摘要： 针对传统长相关网络流量Hurst指数估计算法估计结果不准确、可变信息受损严重的情况，提出时域内滑窗时变方差之差Hurst指数估计算法，采用已知参数的人工分形高斯噪声序列及Bellcore采集的真实网络流量序列BC-pOct89对其进行验证。结果表明该算法减少了可变信息损失，能动态地刻画全域上的长相关特性，具有较高的准确性和鲁棒性。
关键词:
长时相关,
滑窗时变,
分形高斯噪声,
鲁棒性
Abstract: Because result of long range dependent networks traffic Hurst exponent estimate arithmetic is not exact and loses much changed information, this paper proposes Slide Window Time Variety(SWTV) variance’s dispersion Hurst exponent estimate arithmetic, which uses Fractal Gauss Noise(FGN) list whose parameters is known and real networks traffic list BC-pAug89 collected by Bellcore to test it. The result indicates this arithmetic reduces loss of changed information, depicts Long Range Dependent(LRD) characteristic in all field, and gets well veracity and robust .
Key words:
Long Range Dependent(LRD),
slide window time variety,
Fractal Gauss Noise(FGN),
robust
中图分类号:
张 博;汪斌强;智英建. 长相关网络流量Hurst指数估计算法[J]. 计算机工程, 2009, 35(2): 83-85.
ZHANG Bo; WANG Bin-qiang; ZHI Ying-jian. Long Range Dependent Networks Traffic Hurst Exponent Estimate Arithmetic[J]. Computer Engineering, 2009, 35(2): 83-85.