摘要
在NAND闪存存储系统的垃圾回收过程中,重新定义物理块的年龄和数据热度的计算方法,混合使用热块挑选策略和冷块挑选策略回收热数据块和冷数据块,并将回收块中的有效数据按照逻辑页的热度分为"热"数据和"冷"数据,分别写入到擦除次数最小的块和擦除次数最大的物理块中,减少在回收块的过程中多次对"冷"数据重复无意义拷贝.仿真实验结果表明,与当前主要几种磨损均衡算法相比,在保证回收效率的前提下,取得了很好的磨损均衡效果,同时,擦除次数和拷贝次数大幅度减少.
In the garbage collection process of NAND flash memory storage system,a wear leveling algorithm is presented,which is called GCb AH（ Garbage Collection based Age and Hot）. Three key ideas are realized in the algorithm. First,the age of physical block and the hot degree of data are redefined. Then,a hot block and a cold block selection method are combined to select the victim block.Finally,based on the hot degree of logic page,the valid data are divided into ＂hot＂data and＂cold＂data,which are respectively written back to the physical block with minimum or maximum erase cycle,thus meaningless copy operations of ＂cold＂data are reduced during the garbage process. Simulation results showthat the proposed algorithm performs better than various existing wear leveling schemes in terms of the number of erase operations,the copy operations and the degree of wear-leveling.
出处
《小型微型计算机系统》
CSCD
北大核心
2015年第11期2618-2621,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61172181)资助
关键词
NAND闪存存储系统
磨损均衡
物理块
逻辑页
NAND flash-memory storage system
wear-leveling
physical block
logic page
作者简介
石乐健，男，1990年生，硕士研究生，研究方向为模式识别与智能系统 E-mail： slj_scu@163.com；
严华，男，1971年生，博士，副教授，研究方向为模式识别与智能系统．
