網(wǎng)絡(luò)業(yè)務(wù)量的一種穩(wěn)定多分形模型及其檢驗
A Stable Multifractal Model of Network Traffic and Its Verification
-
摘要: 目前網(wǎng)絡(luò)業(yè)務(wù)量研究中有自相似和多分形兩種主要的先驗?zāi)P?。自相似模型簡潔,運用單一的H 指數(shù)來描述業(yè)務(wù)量特性,但它并不能完全表達業(yè)務(wù)量的信息;多分形模型更準(zhǔn)確,但它運用函數(shù)形式的多分形譜來刻畫信號,不便于實際分析。本文在無窮可分層疊的理論框架下提出了網(wǎng)絡(luò)業(yè)務(wù)量的一種穩(wěn)定多分形模型。運用小波變換方法,通過多個業(yè)務(wù)量記錄的檢驗,證實此模型從尺度行為和邊緣分布兩方面都能準(zhǔn)確刻畫業(yè)務(wù)量。作為一般多分形過程的特例,該模型把多分形指數(shù)的表達式確定下來,并歸結(jié)為4個參數(shù)的估計。
-
關(guān)鍵詞:
- 多分形;穩(wěn)定分布;突發(fā)性;離散小波變換
Abstract: There have been two a priori models up till now concerning network traffic: self-similar and multifractal models. The self-similar model is simple enough to characterize network traffic with the single exponent H, however it can not thoroughly describe the contents. The multifractal model, which is more accurate, is not practical because it uses multifractal spectra as functions to represent signals. Therefore, on the basis of the framework of IDC theory, this paper proposes a stable multifractal model of network traffic. It is tested on several traces by using the method of wavelet transforms that the model fits traffic precisely in terms of both scale behaviors and marginal distributions. As a specific case of multifractal processes, this model, by estimations of four parameters, can specify the representation of the multifractal exponent. -
計量
- 文章訪問數(shù): 2234
- HTML全文瀏覽量: 102
- PDF下載量: 783
- 被引次數(shù): 0