CACC:一种基于分簇的自适应网络坐标计算方法

针对Internet环境中采用基于坐标的方法预测网络节点间的距离问题,提出了一种基于分簇的自适应网络坐标计算方法CACC.首先,提出了E-Binning分簇算法,并在此基础上通过比较不同的地标选择策略,CACC采用了最优的分布式地标策略Hybrid;其次,提出了自适应坐标更新算法ACU以提高坐标计算的收敛速度和精确度;最后,采用相对误差和SRRL度量方法,通过与经典的网络坐标计算方法GNP和Vivaldi对比验证了方法的精确性.实验结果表明,CACC是一种可扩展性强、收敛性好、预测精度高的面向实际应用的网络距离预测方法.

CACC: An Adaptive Network Coordinate Computing Method Based on Clustering

Aiming at the problem of predicting distances between network nodes using coordinatebased method in the Internet environment,an adaptive network coordinate computing method CACC is put forward based on clustering.Firstly,E-Binning clustering algorithm is proposed,and based on which different strategies of selecting landmarks have been studied.The CACC adopts the optimal distributed landmark strategy,Hybrid.Secondly,an adaptive coordinate updating algorithm ACU is proposed to improve the convergency and accuracy of the coordinate computing.The accuracy of CACC is validated by comparing it to the classic network computing methods GNP and Vivaldi under the relative error and SRRL metrics.The results of the experiment show that CACC is an applicationoriented network prediction method that has good scalability,convergency and accuracy.