基于关联规则的P2P视频点播缓存替换算法  

Association Rule-based P2P Video-on-Demand System Cache Replacement Algorithm

作　　者：侯浦[1] 

机构地区：[1]北京邮电大学信息与通信工程学院,北京100876

出　　处：《软件》2012年第12期142-144,162,共4页Software

摘　　要：传统的缓存替换算法缺乏对视频点播系统的针对性,不能有效的重用缓存数据,影响缓存查询命中率。为此,本文针对视频点播系统用户VCR操作的特点,提出了一种基于缓存数据块关联规则的P2P视频点播系统客户端缓存替换算法。此算法根据视频点播用户的VCR操作特征,在每个节点记录用户的播放历史,并利用P2P网络的优势,在节点间互换播放记录。节点从自身收集到的播放记录中提取出视频数据块之间的关联规则,根据关联规则更新数据块的流行度。缓存区被填满后流行度低的数据块就会被流行度较高的数据块替换。实验结果表明,与传统缓存替换算法相比,本算法能够提高缓存查询命中率,减轻服务器的负载。Traditional cache replacement algorithm lacks pertinence to video-on-demand system, so it cannot reuse the data in cache efficiently, and the performance of cache is affected. This paper proposes a cache data association rules based P2P video-on-demand system client cache replacement algorithm based on the characteristics of VoD user＇s VCR operations. According to user＇s VCR operations, every peer in the system records its play history. Peers can exchange their records through P2P network. A peer extracts association rules from the records it collected. Then the popularity of a data block can be updated based on association rules. A data block with low popularity will be replaced with the data block with high popularity when the buffer is filled. Experimental results show this algorithm can efficiently improve cache hit ratio than traditional algorithms. Then the overall performance of system is improved.

关 键 词：视频点播 缓存替换 VCR操作 关联规则 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]