Performance analysis of buffered random multiple access protocols

by Wan, Tao

Abstract (Summary)
This thesis deals with the performance anaiysis of random multiple access (MA) systems, particularly the systems where each user has a buffer capacity for more than one customer (packet or message). A unified approach, ├žalled the Tagged User Approach (TUA), is proposed in this thesis for the approximate analysis of the performance of buffered random MA systems. In buffered MA systems, each user is a queueing system whose customer service time is detennined by the behaviors of ail users in the system. However, among the available literatures on the performance analysis of these systems, only few of hem employe existing queueing theory because of the difficulty to find the customer service time distribution. This difficulty is overcorne in TUA where the state flow graph is introduced to obtain the vimial customer service tirne distribution for each user efficiently, which is then applied to the user queueing analysis using the classical queueing theory. Consequently, the analyses on chamel contention scheme and user queueing behaviors are decoupled. This leads to the significant simplification of the system analysis compared to the available approaches. Hence, TUA is able to handle very complicated systems where the available approaches rnay fail to work. Moreover, TUA is able to analyze the stability of buffered MA systems. In this thesis, the stability of buffered S-ALOHA systern is analyzed, for the first time, using TUA. TUA is an analytical fiamework. Particular techniques may be required when TUA is applied to the performance analysis of a random MA system. The techniques for analyzing iii several typical protocols such as ALOHA, CSMA/CD, R-ALOHA and PRMA have been developed in this thesis. Finally, the accuracy of the TUA analysis has been venfied to be very good by simulation.
Bibliographical Information:



School Location:

Source Type:Master's Thesis



Date of Publication:01/01/1999

© 2009 All Rights Reserved.