论文部分内容阅读
延迟是评价网络性能的重要指标,也是进行其他网络性能指标测量的基础.基于全球定位系统(GPS)的端到端(end-to-end)时钟同步是测量网络单向指标的常用方法,但是其代价昂贵且缺乏灵活性.在无端到端时钟同步机制下进行网络单向延迟指标测量的关键是消除时钟偏差效应的影响.基于对时间序列分段技术的分析,提出了一种新的时间序列分段标准与改进的分段算法,实现序列的自动聚类,其时间复杂度为O(N2).将该算法应用于检测端到端时钟的动态性,识别测量过程中时钟跳变和时钟频率调整位置,实现对网络单向延迟的测量,弱化了同类工作中对时钟动态性的严格假设.同时提出了基于滑动窗的在线实时时钟动态性检测算法.实际测试实验表明,该算法是行之有效的.
Delays are an important indicator of network performance and the basis for making measurements of other network performance measures.General Positioning System (GPS) -based end-to-end clock synchronization is a common method of measuring network unidirectional metrics, but Which is costly and inflexible.The key to measure unidirectional delay in network under the mechanism of end-to-end clock synchronization is to eliminate the effect of clock bias effect.Based on the analysis of time series segmentation technology, a new time Sequence segmentation standard and improved segmentation algorithm to achieve automatic clustering of sequences, the time complexity is O (N2) .The algorithm is applied to detect the dynamics of the end-to-end clock, to identify the clock transition in the measurement process and Clock frequency to adjust the position, to achieve the one-way network delay measurement, weakening the similar work on the clock dynamics of the strict assumptions.At the same time, based on the sliding window-based real-time clock dynamic detection algorithm.Experimental results show that the algorithm is effective.