论文部分内容阅读
新近对局域网和广域网上大量突发业务流量的监测结果表明，采用自相似建模表征业务到达过程的长时间相关特性具有较高精度，其中Hurst系数是表征业务突发特性的重要参数，因此在一定的观察时间内对突发业务的Hurst系数进行快速、准确的估计是高速宽带网络（如ATM）实施流量控制和缓冲资源分配的前提．本文提出一种基于多分辨率采样和小波分析的Hurst系数快速估计方法，对严格二阶自相似模型下Hurst系数的估计为最大似然估计．此方法还可对不同观察时隔下业务源Hurst系数的一致性进行检验，而且比传统估计方法大大减少了对突发业务流进行离散采样的计数样本总量．采用分形高斯噪声和真实突发业务数据的仿真结果均表明，本文所述方法比传统的R/S、方差时间分析等估计方法具有更高的估计精度，而且在样本量不足时具有更好的鲁棒性．本文的方法可望应用于ATM网络的业务量管制和拥塞控制．
Recent monitoring results of a large number of burst traffic on LANs and WANs show that the long-term correlation characteristics of service arrival process using self-similar modeling are highly accurate. Among them, Hurst coefficient is an important parameter for characterizing service burst characteristics. Therefore, A fast and accurate estimation of the Hurst coefficient of a burst service during a certain observation time is a prerequisite for implementing traffic control and buffering resource allocation in a high-speed broadband network such as an ATM. In this paper, a fast Hurst coefficient estimation method based on multiresolution sampling and wavelet analysis is proposed. The maximum likelihood estimation of Hurst coefficients under the strict second-order self-similar model is proposed. This method also verifies the consistency of Hurst coefficients of different business sources under different observation intervals and greatly reduces the total number of counting samples discretely sampled for burst traffic compared with the traditional estimation method. The simulation results using fractal Gaussian noise and real burst service data both show that the proposed method has higher estimation accuracy than the traditional R / S and variance time analysis methods, and has a better estimation accuracy when the sample size is not enough Robustness. The method in this paper is expected to be applied to traffic control and congestion control in ATM networks.