逆码:一种可容3错的低密度MDS横式阵列码方法

发布时间：2018-06-12 06:48

  本文选题：磁盘阵列 + 横式阵列码　； 参考：《工程科学与技术》2017年05期

【摘要】：磁盘阵列存储可采用阵列码技术提高系统的容错能力。随着对阵列存储系统的深入认识,阵列码的更新效率也逐步成为一项重要的性能指标。针对当前可容3错横式阵列码更新效率低的问题,提出了一种具有低密度特性的横式阵列码构造方法,称为逆码。不同于传统阵列码是利用特殊几何方法确定编码过程,逆码是从生成矩阵角度出发,通过构造编码分布矩阵确定编码过程。首先,基于域GF(2~w)给出了一种具有超正规性质但只有3行元素的矩阵结构,称为逆结构矩阵;然后,利用w×w大小的比特方阵表示域GF(2~w)中元素;最后,通过提出的优化算法得到具有低密度性质的编码分布矩阵,进而确定逆码的编码过程。理论分析表明:逆码满足最大距离可分性质,可取得最优的存储效率;与STAR码、RTP码等容3错的阵列码相比,逆码的参数取值范围将不受素数的限制,参数设置更加连续。实验分析表明:相比于同样从生成矩阵确定编码过程的CRS码,逆码的稀疏度、更新效率以及编译码效率均有明显优势;相比于STAR码、RTP码,逆码的更新效率平均可提高20%;为了提高逆码的译码效率,文中也尝试了不同的异或序列技术。
[Abstract]:The disk array storage can use array code technology to improve the fault tolerance of the system. With the deep understanding of the array storage system, the update efficiency of array code has gradually become an important performance index. In view of the problem of low update efficiency of the current capacitive 3 error transverse array code, a transverse array code with low density characteristics is proposed. The method is called inverse code. Different from the traditional array code, the encoding process is determined by the special geometric method. The inverse code is based on the generation matrix, and the encoding process is determined by constructing the coded distribution matrix. First, a matrix structure with super normal property but only 3 elements is given based on the domain GF (2~w), which is called the inverse structure matrix; After that, the W x W bit matrix is used to represent the elements in the domain GF (2~w). Finally, the coding distribution matrix with low density properties is obtained by the proposed optimization algorithm, and then the encoding process of the inverse code is determined. The theoretical analysis shows that the inverse code satisfies the maximum distance can be divided and the optimal storage efficiency can be obtained; and the array of 3 errors, such as STAR code, RTP code, etc. Compared to the column code, the parameter range of the inverse code will not be restricted by the prime number and the parameter setting is more continuous. The experimental analysis shows that compared to the CRS code that also determines the coding process from the generation matrix, the sparsity of the inverse code, the update efficiency and the coding efficiency have obvious advantages, and the update efficiency of the RTP code, the RTP code and the inverse code can be increased by an average of 2 compared to the STAR code. 0%, in order to improve the decoding efficiency of the inverse codes, different XOR sequence techniques are also tried.
【作者单位】： 中国科学院成都计算机应用研究所;中国科学院大学;
【基金】：国家自然科学基金青年科学基金资助项目(61501064) 四川省科技厅支撑计划项目资助(2015GZ0088)
【分类号】：TP333
，

本文编号：2008774