Font Size: a A A

A Research On Key Technology For Sub-division Image Data Storage Based On Flash Memory

Posted on:2013-10-29Degree:MasterType:Thesis
Country:ChinaCandidate:Q LiuFull Text:PDF
GTID:2298330422973978Subject:Electronic Science and Technology
Abstract/Summary:PDF Full Text Request
Sub-division image data storage has been a key issue in the building of geographicinformation space system. The performance of mechanical disk has reach its bottleneck,which is restraining the rapid development of the sub-division image data storagesystem. With the superiority on large capacity, high speed, low power consumption andshock resistance, the solid-state hard disk based on flash memory is becoming a risingstar in the field of storage who is replacing the mechanical drive gradually. How toconduct a research on the key technology of sub-division image data storage based onthe flash memory to improve the entire system performance is a valuable issue.LazyFTL is a address mapping scheme which is based on Multi-Channel ParallelStorage Architecture. However, when used on sub-division image data storage, itsstorage and access speed is limited, which is the result of ignorance of parallel operationbetween channels. In this paper, the storage characteristics such as large capacity andhigh real-time requirement of random access is analyzed, the sub-division image datastorage need such as high storage speed is cleared, and a Hilbert-curve-based blockingaddressing mapping algorithm is proposed. In this algorithm, a lookup table approach isused to improve the calculating efficiency,a storage mode which separates the realimage data and metadata and storage the index node into a single page is adopted,resulting in the reduce of flash pages scanning numbers when the storage equipment isstarted, and thus, decreasing the initiating time of disk. In addition, thewindow-query-based local image cache replacement algorithm is proposed after deepanalysis of the local accessing property and the spatial corresponding feature ofsub-division image data. In the algorithm, general disk cache replacement scheme basedon the LRU accessing frequency statistics is changed, while a cache replacementscheme based on the spatial corresponding property is adopted, which highly improvesthe data hit rate and data access speed.As experiment shows, the address mapping algorithm based on Hilbert curve canreach1.6GB/s on the read and write speed, which outperform LazyFTL. The cachereplacement algorithm based on the window query can get32.7%on the cache hit rate,which is obvious superior than those approach based on LRU and significantlyimproves the data access speed.
Keywords/Search Tags:sub-division image data, solid state storage, Hilbert coding, window query
PDF Full Text Request
Related items