A model for computation of the transmission delay distribution in infinite population nonpersistent CSMA/CD is presented, together with computational approximations. Recent extensions of the use of CSMA/CD to heavy load realtime applications, such as packet voice communication and process control, motivated this work. The realtime condition demands quantiles of packet delay distributions, which in turn requires that the entire distribution be calculated rather than representative values such as mean delay and coefficient of variation. The model decomposes into two submodels describing channel state and delay distribution, respectively. The channel model is of infinite population justifying the assumption that the channel behavior affects an individual station's behavior, whereas the converse does not take place. The formulation is by discrete time Markov chain; time is grained into slots or backoff unit time. An arbitrary packet size distribution is allowed. The solution is obtained in closed form; measures such as throughput and probability of trial success are also available in closed form. The delay model assumes that channel access trials are independent of each other; the calculation consists of convolutions and weighted sum of distributions. Because this computation is rather heavy, approximations are included.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Kiyoshi YONEDA, "Percentile Delay Calculation for the Infinite Population Nonpersistent CSMA/CD" in IEICE TRANSACTIONS on transactions,
vol. E68-E, no. 6, pp. 371-375, June 1985, doi: .
Abstract: A model for computation of the transmission delay distribution in infinite population nonpersistent CSMA/CD is presented, together with computational approximations. Recent extensions of the use of CSMA/CD to heavy load realtime applications, such as packet voice communication and process control, motivated this work. The realtime condition demands quantiles of packet delay distributions, which in turn requires that the entire distribution be calculated rather than representative values such as mean delay and coefficient of variation. The model decomposes into two submodels describing channel state and delay distribution, respectively. The channel model is of infinite population justifying the assumption that the channel behavior affects an individual station's behavior, whereas the converse does not take place. The formulation is by discrete time Markov chain; time is grained into slots or backoff unit time. An arbitrary packet size distribution is allowed. The solution is obtained in closed form; measures such as throughput and probability of trial success are also available in closed form. The delay model assumes that channel access trials are independent of each other; the calculation consists of convolutions and weighted sum of distributions. Because this computation is rather heavy, approximations are included.
URL: https://global.ieice.org/en_transactions/transactions/10.1587/e68-e_6_371/_p
Copy
@ARTICLE{e68-e_6_371,
author={Kiyoshi YONEDA, },
journal={IEICE TRANSACTIONS on transactions},
title={Percentile Delay Calculation for the Infinite Population Nonpersistent CSMA/CD},
year={1985},
volume={E68-E},
number={6},
pages={371-375},
abstract={A model for computation of the transmission delay distribution in infinite population nonpersistent CSMA/CD is presented, together with computational approximations. Recent extensions of the use of CSMA/CD to heavy load realtime applications, such as packet voice communication and process control, motivated this work. The realtime condition demands quantiles of packet delay distributions, which in turn requires that the entire distribution be calculated rather than representative values such as mean delay and coefficient of variation. The model decomposes into two submodels describing channel state and delay distribution, respectively. The channel model is of infinite population justifying the assumption that the channel behavior affects an individual station's behavior, whereas the converse does not take place. The formulation is by discrete time Markov chain; time is grained into slots or backoff unit time. An arbitrary packet size distribution is allowed. The solution is obtained in closed form; measures such as throughput and probability of trial success are also available in closed form. The delay model assumes that channel access trials are independent of each other; the calculation consists of convolutions and weighted sum of distributions. Because this computation is rather heavy, approximations are included.},
keywords={},
doi={},
ISSN={},
month={June},}
Copy
TY - JOUR
TI - Percentile Delay Calculation for the Infinite Population Nonpersistent CSMA/CD
T2 - IEICE TRANSACTIONS on transactions
SP - 371
EP - 375
AU - Kiyoshi YONEDA
PY - 1985
DO -
JO - IEICE TRANSACTIONS on transactions
SN -
VL - E68-E
IS - 6
JA - IEICE TRANSACTIONS on transactions
Y1 - June 1985
AB - A model for computation of the transmission delay distribution in infinite population nonpersistent CSMA/CD is presented, together with computational approximations. Recent extensions of the use of CSMA/CD to heavy load realtime applications, such as packet voice communication and process control, motivated this work. The realtime condition demands quantiles of packet delay distributions, which in turn requires that the entire distribution be calculated rather than representative values such as mean delay and coefficient of variation. The model decomposes into two submodels describing channel state and delay distribution, respectively. The channel model is of infinite population justifying the assumption that the channel behavior affects an individual station's behavior, whereas the converse does not take place. The formulation is by discrete time Markov chain; time is grained into slots or backoff unit time. An arbitrary packet size distribution is allowed. The solution is obtained in closed form; measures such as throughput and probability of trial success are also available in closed form. The delay model assumes that channel access trials are independent of each other; the calculation consists of convolutions and weighted sum of distributions. Because this computation is rather heavy, approximations are included.
ER -