For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
A New Approach to Cell Loss Analysis for Long-Range Dependent Network Traffic
Chien Trinh NGUYEN Shinji SUGAWARA Tetsuya MIKI
IEICE TRANSACTIONS on Communications
Publication Date: 2005/09/01
Print ISSN: 0916-8516
Type of Manuscript: PAPER
cell loss, buffer overflow, network QoS, traffic prediction, long-range dependent traffic, multifractal wavelet model,
Full Text: PDF(866.1KB)>>
A new expression for cell/packet loss probability in an ATM and packet switched queue system with a finite buffer is presented. Cell and packet loss analysis is based on the new concept of a "buffer overflow cluster" and the overflow probability for a queue with an infinite buffer. This approach holds for a wide variety of long-range dependent traffic sources typical of wide-area networks, as well as Internet and other communication traffics. The method is verified by simulations of two long-range dependent traffic scenarios: fractional Gaussian noise and multifractal wavelet model traffic with a beta marginal distribution.