闪存损耗均衡的内存优化实现

Implementation of Memory Optimizing for Flash Memory Wear Leveling

广泛应用于嵌入式系统的闪存具有非易失性、低功耗和抗辐射等特点,但是在擦除后写入更新和擦写次数上限等方面具有局限性,经常需要通过闪存映射层进行存储管理,利用损耗均衡机制将擦写操作分布到闪存块,以延长闪存使用寿命。在已有闪存损耗均衡算法基础上,针对其占用较多内存的问题,研究并实现了闪存损耗均衡的内存优化。主要利用冷数据识别机制,结合动态损耗均衡和静态损耗均衡,在确保相近损耗均衡效果的同时,实现损耗均衡占用内存空间的优化。最后通过实验和分析,说明了损耗均衡占用内存空间的优化效果。

Because of flash memory’s advantages including non-volatile, shock-resistant and power-economic, it has been widely utilized in embedded system. However, considering flash memory’s characteristics about erase-before-write and erase-cycle-limit hardware, Flash Translation Layer (FTL) has to be employed. And in order to improve flash memory’s endurance, the wear-leveling is usually used to distribute erase operation evenly. Based on several wear-leveling algorithms for flash memory, in this paper it presented the implementation of limiting memory space requirement for a flash memory wear leveling. It is mainly implemented by using cold data identification mechanism, and combining static wear leveling with dynamic wear leveling. The experiment results show that this implementation can reduce the requirement for memory while achieving the efficient wear leveling effect for the whole storage space of flash memory.