网络距离预测方法的研究  

Study on Network Distance Prediction

作　　者：张一飞[1] 张宏莉[1] 

机构地区：[1]哈尔滨工业大学计算机网络与信息安全技术研究中心,哈尔滨150001

出　　处：《电信科学》2010年第12期56-62,共7页Telecommunications Science

基　　金：国家重点基础研究发展计划"973"基金资助项目(No.2005CB321806);国家"863"计划基金资助项目(No.2007AA01Z442;No.2009AA01Z437);国家自然科学基金资助项目(No.60703014)

摘　　要：在Internet中,节点间的网络距离是影响网络应用性能的重要原因之一。虽然通过直接测量可以优化它们的性能,但是由于会产生较大的测量开销和网络拥塞而不适用于实际的大规模网络应用。网络距离预测方法既可以满足实际应用的需要又可以减少由直接测量产生的网络侵扰,是当前的研究热点之一。首先介绍了它们的研究历程并按照不同的标准进行分类,接着在以上分类基础上对这些方法进行详细的分析和讨论,最后探讨未来的研究方向并总结全文。Latencies between nodes are one of important reasons that affect the performance of network applications. Their performance can be optimized by direct measurement. However, measurement on-demand is impractical in large-scale network applications because of the high measurement overhead and network congestion. By using the network distance prediction technology the needs of improving the performance of network applications and reducing network intrusion can be met. In this paper, the research history of network distance prediction and its classification which based on different criteria are introduced firstly. Then detailed analyses on these methods are investigated depending on different classification. Lastly, a few suggestions on future work are put forward.

关 键 词：网络距离预测 网络坐标 网络测量 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]