基于柯西矩阵的最小带宽再生码研究

Study of Minimum Bandwidth Regeneration Codes Based on Cauchy Matrix

节点的失效在大规模分布式存储系统中是常见现象.为防止数据的丢失,系统必须解决失效节点的自修复问题.利用再生码可以在无需下载整个源文件的情况下即可恢复出失效节点的数据,从而能有效节省修复带宽.本文利用柯西矩阵作为编码矩阵,构造了一种精确修复最小带宽再生码(ER-MBR),可以精确修复失效节点,并通过实例演示了在有限域上进行编码解码及节点修复的过程.理论分析和仿真实验都表明利用柯西矩阵作为编码矩阵,其算法的运算效率优于利用范德蒙矩阵或者随机矩阵.

The failures of node are the common phenomena in the massive distributed storage system. To prevent the data loss, the system must solve the problem of self-repairing for failed nodes.Using regen- erating codes, the data of failed nodes can be recovered without downloading the whole source file, so repai- ring bandwidth can be effectively saved.This paper presents an exact-repair minimum bandwidth regenera- tion code （ER-MBR） by using Cauchy matrix as coding matrix. ER-MBR can exactly repair the failed nodes.The whole process of coding, decoding and node repairing is demonstrated in the finite fields by an instance.Theoretical analysis and simulation experiments prove that by using Cauchy matrix as coding ma- trix,the operation efficiency of this algorithm is better than that of using Vandermonde matrix or random matrix.