论文部分内容阅读
网络流量整形、调度、异常检测、管理与控制及保障QoS需求等都需要了解业务流的局部变化特性.本文给出离散小波及其模极大值的网络流量奇异谱估计算法及影响因素,并通过真实的网络业务数据对算法进行了评估和比较.实验结果表明,两种方法的奇异谱估计能有效刻画网络业务流的局部变化特征,并且能通过奇异谱特征参数之间的差别描述不同业务流之间的差异性,也表明了在一定条件下,离散小波模极大法更加优越.
Network traffic shaping, scheduling, anomaly detection, management and control and QoS requirements need to know the local variation characteristics of the service flow.In this paper, we give the algorithm of estimating the singular spectrum of the network traffic and its influencing factors The real network business data is used to evaluate and compare the algorithm.The experimental results show that the singular spectrum estimation of the two methods can effectively characterize the local variation of the network traffic flow and describe the different services through the differences of the singular spectrum feature parameters The difference between the flow, also shows that under certain conditions, discrete wavelet modulus maxima more superior.