Estimation of Packet Traffic Modeling in High-speed Networks with a Fuzzy-AR Approach

• Main Authors: Wang, Ku-Chen, 王谷辰
• Other Authors: Chen, Bor-Sen
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/27130556739497575175

碩士 === 國立清華大學 === 電機工程學研究所 === 86 === 　　In general, high-speed network traffic is a complex, nonlinear, nonstationary process and significantly affected by unmeasurable parameters and variables. Thus a precise model of the process becomes increasingly difficult to proceed as the complexity of the process increases. Recently, fuzzy model is found to be a powerful method to discribe a real complex process with its nonlinear and time-varying properties. In this study, a fuzzy autoregressive (Fuzzy-AR) model is proposed to describe the traffic characteristics of high-speed networks. 　　The Fuzzy-AR model approximates a nonlinear time-variant process with a combination of several linear AR process via fuzzy clustering method. We advocate the use of Fuzzy-AR model because of its flexibility in capturing the traffic characteristics at different high speed network's structures. Furthermore, the Fuzzy-AR model is more suitable for the development of congestion control scheme of packet network traffic. A parameter estimation problem of Fuzzy-AR modeling is treated by a clustering algorithm from actual traffic data in networks. Finally, based on the packet traffics on Ethernet-LAN network in Bellcore, several illustrative examples are given to demonstrate the validity of the proposed Fuzzy-AR model in high speed network traffic.