基于OWGA缓存的海量遥感瓦片数据发布  

Massive Remote Sensing Tile Data Releasing Based on OWGA Cache

作　　者：苏焕焕[1,2] 沈夏炯[1] 赵亚萌[2] 徐鹏[1,2] 黄祥志[2] 

机构地区：[1]河南大学计算机与信息工程学院,河南开封475004 [2]中国科学院遥感与数字地球研究所,北京100101

出　　处：《地理空间信息》2017年第5期67-70,共4页Geospatial Information

基　　金：高分重大专项课题资助项目(Y4D00100GF;Y4D0100038);中科院战略先导专项课题资助项目(Y1Y02230XD)

摘　　要：通过模拟实现传统的WMTS瓦片服务对客户端发来下载瓦片数据的HTTP请求进行响应,计算得到各处理阶段所用的平均时间,分析测试结果可知,将请求的本地储存瓦片读取到内存中所花费时间占总响应时间的70%以上。为缩短服务器响应时间,提出了基于OWGA缓存的发布方法,将客户端最近频繁请求的、高分辨率的瓦片数据常驻内存,以减少文件I/O操作;并在缓存不足时移除OWGA值较低且低分辨率的瓦片数据。通过与常用的内存缓存置换算法LRU、LFU对比可知,在随机访问模式下,该算法的时间效率优于其他两种算法。Through simulating the traditional WMTS tile service responding to HTTP request of downloading tiles, we calculated the average time which spend in each processing phase, and analyzed test results. The result showed that the obtained response time was mostly spent in reading local tile requested into memory, accounting for more than 70% of the total response time. To shorten the response time, we proposed a release method based on OWGA cache to achieve saving data tiles recently and frequently requested and with high resolution in memory, reducing time-consuming file I/O operations, and removing tiles data with low OWGA value and low resolution when cache was insufficient. Compared with the conventional memory cache replacement algorithm, the algorithm has the advantage over the other two algorithms on the time efficiency.

关 键 词：五层十五级组织结构 数据发布 HTTP协议 OWGA算子 内存缓存 

分 类 号：P237[天文地球—摄影测量与遥感]