IEEE Wireless Communications (February)

75 Pages·2006·4.036 MB·English

IEEE • PRICINGVOWLAN THROUGHAMICRO-ECONOMICFRAMEWORK
W
February 2006, Vol. 13 No. 1 • QOS INIEEE 802.11EWIRELESSLANS
IRELESS • POLLING-BASEDPROTOCOLSFORPACKETVOICETRANSPORT
• IEEE 802.11EENHANCEMENTFORVOICESERVICE
• A CALLADMISSIONCONTROLFRAMEWORKFORVOWLANS
C
OMMUNICATIONS • A QUALITY-AWAREVOWLAN ARCHITECTUREANDITSQUANTITATIVE
EVALUATIONS
V
OICE OVER
W L
IRELESS OCAL
A N
REA ETWORKS
¤
A Publication of the IEEE Communications Society
In cooperation with IEEEComputer and VehicularTechnology Societies
®
LYT-TOC-FEB 2/2/06 11:01 AM Page 1
FEBRUARY 2006/VOL. 13, NO. 1
IEEE
W
IRELESS
C
OMMUNICATIONS
SPECIAL ISSUE
VOICE OVER WIRELESS LOCAL AREA NETWORK
4 36
GUESTEDITORIAL: PROVIDINGSTATISTICALQOS GUARANTEEFOR
VOICEOVERWIRELESSLOCALAREANETWORK WVLANO vIoCicEe mOanVagEeRr IP INTHEIEEE 802.11 WIRELESCSellLulAar nNetSwork
1 XUEMIN SHEN, YI-BING LIN, AI-CHUNG PANG, WLAN HONGQIANGZHAI,C JaIlAl aNgeFnEtNGInWterAneNtG, AANccDessY gUatGewUaAyNGFANG
AND JIANPING PAN Router (RAN)
0.8 Router Router
AP BTS
Resource share00..46 WLAN AP Router AcRcoeusst egratewaRyouter Trunking gatewayPSTN
User 1 4(RA4N)
User 2
0.2 User 3 WLAN voice manager
User 4 6
UUsseerr 56 A CALLADMBTSISSIONCONTROLFRAMEPWhoOneRKFPOhoRne
0 0 P5R0IC1I0N0G150V2O00W25L0A3N00 S35E0R4V00IC4E5S0 T5H00RO550UG60H0 A VOICEOVCeEllRulaWr netLwAorkNS
MICRO-ECOTNimOe (sM) ICFRAMEWORK YIQIAN, ROSEQINGYANGHU, ANDHSIAO-HWACHEN
LEONARDO BADIA, SIMONE MERLIN, ANDREA ZANELLA,
52
AND MICHELE ZORZI
A QUALITY-AWAREVOWLAN ARCHITECTUREANDITS
14
QUANTITATIVEEVALUATIONS
QOS GUARANTEEANDPROVISIONINGATTHE HIROYUKIKOGA, SHIGERUKASHIHARA, YUTAKAFUKUDA,
CONTENTION-BASEDWIRELESSMAC LAYERINTHE KATSUYOSHIIIDA, ANDYUJIOIE
IEEE 802.11EWIRELESSLANS
YANG XIAO ACCEPTED FROM OPEN CALL
60
22
FUTUREPERSONAL“E-PAYMENT”: IRFM
POLLING-BASEDPROTOCOLSFOR
PACKETVOICETRANSPORTOVER PIHUANGANDA.C. BOUCOUVALAS
IEEE 802.11 WIRELESSLOCALAREANETWORKS 68
RAY Y. W. LAM, VICTOR C. M. LEUNG,
AND HENRY C. B. CHAN ORTHOGONALCOMPLEMENTARYCODESFOR
INTERFERENCE-FREECDMA TECHNOLOGIES
30
HSIAO-HWACHEN, HSIN-WEICHIU, ANDMOHSENGUIZANI
IEEE 802.11EENHANCEMENTFORVOICESERVICE
PING WANG, HAI JIANG, AND WEIHUA ZHUANG
EDITOR’S NOTE — 2
2005 EDITORIAL INDEX — 80
Cover illustration: Getty Images
IEEE Wireless Communications • February 2006 1
TLFeBOOK
LYT-EDIT NOTE-FEB 2/2/06 11:02 AM Page 2
MESSAGE FROM THE EDITOR-IN-CHIEF
IEEE W C G D U
IRELESS OMMUNICATIONS OES OWN NDER
D
ear readers, in the professional commu-
As announced in nity, and I have no doubt he
my last editorial, my will continue in the tradi-
term ended last Decem- tion of excellence our maga-
ber, and I am happy to zine has come to enjoy since
introduce the new Editor- its inception. With this I
in-Chief for the magazine, would like to thank all the
Prof. Abbas Jamalipour people who have been work-
from the University of ing with me on the maga-
Sydney, Australia (making zine and all the readers for
this the first time the mag- their support and apprecia-
azine leadership goes to MICHELEZORZI ABBASJAMALIPOUR tion, and hand over the rest
the Southern Hemi- to Abbas.
sphere). He has been on
the Editorial Board of the magazine for many years, Wireless LAN and Voice Communications
and has been a promoter and organizer of a number of
IEEE Wireless Communications since its start in 1994 has
timely and interesting special issues, which I am sure
been on top of the list of telecommunications journals. In
our readers have noticed and appreciated through the 2004 the magazine was ranked fourth for impact among all
years. Abbas is a capable researcher and is very active telecommunications related publications worldwide accord-
IEEE
Director of Magazines W IEEEWireless Communications (ISSN 1536-1284) is pub-
Thomas F. La Porta, Penn. State Univ., USA IRELESS lished bimonthly by The Institute of Electrical and
Electronics Engineers, Inc. Headquarters address: IEEE, 3
Editor-in-Chief
Park Avenue, 17th Floor, New York, NY 10016-5997;tel:
Abbas Jamalipour, Univ. of Sydney, Australia C 212-705-8900; fax: [phone]; e-mail: c.kemelmach-
Senior Advisors OMMUNICATIONS [email protected]. Responsibility for the contents rests upon
Hamid Ahmadi, AT&T Labs, USA authors of signed articles and not the IEEE or its members.
Thomas F. La Porta, Pennsylvania State Univ., USA Unless otherwise specified, the IEEE neither endorses
Mahmoud Naghshineh, IBM, USA nor sanctions any positions or actions espoused in IEEE
Michele Zorzi, University of Padova, Italy Wireless Communications.
Advisory Board 2005 Communications Society Annual subscription: Member subscription: $40 per
Donald Cox, Stanford University, USA year; Non-member subscription:$250 per year. Single copy:
David Goodman, Polytechnic University, USA Board of Governors $50.
Tero Ojanperä, Nokia, Finland Officers Editorial correspondence: Manuscripts for considera-
Kaveh Pahlavan, Worcester Polytech. Inst., USA Nim Cheung, President tion may be submitted to the Editor-in-Chief: Michele
Mahadev Satyanarayanan, CMU, USA Sergio Benedetto,VP–Technical Activities Zorzi,Universita Delgi Studi Di Ferrara, Dipartimento Di
IEEEVehicular Technology Liaison Harvey Freeman, VP–Membership Services Ingegneria, Via Saragat, Ferrara, Italy 1-44100. Electronic
Theodore Rappaport, Univ. of Texas, Austin, USA Byeong Gi Lee, VP–Membership Development submissions may be sent in postscript to: a.jamalipour@
ieee.org.
IEEEComputer Society Liaison Doug Zuckerman, VP–Society Relations
Mike Liu, Ohio State University, USA Curtis A. Siller, Jr., Past President Copyright and reprint permissions: Abstracting is per-
Stan Moyer, Treasurer mitted with credit to the source. Libraries permitted to pho-
Prathima AgTreacwhanl,i Acaulb Eudrnit Uornsiversity, USA John M. Howell, Secretary tuosceo opfy pbaetyroonnsd: tlhimosiets p oosft -U19.S77. Carotipcylersi gthhta tl acawr rfyo ar cpordiev aotne
Sonia Aissa, U. of Quebec, Canada Members-at-Large the bottom of first page provided the per copy fee indicated
Benny Bing, Georgia Tech, USA Class of 2006 in the code is paid through the Copyright Clearance
Anthony C. Boucouvalas, U. of Bournemouth, U.K. T. Scott Atkinson • Stan Moyer Center, 222 Rosewood Drive, Danvers, MA 01923. For
other copying, reprint, or republication permission, write
Jyh-Cheng Chen, Natl. Tsing Hua Univ., R.O.C Naohisa Ohta • Heinrich Stüttgen to Director, Publishing Services, at IEEE Headquarters. All
Hsiao-Hwa Chen, Natl. Sun Yat-Sen Univ., R.O.C. Class of 2007 rights reserved. Copyright © 2006 by The Institute of
Yuguang Fang, Univ. of Florida, USA Robert Blake • Len Cimini Electrical and Electronics Engineers, Inc.
Jennifer Hou, U. of Illinois, USA Jerry Gibson • Larry Greenstein Postmaster: Send address changes to IEEE Wireless
Nei Kato, Tohoku University, Japan Class of 2008 Communications, IEEE, 445 Hoes Lane, Piscataway, NJ
Mario Marchese, University of Genoa, Italy Thomas M. Chen • Andrea Goldsmith 08855-1331; or email to [email protected]. Printed
Archan Misra, IBM Research, USA Khaled Ben-Letaief • Peter J. McLane in USA. Periodicals postage paidat New York, NY and at
additional mailing offices. Canadian GST #40030962.
Giacomo Morabito, U. di Catania, Italy 2005 IEEE Officers Return undeliverable Canadian addresses to:Frontier, PO
Aria Nosratinia, Univ. of Texas, Dallas, USA Michael R. Lightner, President Box 1051, 1031 Helena Street, Fort Eire, ON L2A 6C7.
Tomoaki Ohtsuki, Keio University, Japan Leah H. Jamieson, President-Elect
Subscriptions: Send orders, address changes to: IEEE
Ramachandran Ramjee, Lucent Tech., USA J. Roberto B. de Marca, Secretary Service Center, 445 Hoes Lane, Piscataway, NJ 08855-1331;
Apostolis Salkintzis, Motorola, Greece Joseph V. Lillie, Treasurer tel.: [phone].
YTachoovm Yaasc Woboi,o M, Liucrcoesnotf, tU, USASA JWef.f rCyl eWo.n R Aanyndeesr,s oEnx,e cPuatsitv eP rDesiridecetnotr Apudbvlieshrteirs. iAndgd:rAesds vceorrtriseisnpgo insd aecncceep tteod: Aatd tvheer tdisisicnrge tMioann oafg tehre,
Homayoun Yousefizadeh, U. of Calif, Irvine, USA Mark Karol, Director, Division III IEEE Wireless Communications, 3 Park Avenue, 17th Floor,
Theodore Zahariadis, Ellemedia Techn., Greece New York, NY 10016.
IEEEProduction Staff
Department Editors
Scanning the Literature Joseph Milizzo, Assistant Publisher
Songwu Lu, UCLA,USA Eric Levine, Associate Publisher
Tarik Taleb, Tohoku University, Japan Susan Lange, Digital Production Manager
Spectrum Policy and Reg. Issues Catherine Kemelmacher, Associate Editor ¤(cid:13)
Michael Marcus,Marcus Spectrum Solns., France Jennifer Porcello, Digital Production Associate
2 IEEE Communications Magazine • February 2006
TLFeBOOK
LYT-EDIT NOTE-FEB 2/2/06 11:02 AM Page 3
MESSAGE FROM THE EDITOR-IN-CHIEF
ing to the ISI Journal Citation Report. This great success is deal with various issues regarding QoS control and provi-
the result of the high-quality and hard work of past Editors- sioning, call admission control, architectural design, and
in-Chief Hamid Ahmadi, Tom La Porta, Mahmoud pricing for the transport of voice packets over the wireless
Naghshineh, and Michele Zorzi, the Guest Editors who LAN. I hope you enjoy this issue and join me in congratu-
organized the magazine’s special issues, technical editors, lating Xuemin, Yi-Bing, Ai-Chung, and Jianping for their
reviewers, authors, and, most important, the readers. As of excellent job.
this year I will be taking charge of the magazine with a This issue is also complemented by two articles accepted
great responsibility to maintain its grand position in the from open call. The article “Future Personal ‘e-Payment’:
field of telecommunications. I would like to thank Michele IrFM” by Huang and Boucouvalas describes a new paradigm
for the past three years when I was working with him as a for payment with mobile devices, Infrared Financial Mes-
technical editor and particularly in the past several months saging (IrFM), which promises to change the way transac-
when I worked very closely with him through the important tions are made. The article “Orthogonal Complementary
transition stage. I have learned a lot from Michele, and I Codes for Interference-Free CDMA Technologies” by
hope with his guidance and the help of the magazine’s tech- Chen, Chiu, and Guizani discusses new code design, using
nical editorial team I can deliver the magazine to its best the newly proposed REAL approach, for interference-free
possible position during my term. One important factor in CDMA channels for applications in next-generation mobile
the magazine’s success in the past was its role in delivering networks. These two articles provide additional flavors to
timely special issues to the research community. My first the topics presented in the special issue. I hope you find
goal will be to continue soliciting important topics for the them useful too.
magazine. This first issue of 2006 is a good example, bring-
ing timely samples of research papers on the topic of voice INFUTUREISSUESOF
over wireless LAN. IEEE WIRELESSCOMMUNICATIONSMAGAZINE
Wireless communications is headed in its new direction
•INTER-VEHICULARCOMMUNICATIONS
of more data delivery with the development of advanced
coding and network technologies. Quality of service provi- •MULTIMEDIAINWIRELESS/MOBILEADHOCNETWORKS
sioning to real-time applications still remains a challenge •RADIORESOURCEMANAGEMENTAND
for the wireless channel due to mobility of users and fluctu- PROTOCOLENGINEERINGIN
ations in channel conditions. Despite these obstacles, FUTUREIEEE BROADBANDWIRELESSNETWORKS
researchers try to develop new ways to transport voice traf-
fic over the wireless channel, not by its
traditional method but through the
use of the IP network. The global
Internet provides an inexpensive and
ubiquitous means of delivery for voice
irrespective of the physical distance
between end users, and wireless com-
munications needs to get involved in
this important field. Wireless LAN, as
an important player in future wireless
data networks, has been widely investi-
gated to incorporate with the networks
and other broadband wireless services
such as WiMax. It is now time that its
advantageous characteristics initiate
the wireless LAN’s role in transmitting
voice over the Internet. This special
issue, organized by Xuemin Shen, Yi-
Bing Lin, Ai-Chung Pang, and Jian-
ping Pan, presents some research
results related to the implementation
of voice over IP via the wireless LAN.
A major difficulty in such an arrange-
ment is the fact that the conventional
wireless LAN does not support QoS.
The channels are accessed by users in
a competitive pattern, which subse-
quently does not provide any provi-
sioning to voice traffic and its delay
sensitivity. As either an extended seg-
ment of the wired Internet or part of a
cellular network in an integrated wire-
less LAN/cellular system, or any other
configuration, the wireless LAN needs
to overcome its QoS inefficiencies. As
detailed in the guest editorial and the
articles themselves, the seven articles
IEEE Communications Magazine • February 2006 3
TLFeBOOK
LYT-GUEST EDIT-Shen 2/2/06 11:03 AM Page 4
GUEST EDITORIAL
V W L A N
OICE OVER IRELESS OCAL REA ETWORK
XUEMINSHEN YI-BINGLIN AI-CHUNGPANG JIANPINGPAN
T
he market for wireless local area network (WLAN) has try in networking, wireless communications, and mobile
been experiencing tremendous growth in recent years, computing, with the goal of fostering interaction among
as evidenced by the fast increasing popularity of them. It is timely and valuable for those involved in the research
WLAN hotspots deployed in residence, enterprise, and of VoWLAN.
public areas such as airports, campuses, conference venues, In the first article “Pricing VoWLAN services through a
shopping malls, and exhibitions. Meanwhile, WLAN ser- Micro-Economic Framework,” Badia et al.investigate the issue
vices are evolving from best effort data services to real-time of determining an appropriate pricing strategy for voice-
applications with a certain level of quality of service (QoS) pro- over-WLAN provisioning. The voice services are framed in
visioning. In the near future, voice over IP (VoIP) is expect- a tunable QoS scenario. Analysis is then performed with the
ed to be extended from the Internet to the wireless domain awareness that in a WLAN system the tariff payment deter-
via WLAN. An 802.11-based phone system is generally less mines price-based access regulation. The authors also apply
expensive to install and support than a wired system. A sig- a micro-economic framework which considers the trade-off
nificant benefit of mixing telephone traffic with data on a WLAN between perceived QoS and paid price in the users’ request.
is that it can support user mobility and make use of a com- A contention-based medium access mechanism is a manda-
mon infrastructure. tory part of the IEEE 802.11 standard series. In the second arti-
However, unlike cellular networks where dedicated chan- cle, “QoS Guarantee and Provisioning at the Contention-Based
nels are assigned to voice traffic, the voice packets in WLAN Wireless MAC Layer in IEEE 802.11e Wireless LANs,”
are multiplexed with other data traffic. The distributed coor- Xiao provides a survey of previous work on QoS provision-
dination ffunction (DCF) is the primary operation mode of ing with contention-based medium access control (MAC). Dis-
WLAN, which introduces significant delay jitter. Investiga- tributed admission control, data control mechanisms, bandwidth
tion of voice packet delay/jitter performance over 802.11/DCF allocation, and handoff/roaming are discussed. Further research
mode is a necessary and challenging problem. The 802.11e group directions are also pointed out.
is currently working on QoS support in WLAN, but the rati- The next two articles focus on polling, which is an effec-
fication of the standard has a long way to go. Until then, tive solution for supporting real-time traffic over WLANs
development of VoWLAN QoS proprietary schemes is with stringent delay requirements. In the article “Polling-Based
essential to prioritize traffic on the network for QoS support for Protocols for Packet Voice Transport over IEEE 802.11
real-time traffic. In addition, it is necessary that voice over WLAN Wireless Local Area Networks,” Lam et al.present a survey
has handover capability across WLANs and/or hybrid of different polling-based protocols for supporting VoWLAN.
WLAN-cellular networks. Improved WLAN security, WPA, Three key issues are discussed: managing a polling list,
and WPA2 introduce unacceptably high delays for VoWLAN determining the polling sequence and reducing polling
during session switching, and 802.11r is going to address the overhead. An isochronous control function (ICF) is also
issue. The next-generation wireless access technology (NGWA), proposed. In the article “IEEE 802.11e Enhancements for
such as WiMax and its mobility enhancements 802.16e and IEEE Voice Service,” Wang et al.propose mechanisms to enhance
802.20, is above the horizon. The new technology and stan- IEEE 802.11e with voice QoS provisioning capability. The delay
dard migration and investment protection should also be requirement of real-time voice is guaranteed by controlled
considered. channel access, while bandwidth efficiency can be greatly
This special issue is to disseminate the state-of-the-art R&D improved by overhead suppression and statistical multiplex-
results on VoWLAN, facilitate the deployment of VoWLAN, ing. A call admission control scheme is also presented to
and bring together researchers from both academia and indus- admit voice stations into the system with QoS guarantee.
4 IEEE Wireless Communications • February 2006
TLFeBOOK
LYT-GUEST EDIT-Shen 2/2/06 11:03 AM Page 5
GUEST EDITORIAL
Admission control is essential to QoS provisioning of than 200 papers and book chapters in wireless communications and net-
VoWLAN. In the article “Providing Statistical QoS Guar- works, control, and filtering. He was Technical Co-Chair for IEEE GLOBECOM ’03,
ISPAN ’04, QShine ’05, IEEE Broadnets ’05, and WirelessCom ’05, and is Spe-
antee for Voice over IP in IEEE 802.11 Wireless LANs,” cial Track Chair of the 2005 IFIP Networking Conference. He serves as Associ-
Zhai et al.propose a call admission control scheme that ate Editor for IEEE Transactions on Wireless Communications; IEEE
Transactions on Vehicular Technology; Computer Networks; ACM/Wireless
runs over the MAC layer to support VoIP services. Voice
Networks; Wireless Communications and Mobile Computing(Wiley); and
traffic is regulated to efficiently coordinate medium contention International Journal Computer and Applications. He has also served as
among voice sources. Non-voice traffic is regulated by a Guest Editor for IEEE JSAC, IEEE Wireless Communications, and IEEE Commu-
nications Magazine. He received the Premier’s Research Excellence Award (PREA)
rate control mechanism to control its impact on the perfor-
in 2003 from the Province of Ontario, Canada, for demonstrated excellence
mance of voice traffic. Statistical QoS guarantee for voice of scientific and academic contributions, and the Distinguished Performance
traffic is provided, and a stable high throughput is main- Award in 2002 from the Faculty of Engineering, University of Waterloo, for
outstanding contributions in teaching, scholarship, and service.
tained for non-voice traffic at the same time. In the article
“A Call Admission Control Framework for Voice over WLANs,” YI-BINGLIN[F] ([email protected]) is chair professor of the Department
Qian et al.present a call admission control framework of Computer Science and Information Engineering (CSIE), National Chiao
Tung University (NCTU). He also serves as vice president of the Office of
called WLAN Voice Manager. The WLAN Voice Manager
Research and Development, NCTU. He is a senior technical editor of IEEE
interacts with WLAN MAC layer protocols, soft switches (VoIP Network, an editor of IEEE Transactions on Wireless Communications, an
call agents), routers, and other network devices to perform end- associate editor of IEEE Transactions on Vehicular Technology, an editor of
IEEE Wireless Communications, an editor of ACM/Baltzer Wireless Networks, Pro-
to-end (ETE) QoS provisioning and control for VoIP calls orig- gram Chair for the 8th Workshop on Distributed and Parallel Simulation,
inated from WLANs. By implementing the proposed General Chair for the 9th Workshop on Distributed and Parallel Simulation,
Program Chair for the 2nd International Mobile Computing Conference,
WLAN Voice Manager in the WLAN access network, a
Guest Editor for the ACM/Baltzer MONETspecial issue on Personal Commu-
two-level ETE VoIP QoS control mechanism can be nications, a Guest Editor for IEEE Transactions on Computers special issue on
achieved. Mobile Computing, a Guest Editor for IEEE Transactions on Computersspe-
cial issue on Wireless Internet, and a Guest Editor for IEEE Communications
In the final article, “Quality-Aware VoWLAN Architecture
Magazinespecial issue on Active, Programmable, and Mobile Code Network-
and Its Quantitative Evaluations,” Koga et al.propose a ing. He is co-author of the book Wireless and Mobile Network Architecture(with
media optimization network architecture (MONA). An Imrich Chlamtac; Wiley). He received 1997, 1999, and 2001 Distinguished Research
Awards from National Science Council, ROC; 2004 K.-T. Li Outstanding
intercarrier handover mechanism based on a cross-layer Award; 2005 Pan WY Distinguished Research Award; 2005 Teco Award; and
approach using the number of layer 2 retransmissions is the 1998 Outstanding Youth Electrical Engineer Award from CIEE, ROC. He
also received the NCTU Outstanding Teaching Award in 2002. He is an
evaluated. Voice quality management in WLAN is also dis-
adjunct research fellow of Academia Sinica, and an adjunct chair professor
cussed. of Providence University. He is an ACM Fellow, an AAAS Fellow, and IEE Fel-
In closing, the guest editors would like to thank all the authors low.
who responded to the Call for Papers, regardless of whether their
AI-CHUNPANG([email protected]) received B.S., M.S., and Ph.D.
papers have been included in this issue or not due to space degrees in computer science and information engineering from National
limitations. The editors would also like to acknowledge the Chiao Tung University, Taiwan, in 1996, 1998, and 2002, respectively. She joined
the Department of Computer Science and Information Engineering (CSIE), Nation-
contribution of many experts who participated in the review pro-
al Taiwan University (NTU), Taipei, Taiwan, as an assistant professor in 2002.
cess, and provided helpful and valuable suggestions to the authors From August 2004 to July 2005 she served as an assistant professor at the
on improving the content and presentation of the articles. Graduate Institute of Networking and Multimedia (INM), and an adjunct
assistant professor at CSIE/NTU, Taipei, Taiwan. Currently, she is an associate
The advice and support of Dr. M. Zorzi, Editor-in-Chief of IEEE
professor with INM and CSIE of NTU, Taipei, Taiwan. Her research interests include
Wireless Communications, and the assistance of Sue Lange design and analysis of personal communications services networks, mobile com-
are greatly appreciated. puting, voice over IP, and performance modeling.
BIOGRAPHIES JIANPINGPAN[M] ([email protected]) is currently an assistant professor of computer
science at the University of Victoria, British Columbia, Canada. He received
XUEMIN(SHERMAN) SHEN[M’97, SM’02] ([email protected]) received a his B.S. and Ph.D. degrees in computer science from Southeast University,
B.Sc. (1982) degree from Dalian Maritime University, China, and M.Sc. Nanjing, China, in 1994 and 1998, respectively. From 1999 to 2001 he was a
(1987) and Ph.D. (1990) degrees from Rutgers University, New Jersey, all in postdoctoral fellow and then research associate at the University of Water-
electrical engineering. Currently, he is with the Department of Electrical and loo, Ontario, Canada; from 2001 to 2005 he was a member of research staff
Computer Engineering, University of Waterloo, Canada, where he is a pro- at Fujitsu Labs and a research scientist at NTT MCL in Silicon Valley, Califor-
fessor and associate chair for graduate studies. His research focuses on nia. His area of specialization is distributed systems and networks, and his
mobility and resource management in interconnected wireless/wireline net- recent research interests include protocols for advanced networking, perfor-
works, UWB wireless communications systems, wireless security, and ad hoc and mance analysis of networked systems, and applied network security. He is a mem-
sensor networks. He is a co-author of two books, and has published more ber of the ACM.
IEEE Wireless Communications • February 2006 5
TLFeBOOK
BADIA LAYOUT 2/2/06 11:04 AM Page 6
VOICE OVER WIRELESS LOCAL AREA NETWORK
P V WLAN S
RICING O ERVICES THROUGH A
M -E F
ICRO CONOMIC RAMEWORK
LEONARDO BADIA, UNIVERSITY OF FERRARA
SIMONE MERLIN, ANDREA ZANELLA, AND MICHELE ZORZI, UNIVERSITY OF PADOVA
ABSTRACT neers is to improve the transmission efficiency
for VoWLAN [2] to obtain reliable service. This
This article investigates the issue of determin- impacts on the economic aspects, since enhanced
ing an appropriate pricing strategy for voice over quality of service (QoS) might imply higher com-
WLAN provisioning. This is performed by first mercial value of the service and, consequently,
framing the voice services in a tunable QoS sce- larger revenues for the operator in the long run.
nario. The analysis is then performed with the On the other hand, VoWLAN provisioning can
awareness that in a WLAN system the tariff pay- be faced as a distributed allocation, a competi-
ment determines price-based access regulation, tive game, and a constrained optimization prob-
which implies a different service perception lem, which are concepts derived from economics
because of the modified network conditions. For [3]. Finally, another economic aspect concerns
this reason, we apply a micro-economic frame- the design of appropriate pricing strategies for
work that considers the trade-off between per- provisioning such a service that are complicated
ceived QoS and paid price in the users’ request by the difficulties in capturing the capacity of a
(i.e., including in a tunable way both service WLAN.
The simplicity of
requirements and willingness to pay). This allows The investigation of effective pricing strate-
connection and the us to investigate the provider’s task of having a gies for telecommunication services has been
suitable price policy, both identifying the subse- widely studied mainly for Internet-like environ-
low cost of Wireless quently involved trade-offs and providing insight ments in the wireline case [4]. This has shown,
on how to efficiently cut them. for example, the difficulty of identifying a good
Local Area Network tariff function that, besides generating proper
INTRODUCTION revenue, is both easy to understand and appreci-
cards have led to a
ate by users, and efficient in terms of regulating
The simplicity of connection and the low cost of access to the service. Indeed, this is also a prob-
large diffusion of
wireless local area network (WLAN) cards have lem for services like VoWLAN, with the addi-
devices provided led to a large diffusion of devices provided with tional limitation of system capacity, which is
WLAN connectivity, especially based on the inherently difficult to treat analytically. Different
with WLAN IEEE 802.11b standard. In this article we refer from other kinds of networks, the medium access
interchangeably to WLANs and IEEE 802.11b- control (MAC) strategy adopted in WLANs is
connectivity, based networks, using the distributed coordina- based on the carrier sense multiple access with
tion function (DCF). WLAN connectivity, at collision avoidance (CSMA/CA) mechanism that
especially based on
first limited to laptops, is now migrating toward does not provide QoS and/or fairness guarantees
palmtops and even mobile phones. Following to users. Also, the capacity is not a well defined
the IEEE 802.11b
such a diffusion of WLAN devices, the number quantity, since it depends on many factors relat-
standard. of hotspots is also rapidly increasing, and the ed to channel conditions and traffic patterns.
services offered are going to cover a broad set of Thus, the service provided to users depends on
applications. the actual working conditions of the systems and
Although WLAN systems have mainly been the admission control policies possibly imple-
designed to carry best effort traffic, it is expected mented by the system manager.
that such systems will soon be required to also In our view, such difficulties are probably one
carry many types of audio, video, and multime- of the reasons for the lack of significant contri-
dia traffic. In particular, voice over Internet Pro- butions concerning VoWLAN pricing. Such an
tocol (VoIP) [1] is gaining relevance in the issue, indeed, is generally washed out by relying
wireless environment, opening the way to the on the large basic potential of the WLAN, with-
appealing voice over WLAN (VoWLAN) sce- out any consideration of the efficiency of system
nario. management, the dimensioning of economic
The analysis of VoWLAN is an important quantities, and the access control capability of
aspect for researchers in both engineering and price-based regulation. On the contrary, we
economics. On one hand, the challenge for engi- believe that it is essential to consider these
6 1536-1284/06/$20.00 © 2006 IEEE IEEE Wireless Communications • February 2006
TLFeBOOK
BADIA LAYOUT 2/2/06 11:04 AM Page 7
aspects as well in order to achieve larger and
larger diffusion of VoWLAN services, and obtain Voice user
economic benefits.
Therefore, we aim here at discussing a design
Service Network Revenue
methodology that might be a useful instrument
request capacity
for a provider operating VoWLAN services. In
more detail, we capture the allocation problem
in a micro-economic framework, where users’
QoS perception and willingness to pay are Simulation
described by means of mathematical instru-
ments. Furthermore, we specifically take into
account the CSMA/CA capacity of the WLAN. Perceived QoS
The whole procedure is aimed not only at inves-
tigating certain specific tariff strategies, but
more widely at identifying the trade-off between
the objectives of users and provider, in order to Network manager
User
give an overview of how a network manager Acquired
preferences
might set up pricing for VoWLAN service.
This framework relies on two separate parts.
First of all, the specific characteristics of the
WLAN capacity must be accounted for. In order
to have a better treatable model, we also intro-
duce some assumptions about QoS and fairness
among users, which are discussed in the follow- Price
ing section. Second, we also need to model the
users’ perception of the service through a micro-
economic framework for wireless systems. This is nFigure 1.The estimation scheme and the micro-economic model application
needed since pricing strategies impact not only to VoWLAN service.
the regulation of service demand, as in classic
economic models, but also network operating
conditions, thus indirectly affecting the QoS per- emphasis and are discussed in the following sec-
ceived by users. tions. In particular, as explained later, our
For this latter part, we make use of the model approach in this article is to investigate network
presented in [5], which considers the trade-off capacity by means of simulation, which probably
between QoS and pricing by defining a probabil- represents an easier and faster way for a net-
ity measure of user satisfaction. The allocation is work manager to investigate WLAN capacity.
thus evaluated with respect to a specific perfor- Clearly, approaches based on analytical evalua-
mance metric, such as generated revenue or tions are possible as well, although the simplifi-
number of users provided with satisfactory QoS, cations usually assumed by such models may
which is weighted with the user satisfaction conceal some of the complex interactions that
probability. This might be regarded as an implic- might arise in reality.
it admission control mechanism, since we may The rest of this article is organized as follows.
think of unsatisfied users as rejecting the service In the next section we discuss WLAN capacity
conditions and therefore not taking part in the and explain the way in which soft QoS might be
system over the long term. required by VoWLAN users and also perceived,
Finally, these two components might be inte- depending on limitations in system capacity. We
grated by means of a micro-economic scheme as discuss an economic model in order to describe
follows. Investigation of the pricing policy is pos- how this impacts users’ choices, also driven by
sible given that the following preliminary evalua- the pricing policy chosen by the operator. Final-
tions have been performed: ly, we discuss with a numerical example how this
•QoS requirements of users. This can be done can be useful for the provider of the WLAN in
via subjective testing of the VoWLAN service. order to identify a suitable tariff plan that both
The result of this estimation will be numeri- matches users’ QoS requirements and generates
cally treated in the following as a user utility adequate revenue.
function. The provider might even store a
user’s profile that describes the utility parame- QOS PROVISIONING OVER WLAN
ters specific to each user.
•Sensitivity of users to price and utility varia- The IEEE 802.11b standard does not encompass
tion. This evaluation impacts the satisfaction any QoS mechanism for supporting transmission
probability, and might be performed via exten- of voice or, in general, multimedia contents. The
sive sampling. channel access is ruled by a contention-based
•Expected number of incoming users and dura- strategy, DCF, which does not natively provide
tion of calls, in order to evaluate the aggre- any service differentiation mechanism. Limited
gate performance of multiple connections. service differentiation might be provided by
It is important to observe that due to the using the point coordination function (PCF) to
nature of the VoWLAN service, the whole sys- rule channel access. However, the PCF function-
tem is on a feedback loop, where all the values ality is defined as optional in IEEE 802.11b and
mentioned above impact each other. A scheme is typically not implemented in commercialized
to represent this behavior is outlined in Fig. 1. IEEE 802.11 devices.
Several aspects of this figure are worth more Problems arising from distributed sharing of
IEEE Wireless Communications • February 2006 7
TLFeBOOK
BADIA LAYOUT 2/2/06 11:04 AM Page 8
original request, thus providing a generalized
1 best effort quality.
Even though a detailed study of a service
differentiation mechanism for IEEE 802.11b is
0.8 beyond the scope of the present article, in the
following we consider a specific strategy for
the purpose of our investigation. For the sake
e
ar 0.6 of simplicity, we refer to the simplest tech-
h
e s nique of those proposed in [8], which, without
urc substantial modifications to the standard, acts
eso 0.4 on the length of generated packets . Since
R
DCF provides equal access probability (in the
User 1
User 2 long term) for all the contending nodes
0.2 User 3 (including the access point), under the assump-
User 4
User 5 tion that all the nodes in the cell served by an
User 6 access point work with the same physical rate,
0
0 50 100 150 200 250 300 350 400 450 500 550 600 the resource share gained by each node is thus
proportional to the length of the transmitted
Time (s)
packets. Note also that more complex and
nFigure 2.Proportional share of resource for a WLAN system with six users realistic service differentiation mechanisms
with increasing priority. This confirms and extends to a larger system the can be defined, but we stress that our approach
results already obtained by [8]. can be kept almost unchanged since it simply
involves evaluation of the perceived service,
and hence does not rely critically on a given
available resources are the lack of throughput differentiation mechanism.
and maximum access delay guarantees [6]. For Following this resource sharing strategy,
light traffic, the DCF mechanism is generally when the network approaches the saturation, the
able to satisfy the packet rate and delay require- average transmission rate obtained by each user
ments for voice applications. However, when the results to be proportional to its packets length.
allocated traffic approaches system capacity, the Furthermore, the ratio between the traffic of any
access delay may result in total inadequacy for two users will be the same both in the non satu-
support of multimedia and voice services. The rated and saturated case. Our evaluation with
problem is exacerbated by the coexistence of the ns2-simulator [9] confirms the results shown
voice and data traffic, since the transmission of in [8] over a wider population range. Figure 2
long data packets may worsen the delay and jit- shows, for example, the results of the subsequent
ter of voice packets. allocation of up to 6 users in the WLAN sce-
These impairments have direct effects on the nario, where a new user is added every 100 sec-
perceived voice quality, which will gradually onds. The resource requirements of each user
decrease as the system approaches the saturation are subsequently increasing, so that the rate
condition, with a sensitivity depending on the request of the i-th user, i > 1, is i times the
specific codec adopted. request of the first one. In other words, user 2,
Nonetheless, different quality levels can be 3, … requires a rate equal to 2, 3, … times the
obtained by using adaptable codecs. Such codecs request of user 1. As shown in the figure, the
allow for adjusting the service quality and correctness of the assumption of proportionally
amount of required resources, according to user fair share of resource holds.
requests or available resources [7]. Notice that, We notice that the AP will get just the same
in general, scaling of service quality is obtained amount of resources as a generic user, even
by relaxing the requirements in terms of band- though the AP has to serve all the downlink traf-
width or delay. fic that is generally much higher than the uplink
We might therefore focus on a VoWLAN flow generated by each terminal. In particular,
scenario, where users require different service with voice connections the AP has to serve as
levels, according to their personal preferences many inbound flows as the number of active
and the price paid for the service [5]. As long as voice sessions, thus rapidly becoming the bottle-
the total requested rate is below the network neck of the system [10]. However, solutions to
capacity, the allocation does not present prob- alleviate this problem and consequently increase
lems, and all users are entitled to obtain the the voice capacity of WLANs have recently
requested rate (i.e., to experience the desired appeared in the literature [11].
QoS). However, as the total requests approach Finally, we remark that in the saturation case
the available capacity, the rate experienced by the proportionally fair share of resources can
users has to be reduced in order to maintain the determine an unsatisfactory allocation for some
packet delay within the limits required for the users. In order to investigate this point in detail,
voice/video service to be acceptable. The policy we introduce a micro-economic model, with the
used for subdividing the available resources goal of quantifying the wireless resource alloca-
among users can greatly impact the QoS per- tion, which is described in the next section.
ceived by the users, determining their behavior. Under this framework, specific evaluation of
We will assume a rate scaling policy that assigns VoWLAN service provisioning is performed and
to each user a fraction of the available resources useful insights are given, in particular concerning
proportional to the requested rate (i.e., service the design of an appropriate pricing strategy,
level). In this way, the degradation of service capable of both coordinating users’ requests and
experienced by each user is proportional to its achieving adequate provider revenue.
8 IEEE Wireless Communications • February 2006
TLFeBOOK
BADIA LAYOUT 2/2/06 11:04 AM Page 9
MICRO-ECONOMIC
1
RESOURCE ASSIGNMENT MODEL
Changing
the slope (ζ)
The offer of a VoWLAN service has a side-
0.8
effect on the network manager. In order to reach
a sustainable business model, the service
provider must achieve adequate revenue from
tnhuem nbeetrw oofr cko onpneercatitoionns,. aHnedn tchee, tmhoer hei rgehseoru trhcee u (r) 0.6
y
dreevmenauned. iHngo wtheevye ra, rper,i ctihneg hanigdh aelrl otchaet ipono tsetnrattiea-l Utilit 0.4 Shifting
gies heavily influence the behavior of the users. the curve (K)
In fact, users who are faced with inadequate
QoS or very high price are likely dissatisfied. For
these reasons, it is important to also incorporate 0.2
economic considerations into the analysis. ζ = 6, K = 0.12
To describe the allocation of soft QoS, many ζ = 13, K = 0.12
ζ = 6, K = 0.41
researchers, even in the telecommunication field, 0
have proposed employing utility functions [3]. 0 r0 r1
These are micro-economic instruments intended Rate r
to map the welfare of purchasing a good or ser-
vice. The capability of utility functions to capture nFigure 3.Sigmoid utilities chosen to represent the QoS perceived by the users
the QoS evaluation by users of a wireless net- as associated with different rates.
work becomes even more important when eco-
nomic aspects such as service pricing are
considered. This is especially needed for WLAN factors such as delay or delivery ratio (which are,
systems, where it has been shown that resource however, correlated with the rate). Nevertheless,
pricing is also a way to implicitly regulate the this would lead to more complex expressions
access of the terminals [12]. without affecting the method here proposed.
Therefore, our investigations about micro- From a mathematical point of view, we adopt
economics of WLANs are made through the the following descriptions, proposed in [5]: both
adaptation to this particular scenario of the utility and pricing functions (u(r) and p(r),
model presented in [5]. The very general idea respectively) increase as the allocated resource
adopted in this framework is that users’ service increases. However, whereas the price might
perception is determined by the trade-off indefinitely increase as the allocation becomes
between utility (representing the perceived QoS) larger, the utility must saturate after a certain
and price. Formally, this consists of evaluating a point. In particular, we assume that the rates
numerical value for each user measuring satis- available at the codec span over an interval
faction probability, which increases as utility [r ,r ].
0 1
increases and/or price decreases. The utility u(r) might be obtained, as previ-
This value is then used in order to weigh all ously discussed, through subjective testing. This
performance metrics, since from the economic results in a different utility function for each
point of view an efficient resource usage not user; also, since the set of rates is discrete, this
only aims at satisfying the users as much as pos- would determine a stepwise constant function.
sible, but also evaluates the performance mostly However, for the mathematical approach, we
on satisfied users. Metrics like throughput and assume that u(r) can be approximated with a
even more the provider’s revenue obtained from continuous function. This is reasonable if the
service provisioning are not meaningful if the number of available rates is large enough; and,
associated service appreciation is low. If a peak in any case, the analysis presented in the follow-
allocation is obtained without respecting users’ ing might be replaced by taking a quantized ver-
requests, it is likely that the users will abandon sion of the utility values. Moreover, utilities are
the service on the long run. This is why the eco- assumed to be normalized between 0 and 1, and
nomic model we employ considers only satisfied for every user we impose that u(r) is always 0
users (or better, it weighs the performance met- when r< r and 1 when r> r .
0 1
rics with the satisfaction probability of users) as Formally, the utilities we consider are sig-
concurrent with the evaluation of allocation effi- moid-shaped functions, where the middle point
ciency. and curvature are regulated by means of numeri-
According to the assumptions made, both cal parameters, indicated with Kand ζ, respec-
utility and pricing, and hence users’ satisfaction, tively. This means we consider u(r) as follows:
depend on the allocated resource, which we
describe through the average transmission rate r u(r)=
otrhabtetea dirnifiesfd ed rbeeytne atri mautsiineoren.d W mbyee ctrhheeam nlaeirsnkmg tt hhcoa otn fas ticdhceeo rrpedadinc,k gte httoes  K1−ζrr1−−rr00 ζ for r0≤r≤r0+K(r1−r0))(1)
tarcatnusaml isttiegdn aolvienrg t hraet we iaretl ethsse c phhanysniecla, lw lhaiylee rth ies 1−(1−K)1−ζrr1−−rr00 for r0+K(r1−r0)≤≤r≤r1
assumed to be equal for all users and constant
over time. Note that it would be possible to where 0 < K < 1 and ζ > 1 are adjustable
extend this approach by also considering other parameters. Note that keeping track of specific
IEEE Wireless Communications • February 2006 9
TLFeBOOK