摘要

A simple and systematic path-level performance evaluation method mainly composed of a decomposition technique and a parameter-fitting algorithm is proposed in this paper. With comprehensive considerations, this method works for a discrete-time tandem network with input of two-state Markov-modulated Bernoulli processes (MMBPs), which have potential applications in computer and communication networks. Of course, this proposed method offers convenience of traffic aggregation and decomposition for multiple MMBPs. Furthermore, it can be easily extended to work for tree-type networks. To validate the proposed method, several experiments are conducted to demonstrate that the proposed method has good accuracy in terms of delay, delay variance, and packet loss probability. For further checking the effectiveness of this method, comparisons with some closely related methods in the literature are also made.

  • 出版日期2012-4

全文