一种基于主题时空价值的服务器端瓦片缓存算法

A server-side tile caching algorithm based on theme temporal and spatial value

先进先出置换算法(FIFO)、最近最少使用置换算法(LRU)、最不经常使用置换算法(LFU)等传统缓存算法侧重于数据的访问时间和频率,而现有面向瓦片的缓存算法大多只适用单一类型的瓦片数据,不能兼顾瓦片数据具有的多类型与空间位置特性,在服务器端瓦片缓存的应用上存在局限性。通过构建面向多类型瓦片的服务器端缓存索引,并综合考虑瓦片访问中的时间局部性、空间局部性和用户主题倾向性,提出基于主题时空价值的瓦片数据缓存置换算法(G DTST)。实验结果表明,相较于传统服务器端缓存置换算法,GDTST可以提高缓存命中率,并且在缓存容量较大时具有更好的延迟节省率,能够有效降低瓦片源服务端负载,提高用户响应速度。

Traditional cache algorithms such as FIFO,LRU,and LFU focus on the access time and frequency of data,while existing tile cache algorithms are mostly applicable to single-type tile data,therefore failing to balance the multitype and spatial location characteristics of the tile data,and limiting their application on server-side tile caching.By constructing the server-side cache index for multi-type tiles and taking account of the temporal locality,spatial locality and user theme tendency in tile access,a tile data cache replacement algorithm GDTST based on the topic time-space value is proposed.The experimental results show that compared with the traditional server-side cache replacement algorithm,GDTST can improve the cache hit ratio,and has a better latency reduction ratio when the cache capacity is larger,which effectively reduces the load on the tile source server and accelerates the user response speed.