基于时空老化模型的服务端瓦片缓存置换算法

随着网络地理信息服务（network geographic information service，NGIS）向云服务演进，客户端瓦片缓存架构的应用局限性逐渐体现。为提升瓦片服务的性能，在老化算法的基础上，综合分析了瓦片访问长短期流行度和瓦片大小特征，设计了基于时空老化模型的服务端瓦片缓存置换算法（server-side cache replacement algorithm based on spatiotemporal aging model for tiles，SSAT），并利用谷歌全球底图瓦片和瓦片访问日志进行了仿真实验。结果表明，在不同缓存空间下，SSAT的缓存命中率均高于传统算法，缓存空间每增加1 MB，最多可以提高0.24%的请求命中率和0.23%的字节命中率；当缓存空间为500 MB时，SSAT能达到73%的请求命中率和76%的字节命中率，平均访问时长可缩短35%以上。SSAT能兼顾性能与资源消耗，具备高效性和扩展性。

Server-side cache replacement algorithm based on spatiotemporal aging model for tiles

As network geographic information service (NGIS) is developing towards cloud services, the client-side tiles cache architecture has already demonstrated its limitations in applications. To improve the efficiency of tile service, this paper designs a server-side cache replacement algorithm based on spatiotemporal aging model for tiles (SSAT). The SSAT algorithm is based on the aging algorithm and comprehensively analysis on the long-term and short-term popularity of tile access pattern as well as the characteristics of tile size. A series of Google global base map tiles and access logs of tiles are used for simulation experiment. The experimental results show that the SSAT algorithm outperforms traditional algorithms when using different cache sizes. With the increasing of cache size, the SSAT algorithm can increase the request hit rate by 0.24% and the byte hit rate by 0.23% for every additional 1 MB of cache. When the cache space is 500 MB, SSAT algorithm can achieve 73% cache hit rate and 76% byte hit rate, and reduce the average access time by more than 35%. In summary, the SSAT algorithm considers both performance and resource consumption, and has high efficiency and scalability.