based on data replication has high storage overhead and poor reliability, which is not suitable for distributed wireless sensor networks. The network coding is introduced to improve the data storage performance and achieve high reliability for the same redundancy compared to the data replication scheme. The distributed storage based on network coding can achieve high, data reliability with small storage overhead, but the introduction of network coding causes the problem of reparing the failure encoded data. In the distributed storage systems based on network coding, to maintain the reliability of the data, it is necessary to repair the failure encoded data. However, the traditional repair algorithm has great communication overhead, which makes the network resources waste. To reduce the repair communication overhead of the encoded data, this paper presents a two-layer distributed data storage scheme, in which the lower layer nodes store the encoded data blocks, and the upper layer nodes store the re-encoded blocks that are responsible for repairing failure data. Based on the two-layer data storage scheme, a data repair method is proposed to decrease the repair communication overhead with sacrificing lower storage overhead. Compared with MSR (Minimum Storage Regenerating Codes), basic interference alignment and group interference alignment schemes, the proposed method has low repair communication overhead. We prove the proposed method reduces the repair communication overhead to o(1/(?)k) times, and it is suitable to resource-constrained distributed wireless sensor networks. |
