Landelijk Netwerk Mathematische Besliskunde
Course SMTS: Stochastic Models for Telecommunication Systems
Time: |
Monday 10.15 - 12.00 (September 12 - November 7). |
Location: |
Mathematical Building, Room 611AB, Budapestlaan, Utrecht (De Uithof). |
Lecturers: |
Prof.dr. R.D. van der Mei (CWI & VU) and Prof.dr. M.R.H. Mandjes (UvA) |
Course description:
The course consists of two synergistic parts.
Part I: stochastic models for telecommunication systems at the
packet/burst level.
Part II: stochastic models for telecommunication systems at the
call/flow level.
Part I:
At the packet/burst level performance aspects of telecommunication
networks can be modeled well by using fluid systems - think of packet
delays and loss in the buffers of an IP (Internet protocol) based
network. Part I of the course treats these fluid models, focusing on
single-node queues. The most prominent subjects are: (1) the M/D/1
queue; (2) the /D/1 queue fed by periodic sources; (3) the
Anick-Mitra-Sondhi model; (4) general Markov-fluid driven queues; (5)
exponential decay of the overflow probability; (6) effective bandwidth;
(7) reduction of fluid models to the call level, to enable the
application of the results of Part II.
Part II:
Call/flow-level models are used for the dimensioning of
telecommunication networks, and for determining the blocking
probabilities in ATM networks, throughputs in the Internet, and
download times in a wireless access network (WLAN, UMTS, etc.). In part
II the following subjects are treated:
(1) traffic modelling at the call level, and traffic measurements in
operational networks
(2) performance models for streaming traffic (speech, video), such as
the Erlang and Engset models, multi-rate models, call admission
policies like trunk reservation, loss networks, etc.
(3) performance models for elastic traffic (data), such as the
processor-sharing queue
(4) Performance models for TCP (TCP
is the transport protocol underlying the Internet), WLAN and UMTS
(5) models for integration of streaming and elastic traffic
(6) "Quality-of-Service aware" dimensioning of data networks.
Literature:
Lecture notes will be provided.
Prerequisites:
Basic knowledge of probability at the level: S.M. Ross, Introduction to probability models, 8th edition, Academic Press, 2003 (chapters 1-3).
Examination:
Take home problems.
Addresses of the lecturers:
Prof.dr. R.D. van der Mei
CWI, P.O. Box 94079, 1090 GB Amsterdam.
Phone: 020 - 5924129 E-mail: r.d.van.der.mei@cwi.nl
Prof.dr. M.R.H. Mandjes
Korteweg-de Vries Institute for Mathematics,
University of Amsterdam, Science Park 904,1098 XH Amsterdam
Phone: 0 20 - 5255164 E-mail: m.r.h.mandjes@uva.nl
|