TY - JOUR
T1 - New view on adversarial queueing on MAC
AU - Hradovich, Elijah
AU - Klonowski, Marek
AU - Kowalski, Dariusz R.
N1 - Funding Information:
Manuscript received November 22, 2020; accepted December 12, 2020. Date of publication December 29, 2020; date of current version April 9, 2021. This paper is supported by Polish National Science Center NCN grant UMO-2017/25/B/ST6/02553. The associate editor coordinating the review of this letter and approving it for publication was M. H. Cheung. (Corresponding author: Marek Klonowski.) Elijah Hradovich and Marek Klonowski are with the Department of Fundamental Problems of Technology, Wrocław Univeristy of Science and Technology, 50-370 Wrocław, Poland (e-mail: ilya.hradovich@pwr.edu.pl; marek.klonowski@pwr.edu.pl).
Publisher Copyright:
© 1997-2012 IEEE.
PY - 2021/4
Y1 - 2021/4
N2 - This work proposes an average-case performance measurement as a meaningful and non-trivial way of evaluating and comparing queueing protocols on a multiple-access channel (MAC). We prove that Little's Law holds for this way of measuring latency and number of packets in the system, also for adversarial packet arrival (which was not the case for the previously considered worst-case measurement), and analyze performance of popular MAC protocols. Interestingly, some of them have asymptotically the same average and worst case queue sizes, but not packet latencies, while others exhibit entirely opposite behavior. We also propose a consistent methodology for classifying queueing protocols, based on protocol features, such as synchronization, collision detection, queue-size awareness etc.
AB - This work proposes an average-case performance measurement as a meaningful and non-trivial way of evaluating and comparing queueing protocols on a multiple-access channel (MAC). We prove that Little's Law holds for this way of measuring latency and number of packets in the system, also for adversarial packet arrival (which was not the case for the previously considered worst-case measurement), and analyze performance of popular MAC protocols. Interestingly, some of them have asymptotically the same average and worst case queue sizes, but not packet latencies, while others exhibit entirely opposite behavior. We also propose a consistent methodology for classifying queueing protocols, based on protocol features, such as synchronization, collision detection, queue-size awareness etc.
KW - MAC
KW - adversarial queueing
KW - average-case analysis
UR - http://www.scopus.com/inward/record.url?scp=85099102898&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85099102898&partnerID=8YFLogxK
U2 - 10.1109/LCOMM.2020.3047997
DO - 10.1109/LCOMM.2020.3047997
M3 - Article
AN - SCOPUS:85099102898
SN - 1089-7798
VL - 25
SP - 1144
EP - 1148
JO - IEEE Communications Letters
JF - IEEE Communications Letters
IS - 4
M1 - 9310218
ER -