A Tri-Pool Dynamic Wear-Leveling Algorithm for Large Scale Flash Memory Storage Systems

Abstract

Because of the extra ordinary development in electronics technology, flash memories arise as a suitable data storage option for embedded systems. Due to the hardware restrictions and efficiency issues, series of researches has been conducted and are in progress to provide better performance for these devises. One of the popular research areas on flash memories is the state-of-art wear-leveling technique, which is used to distribute erase operations throughout blocks of flash memories, in order to avoid wearing of some blocks earlier than others. In this paper, we have proposed a tri-pool dynamic wear leveling algorithm which uses the method of neutralizing hot and cold blocks. The proposed algorithm mainly focuses on keeping erase counts of blocks as near as possible to the average erase count of the whole blocks; this will insure that no single block will have erase count much larger or lesser than the average. We also analytically prove that the newly proposed technique outperforms previous ones.