TY - JOUR
T1 - Fiber-optic-based protocol for manufacturing system networks. Part II. Statistical analysis
AU - Ayyagari, Arun
AU - Ray, Asok
N1 - Copyright:
Copyright 2004 Elsevier B.V., All rights reserved.
PY - 1990
Y1 - 1990
N2 - Conceptual development, architecture, statistical and simulation models, and the results of test and verification of a fiber- optic-based protocol have been reported in a sequence of two papers. The protocol uses the unidirectional bus topology and is specifically designed for computer integrated manufacturing (CIM) networks. The architecture of the protocol is formulated in Part I. This second part presents a statistical model of the proposed protocol for analyzing the network-induced delays and for generating pertinent results of analysis and simulation for different scenarios of network traffic. The major assumptions in formulating the statistical model are: (1) message arrival processes for real-time (RT) and non-real time (NRT) modes are independent and Markov; and (2) message lengths follow independent general distributions with known statistical properties. The Laplace-Stieltjes transforms of probability distribution functions of queueing delays for RT and NRT messages have been derived. The analytical results for the first two moments of both RT and NRT queueing delays have been compared with those obtained from discrete-event simulation.
AB - Conceptual development, architecture, statistical and simulation models, and the results of test and verification of a fiber- optic-based protocol have been reported in a sequence of two papers. The protocol uses the unidirectional bus topology and is specifically designed for computer integrated manufacturing (CIM) networks. The architecture of the protocol is formulated in Part I. This second part presents a statistical model of the proposed protocol for analyzing the network-induced delays and for generating pertinent results of analysis and simulation for different scenarios of network traffic. The major assumptions in formulating the statistical model are: (1) message arrival processes for real-time (RT) and non-real time (NRT) modes are independent and Markov; and (2) message lengths follow independent general distributions with known statistical properties. The Laplace-Stieltjes transforms of probability distribution functions of queueing delays for RT and NRT messages have been derived. The analytical results for the first two moments of both RT and NRT queueing delays have been compared with those obtained from discrete-event simulation.
UR - http://www.scopus.com/inward/record.url?scp=0025536982&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0025536982&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:0025536982
SN - 0402-1215
JO - American Society of Mechanical Engineers (Paper)
JF - American Society of Mechanical Engineers (Paper)
T2 - Proceedings of the Winter Annual Meeting
Y2 - 25 November 1990 through 30 November 1990
ER -