基于包对测量的路径容量估计方法改进

Improvement of the Packer-pair Based Path Capacity Estimation Methods

测量样本的统计分析是基于包对技术的路径容量估计的关键.提出一种路径容量包对估计方法改进,将端到端路径视为离散控制过程的系统,以路径容量描述其状态.首先以包对探测方式连续发送一定数量的背靠背包对序列,获取足够的路径容量测量样本;再采用卡尔曼过滤算法对测量样本进行统计分析,以准确估计路径容量.建立了路径容量估计的滤波方程,给出了路径容量估计过程.实验表明,与pathrate算法相比,提高了估计的准确性并降低了测量探测量和测量时间.

Statistics analyzing of the measurement samples is the crucial problem in packet-pair based path capacity estimation. In this paper, an improved packet pair based path capacity estimation method has been proposed, which treats the end to end path as dispers- ed control system, and uses path capacity to describe it＇s states. The method first collects enough measurement samples by packet pair probing which send a back to back packet pair sequence; then uses the Kalman Filer algorithm to statistics analyze the measurement samples, to accurately estimate path capacity. The filer equations are built and the path capacity estimation progress is given also in this paper. Experiments show that, when comparing with pathrate, the estimation accuracy is improved while both the probe cost and measurement time are decreased.