Using Series-Parallel Reduction and Stratified Sampling in Estimating Network Relibility

• Main Authors: Chia-Shuan Li, 李佳璇
• Other Authors: Wei-Ning Yang
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/x35ss3

碩士 === 國立臺灣科技大學 === 資訊管理系 === 95 === In order to evaluate the performance of a complicated stochastic network system, network reliability is a decisive key factor to the administrator. The purpose of this paper is to obtain a more accurate estimator within limited time. Because numerical evaluation method for network reliability is an NP-hard problem, an alternative approach to the exact evaluation is to estimate network reliability using computer simulation. Crude Monte Carlo method suffers from requiring large sampling efforts when the network is highly reliable, so variance reduction techniques which reduces the variance of the estimator without increasing the sampling efforts must be used. Cancela,H. and Khadiri,M.E (2003) incorporates series-parallel reductions in a recursive variance reduction algorithm and avoids redundant identical computations. This paper proposes an exhaustively stratified sampling scheme to enhance the variance reduction. Empirical results show that the proposed method outperforms the existing sampling methods.