基于分数阶Fourier变换的云存储系统重复数据删除算法

发布时间：2018-07-02 15:48

本文选题：分数阶Fourier变换 + 云存储　；参考：《计算机科学》2015年07期

【摘要】：云存储系统的重复数据作为大量冗余数据的一种,对其有效及时地删除能保证云存储系统的稳定与运行。由于云存储系统中的干扰数据较多,信噪比较低,传统的重删算法会在分数阶Fourier域出现伪峰峰值,不能有效地对重复数据进行检测滤波和删除处理,因此提出一种改进的基于分数阶Fourier变换累积量检测的云存储系统重复数据删除算法。首先分析云存储系统重复数据删除机制体系架构,定义数据存储点的适应度函数,得到云存储节点的系统子集随机概率分布;采用经验约束函数对存储节点中的校验数据块分存,通过分数阶Fourier变换对云存储系统中的幅度调制分量进行残差信号滤波预处理。采用4阶累积量切片后置算子,把每个文件分为若干个块,针对每个文件块进行重删,进行重复数据检测后置滤波处理,实现存储资源上的重复数据检测及其删除。仿真实验表明,该算法能提高集群云存储系统计算资源的利用率,重复数据准确删除率较高,有效避免了数据信息流的干扰特征造成的误删和漏删,性能优越。
[Abstract]:The repeated data of cloud storage system is one of the redundant data which can be deleted effectively and timely to ensure the stability and operation of cloud storage system. Because there are more interference data and lower signal-to-noise ratio in cloud storage system, the traditional redelete algorithm will appear pseudo peak value in fractional Fourier domain, which can not effectively detect, filter and delete repetitive data. Therefore, an improved algorithm based on fractional Fourier transform cumulant detection for repeated data deletion in cloud storage systems is proposed. Firstly, the structure of repeated data deletion mechanism in cloud storage system is analyzed, the fitness function of data storage point is defined, and the random probability distribution of system subset of cloud storage node is obtained. The empirical constraint function is used to separate the check data block in the storage node and the amplitude modulation component in the cloud storage system is preprocessed by fractional Fourier transform. The fourth order cumulant slice post-operator is used to divide each file into several blocks. The repeated data detection and postfilter processing are carried out for each file block to realize the repeated data detection and deletion on the storage resources. The simulation results show that the algorithm can improve the utilization of computing resources in cluster cloud storage system, and the accurate deletion rate of repeated data is higher. The algorithm can effectively avoid the error deletion and omission caused by the interference characteristics of data information flow, and the performance of the algorithm is superior.
【作者单位】：广西科技大学计算机科学与通信工程学院;武汉理工大学信息工程学院;
【基金】：广西自然科学基金青年基金项目(2013GXNSFBA019268) 广西科技大学自然科学基金项目(校科自1261126) 广西特色专业建设项目(GXTSZY217) 广西教育厅一般项目(YB2014208) 广西教育厅立项项目(LX2014182)资助
【分类号】：TP333

【参考文献】