| 移动环境下的网页预取及其与Web缓存的集成机制研究 |
| 其他题名 | Research on Integrated Prefetching and Web Caching Scheme for Mobile Environment
|
| 田四化
|
| 2007-06-03
|
| 学位授予单位 | 中国科学院软件研究所
|
| 学位 | 博士
|
| 学位授予地点 | 软件研究所
|
| 关键词 | Web缓存
预测
预取
缓存替换
数据挖掘
|
| 摘要 | Web缓存技术是Internet环境下减轻Web服务器负载、缩短用户访问延迟、提高Web访问性能的有效方法。在移动计算环境中，由于受设备和网络的限制，缓存策略的设计必须权衡性能收益和代价两者，才能获得用户满意的效果。另一方面，网页预取技术能够隐藏网络延迟，适应无线网络的速度慢、易断连特征，利用该技术，能改善用户请求的响应时间，有效利用网络带宽。针对Web缓存和预测算法，学术界已经有不少研究，但大多研究工作只是比较孤立地讨论单个算法，而网页预取和缓存机制的有效性验证、替换策略等是相互影响的，它们共同决定着缓存命中率。
本文对移动环境下的网页预测、预取及其与Web缓存的集成机制进行研究。基于数据挖掘理论和信息论提出适用于移动环境的轻量级预测算法，使用序列模式的挖掘方法，以少量的计算代价统计用户访问历史的一阶关联关系，构造预取集合，动态生成预取策略；同时提出一套集成网页预取和缓存系统的机制，包括上下文环境感知的预取策略和基于缓存收益的替换策略。当设备资源不足时，根据当前上下文环境进行权衡，调整预取策略，保证服务质量。发生缓存替换时，收益驱动的替换策略充分考虑了缓存数据大小、网络传输代价、被访问频度以及被预取的机率等影响缓存收益的因素，判断缓存中各数据项的收益，决定缓存替换的优先级，从而将预取和替换紧密的联系在一起。我们构建了移动网页缓存系统EasyCache实现上述机制。模拟实验证明，这种有效的预取和缓存集成机制，不仅可以提高缓存命中率，还具有很低的计算开销，适用于移动手持设备。 |
| 其他摘要 | Web caching is an effective technology to alleviate the load of Web servers, decrease perceived access delays and improve the performance of Web access. In mobile computing environments, because of the constraints of wireless networks and devices, the design of caching scheme must trade off between performance benefits and costs in order to produce the effects satisfied by users. On the other hand, page prefetching can conceal the network delays, which exactly matches the features of mobile environment including slow network speed and frequent disconnections. By taking advantage of prefetching, the response time of user request will be reduced and network bandwidth will be better exploited. There are a lot of researches on Web caching and prefetching, but they mainly focus on a single algorithm. In fact, web prefetching, cache validating and replacement policy influence each other, which finally affect caching hit ratio as a whole.
This thesis studies on page prediction algorithm, prefetching algorithm and their integrated scheme with Web caching for mobile environment. We propose a sequence-mining based prediction algorithm with the less amount of calculation which mines 1-gram association patterns from Web history records to form the prefetching candidate set for mobile client, proposing an adaptive prefetching strategy to automatically decide the appropriate prefetching level, formulating the item’s cache profit and then devising profit-driven caching replacement policy to glue the caching and prefetching in terms of mobile client’s context , which are all implemented in EasyCache. Experiments show that this algorithm not only successfully improves cache hitting rate but also reduces computing overhead, which is very suitable to handheld devices. |
| 页数 | 70
|
| 语种 | 中文
|
| 内容类型 | 学位论文
|
| URI标识 | http://ir.iscas.ac.cn/handle/311060/5918
|
| 专题 | 中科院软件所_中科院软件所
|
推荐引用方式
GB/T 7714 |
田四化. 移动环境下的网页预取及其与Web缓存的集成机制研究[D]. 软件研究所. 中国科学院软件研究所,2007.
|
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论