Top Banner
187
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Mobile Computing

Team-Fly®

Page 2: Mobile Computing

ANY TIME, ANYWHERECOMPUTING

Mobile Computing Concepts andTechnology

Page 3: Mobile Computing

The Kluwer International SeriesIn Engineering and Computer Science

Page 4: Mobile Computing

ANYTIME, ANYWHERECOMPUTING

Mobile Computing Concepts andTechnology

Abdelsalam HelalUniversity of Florida

Gainesville, Florida, USA

Bert HaskellMCC

Austin, Texas, USA

Jeffery L. CarterMotorola

Austin, Texas, USA

Richard BriceDarrell Woelk

Marek RusinkiewiczMCC

Austin, Texas, USA

KLUWER ACADEMIC PUBLISHERS NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW

Page 5: Mobile Computing

eBook ISBN: 0-306-47301-1Print ISBN: 0-792-38610-8

©2002 Kluwer Academic PublishersNew York, Boston, Dordrecht, London, Moscow

All rights reserved

No part of this eBook may be reproduced or transmitted in any form or by any means, electronic,mechanical, recording, or otherwise, without written consent from the Publisher

Created in the United States of America

Visit Kluwer Online at: http://www.kluweronline.comand Kluwer's eBookstore at: http://www.ebooks.kluweronline.com

Page 6: Mobile Computing

CONTENTS

FOREWORD

PREFACE

xi

xiii

1

2

3

INTRODUCTION TO MOBILECOMPUTING1.11.2

1.3

Impressive TechnologyWireless and Mobile Computing ArchitectureLimitations of the Wireless and Mobile Environment 10

124

WIRELESS TELECOMMUNICATIONNETWORKS2.1

2.2

2.32.4

Digital Cellular Systems

2.1.12.1.2

Time-Division Multiple Access (TDMA)

Code-Division Multiple Access (CDMA)

The Wireless Network Technology

2.2.12.2.2

2.2.32.2.4

2.2.5

2.2.6

In-room InfraredIn-room Radio FrequencyIn-building Radio FrequencyCampus/Metropolitan Area Packet NetworksWide-Area Packet/Circuit Switched Data NetworksSatellite Networks

Mobility-Bandwidth Tradeoffs

Systems Issues

2.4.1 Multimedia Applications

1315151618181919202021212225

27PORTABLE INFORMATION APPLIANCES

Page 7: Mobile Computing

v i A N Y T I M E , A N Y W H E R E C O M P U T I N G

3.13.23.3

3.4

3.5

3.63.73.83.9

Historical EvolutionThe Advent of the PDAPalmtop Computers

3.3.1 The Palm PilotHand-held Computers3.4.1 Sharp Power Zaurus

VADEM ClioCommunicators

3.5.13.5.2

Nokia 9000

Motorola Marco

Sub-notebooks (Micro-notebooks)NotebooksLaptopsOther Information Appliances

3.9.13.9.2

HP CapShare

Clarion AutoPC

4

5

FUTURE INFORMATION APPLIANCES4.14.2

4.3

New Challenges

Emerging Portable Information Appliances and Teleservices

4.2.7

Wearable Computing (MIT)

Wearable Computer Systems (CMU)IBM Wearable PC

BodyLAN: A Wearable RF Communications SystemToshiba Desk Area Network (DAN)BlueTooth

Seiko Wristwatch PCNTT PHS Wristwatch PhoneNTT Ring Keyboard

4.2.104.2.11

4.2.12

Display Pad: The Next Generation TVThe Ear Phone

Power Ring and the Magic Wand

Concluding Remarks

FUTURE WIRELESS COMMUNICATIONNETWORKS

27323536373739394041424444464647

49495152525353555555565758596061

63

4.2.14.2.24.2.3

4.2.64.2.54.2.4

4.2.94.2.8

3.4.2

Page 8: Mobile Computing

Contents vii

5.1

5.2

5.3

5.45.5

Future Wireless Teleservices

5.1.15.1.25.1.3

Wireline Network Services

Wireless Service EvolutionMarket Evolution

Emerging Wireless Network Standards5.2.1

5.2.2

5.2.3

IMT-2000

UMTSACTS

Third Generation Wireless Networks

5.3.15.3.2

5.3.3

Time Division/Code Division Multiple Access

Wideband Code Division Multiple Access

Space Division Multiple Access

Fourth Generation Wireless ResearchConcluding Remarks

6

7

STATE OF INDUSTRY: MOBILITYSUPPORT SOFTWARE6.16.2

Competing Philosophies

End-User Client Applications6.2.16.2.26.2.36.2.46.2.5

Oracle Mobile Agents

Oracle Lite

Oracle Software ManagerOracle Replication ManagerSybase SQL Remote

6.3

6.46.5

Mobility Middleware6.3.16.3.2

MobileWare Office Server

Shiva PPPInteroperability and StandardizationShortcomings and Limitations

RESEARCH IN WIRELESS AND MOBILECOMPUTING7.1 Mobile Networking

7.1.17.1.2

7.1.3

Early Approaches: Virtual IP ProtocolsLoose Source Routing ProtocolThe Mobile Internet Protocol (Mobile-IP)

6464

656667676971727477838687

89899191919292929393949597

99100100101102

Page 9: Mobile Computing

viii ANY TIME, ANYWHERE COMPUTING

7.1.47.1.57.1.6

Cellular Digital Packet Data (CDPD)

The GSM General Packet Radio Service (GPRS)Security and Authentication Issues in Mobile Net-works

7.2

7.3

7.4

7.5

7.6

Quality of Service in Mobile Networks

Optimizing TCP/IP for Mobile NetworksQoS Driven, High-Level Communication Protocols

QoS Driven, Full Protocol Stacks

Mobile Access to the World Wide Web

7.3.1

7.3.5

The Wireless WWW (W4)Dynamic DocumentsDynamic URLs

Mobile Browser (MOWSER)WebExpress

Mobile Data Management7.4.1 Mobile Client/Server Data Access

Mobile Data Access in Ad-hoc Networks

Mobile Transactions

7.5.1

7.5.3

7.5.57.5.6

Reporting and Co-Transactions

The Kangaroo Transaction Model

The Clustering ModelIsolation-Only TransactionsThe Two-tier Transaction ModelSemantic-based Nomadic Transaction Processing

Mobile Computing Models

7.6.17.6.27.6.3

The Client/Server ModelThe Client/Proxy/Server ModelThe Disconnected Operation Model

The Mobile Agent Model

The Thin Client Model

APPENDIX-A: GLOSSARY OF COMMONABBREVIATIONS

APPENDIX-B: WIRELESS CELLULARSYSTEMS

106107

108109110112114119119119120120121123123125126126127128129129130131131132133133134

137

145

7.2.17.2.27.2.3

7.3.27.3.37.3.4

7.4.2

7.5.2

7.5.4

7.6.4

7.6.5

Page 10: Mobile Computing

Contents ix

APPENDIX-C: STANDARDS ORGANIZATIONS 151

REFERENCES 157

INDEX 165

Page 11: Mobile Computing

This page intentionally left blank.

Team-Fly®

Page 12: Mobile Computing

FOREWORD

Mobile communications have permeated the globe in both business and socialcultures. In only a few short years, Japan alone has had more than ten millionsubscribers enter the mobile market. Such explosive popularity is an indicationof a strong commercial demand for communications in both the tethered andtetherless environments.

Accompanying the vibrant growth in mobile communications is the growthin multimedia communications, including the Internet. Mobile and multime-dia communications technologies are merging, making mobile computing a keyphrase in the coming advanced information communication era. The growth inthese dynamic industries shows that a change in our chosen method of commu-nications is already well advanced. Reading e-mail and connecting to variousinformation feeds have already become a part of daily business activities.

We are trying to grasp the overall picture of mobile computing. Its shape andform are just starting to appear as personal digital assistants (PDA), handheldpersonal computers (HPC), wireless data communication services, and com-mercial software designed for mobile environments. We are at the cusp of vastpopularization of “computers-on-the-go.”

“Any time Anywhere Computing” provides the reader with an understand-able explanation of the current developments and commercialization of mobilecomputing. The core technologies and applications needed to understand theindustry are comprehensively addressed. The book emphasizes three infrastruc-tures: (1) wireless communication network infrastructure, (2) terminal devices(or ”computers-on-the-go”), and (3) software middleware and architectures thatsupport wireless and mobile computing.

Page 13: Mobile Computing

xii ANY TIME, ANYWHERE COMPUTING

Moreover, the research activities presented in this book provide an insightfullook into the future of mobile computing. I would like to express my sincereapplause to the authors who have completed this enlightening work.

Moriji Kuwabara, Ph.D.Senior Advisor to NTT President

Nippon Telephone and Telegraph Corp.Japan

April 1999

Page 14: Mobile Computing

PREFACE

The proliferation of wireless networks and small portable computing deviceshave led to the emergence of the mobile computing paradigm. Mobile andnomadic users carrying laptops or hand-held computers are able to connectto the Internet through publicly available wireline or wireless networks. In thenear future, this trend can only grow as new exciting services and infrastructuredelivering wireless voice and multimedia data will be deployed.

Any Time, Anywhere Computing: Mobile Computing Concepts and Technologyis intended for technical and non-technical readers. It includes substantialcoverage of the technologies that are shaping mobile computing. Current andfuture portables technology is covered and explained. Similarly, current andfuture wireless telecommunication networks technology is covered and reviewed.By covering commercial solutions and middleware, this book will also help ITprofessionals who are looking for mobile solutions to their enterprise computingneeds. Finally, the book surveys a vast body of recent research in the area ofmobile computing. The research coverage is likely to benefit researchers andstudents from academia as well as industry.

Chapter 1 provides a motivation and an introduction to Mobile Comput-ing. A reference architecture is described along with a discussion of thelimitations of the mobile and wireless environment.

Chapter 2 summarizes two digital cellular systems (TDMA and CDMA)and describes a taxonomy of different wireless network technologies in-cluding in-room infrared, in-room RF, in-building RF, campus area packetrelay networks, wide area packet/switched data networks, and satellitenetworks. The chapter also provides an alternative taxonomy based onthe mobility/bandwidth tradeoffs. Finally, systems issues in wireless net-works are discussed with special focus on the effect of the various wirelessnetworks on multimedia applications.

Chapter 3 provides a historical perspective on the evolution of “PortableInformation Appliances” (PIA). It also describes the emergence and evo-

Page 15: Mobile Computing

xiv ANY TIME, ANYWHERE COMPUTING

lution of the Personal Digital Assistant (PDA). A classification of existingPIAs is presented and sample devices from each category are described.

Chapter 4 extends Chapter 3 by analyzing the suitability of existing PIAs.It describes the new challenges that the PIA industry must face before an“ideal” PIA can be invented. Emerging PIAs and PIAs’ prototypes andconcepts are included in this chapter.

Chapter 5 covers third and fourth generation wireless telecommunicationnetworks and services, which is one of the hotest topics today. The chapterattempts to clarify confusing concepts and sort out the facts from the un-certainties. Wireless network and service evolution is carefully presented toprovide a basis for predictions. Emerging wireless network standards suchas the IMT-2000, UMTS, and ACTS are covered. Future universal air in-terface standards, which is a heavily debated issue, is delicately presented.The TD-CDMA and the W-CDMA proposals for future air interfaces arealso described.

Chapter 6 includes coverage of commercially available mobility softwaresolutions, which is divided into end-user client applications, and mobilitysupport middleware. The chapter addresses interoperability issues anddiscusses shortcomings and limitations of the current state of industry.

Chapter 7 surveys academic and industrial research in mobile comput-ing. It includes research on mobile networking, quality of service, wirelessInternet access, mobile transactions, and mobile computing models. Thesurvey is very condensed but is loaded with citations for further reading.A complete and detailed survey of research in this area is outside the scopeof this book.

Appendix A provides a glossary of acronyms and abbreviations. Thereader is encouraged to consult this appendix if an acronym is encounteredwithout a definition.

Appendix B lists a brief summary of the various wireless cellular systems.It is intended to be used as a quick reference, not as a complete source ofinformation about these systems.

Appendix C includes a listing of standard organizations and consortia.For each organization, a brief description is provided in addition to a Uni-versal Resource Locator (URL) to point the reader to where more infor-mation can be obtained.

Page 16: Mobile Computing

Preface xv

The reader should be cautioned that the book is based on current developments,standards, and trends. The field of mobile computing is relatively new andis constantly changing. Some standards or trends covered in this book arebound to change during the book lifetime. For most up to date informationon standards and products, the reader is advised to consult the latest revisionsand information sheets.

We hope you enjoy reading this book, any time and anywhere.

Abdelsalam HelalBert HaskellJeffery L. CarterRichard BriceDarrell WoelkMarek Rusinkeiwicz

22nd of June 1999

Page 17: Mobile Computing

This page intentionally left blank.

Page 18: Mobile Computing

ACKNOWLEDGEMENTS

The authors would like to acknowledge the contributions of a number of people.Joe Bumblis and Ed Krall contributed to the information gathering on wirelessnetwork technology. Ed White contributed to the writing on the history of thePDA. Jerry West and Melinda Helal assisted in the editing and artwork forthe book. They are responsible for all typos that might be found JohnWilkes helped in gathering and updating some of the information in the Infor-mation Appliances section. A few people contributed to the book cover design.Lin Russinoff designed and produced the wonderful cover with some artworkprovided by Sumi Helal, Sherry Scheetz and James Andrae.

Page 19: Mobile Computing

This page intentionally left blank.

Page 20: Mobile Computing

To my father Ali Abdelsalam Helal for all the love and brotherhood he gaveme.

—A. Helal

To my wife and daughters, who are not impressed by technology, but byutility.

—B. Haskell

Knowledge, experiences, and families are eternal. I would like to thank myWife and Kids for their love and support of my research.

—J. Carter

Page 21: Mobile Computing

This page intentionally left blank.

Team-Fly®

Page 22: Mobile Computing

1INTRODUCTION TO MOBILE

COMPUTING

Mobile computing will be the buzz of the next century. From our first breath,as soon as the umbilical cord is cut, the individual is free and untethered. To betethered is unnatural and soon to be unnecessary for computing environments.Consumers want personalized wireless computing services while they are mo-bile, and companies want to make money offering those services. The infantileparadigm of mobile computing is opening up new markets never dreamed ofbefore. We are presently at the cusp of the mobile multimedia era.

Buzzwords such as mobile, ubiquitous, nomadic, untethered, pervasive, and anytime anywhere, are used by different people to refer to the new breed of comput-ing that utilizes small portable devices and wireless communication networks.Defining and relating some of these buzzwords is an important prerequisite tothis introduction. The difference between nomadic and mobile computing isparticularly important to point out. Both nomadic and mobile computing re-quire small portable devices. However, the kind of network used in nomadiccomputing does not allow mobility, or does so in the confines of a building,at pedestrian speed. Examples of such networks are DIAL-UP lines, whichobviously do not allow any mobility, and Wireless Local Area Networks (W-LAN), which allow for limited mobility within a building facility. Nomadiccomputing refers to the interleaved pattern of user relocation and “in-door”connection. Travelers carrying laptops with DIAL-UP modems are, therefore,nomadic users engaged in nomadic computing. Mobile computing, on the otherhand, requires the availability of wireless networks that support “outdoor” mo-bility and handoff from one network to the next, at pedestrian or vehicularspeeds. A bus traveler with a laptop connected to a GSM phone or a CDPDmodem is a mobile user engaged in mobile computing. Figure 1.1 depicts thistaxonomy. It also shows ubiquitous computing to be the aggregate ability to

Page 23: Mobile Computing

2 CHAPTER 1

compute in both the nomadic and the mobile modes. Mark Weiser, a pioneerand a visionary from Xerox PARC, had different view and definition for ubiq-uitous computing. The reader is referred to his famous 1991 article in ScientificAmerican [91]. We caution the reader that, in this book, the term mobile com-puting is used to refer to both nomadic and mobile computing, to reduce theclutter.

1.1 IMPRESSIVE TECHNOLOGY

An important question to ask is which technology drove mobile computing towhere it is today? Is it the wireless network technology or the miniaturizationand portable computing technology? Unfortunately, there is no easy answer.An individual with a Palm Pilot will probably answer in favor of the portabletechnology, whereas a UPS package delivery worker will be more thankful tothe wireless technology. Whatever the right answer might be, more importantquestions need to be answered: where are we now? and what are the challengesand impediments facing mobile computing? This book attempts to answer thesetwo questions by organizing a morass of information about technologies, stan-dards, research, and commercial products. Satyanarayanan and Zahorjan havearticulated such challenges in their famous articles that we highly recommendto the inquisitive reader [45, 85].

Page 24: Mobile Computing

Introduction to Mobile Computing 3

Figure 1.2 shows how pervasive the portables technology has become. A largecollection of portable devices are available in the market today. These are toomany that we dedicated a chapter in this book to classify and describe eachof them. Another chapter is also dedicated to future portable devices that arecurrently in the prototype development phase or are just pure concepts.

Similarly, the wireless communication technology is growing and expanding at abreath-taking pace. It is changing the way people live and interact. Subscribersare given the power of “ubiquitous” communication at affordable prices. Figure1.3 shows an Ericsson GSM base station in downtown Stockholm. Antenna,power technology, and miniaturization breakthroughs have led to the small-size design of radio equipment and the elimination of large tower and monopoleinfrastructures. It is now amazingly easy to deploy cellular networks (especiallypico-cellular technology) in record times, and at an ever decreasing cost.

There are several applications for mobile computing including wireless remoteaccess by travelers and commuters, point of sale, stock trading, medical emer-gency care, law enforcement, package delivery, education, insurance industry,disaster recovery and management, trucking industry, intelligence and military.Most of these applications can be classified into: (1) wireless and mobile accessto the Internet, (2) wireless and mobile access to private Intranets, and (3) wire-

Page 25: Mobile Computing

4 CHAPTER 1

less and ad-hocly mobile access between mobile computers. This classificationis depicted in Figure 1.4 and will be discussed further in Chapter 7.

An example of a wireless and mobile access to the Internet is shown in Figure1.5, where a traveler, through a Wireless Service Provider (WSP), is able toissue queries based on her location, direction of motion in a particular highway,and perhaps her vehicular speed. Queries such as “nearest Japanese restau-rant”, “nearest hospital”, “Pizza Hut nearest to destination” will soon not beuncanny for the high-tech travelers.

1.2 WIRELESS AND MOBILE COMPUTINGARCHITECTURE

The architectural model of a mobile computing environment is shown in Fig-ure 1.6 and consists of stationary and mobile components. Fixed hosts areconnected together via a fixed high-speed network (Mbps to Gbps). Some ofthe fixed hosts are special computers equipped with wireless interfaces, and areknown as base (radio) stations (BS). They are also known as mobile supportstations (MSS). Base stations, which are placed in the center of a cellular cov-erage areas, act as access points between the mobile computers and the fixednetwork. Mobile computers can be in one of three states. The first state places

Page 26: Mobile Computing

Introduction to Mobile Computing 5

Page 27: Mobile Computing

6 CHAPTER 1

Page 28: Mobile Computing

Introduction to Mobile Computing 7

a mobile computer within a cell and capable of communicating. The secondstate places the mobile computer out of range of any service cell and not ca-pable of communication. The third state places a mobile computer in a cell,communicating, but just ready to cross a cell boundary. These scenarios aredepicted in Figure 1.6.

Figure 1.6 is a generalized architectural overview of a typical wireless/nomadicsystem. Many such systems have been deployed both in the United States andEurope as well as in many other parts of the world. One such European sys-tem is the Global System for Mobile Communications (GSM). GSM, which isdepicted in Figure 1.7, was originally developed by the European Institute forResearch and Strategic Studies in Telecommunications (EURESCOM) as anadvanced mobile communications technology. During early stages of deploy-ment, GSM was hailed as a superior wireless technology because the generalarchitecture supported such features as roaming, minimum disruption whencrossing cell boundaries, and connectivity to any number of public wired infras-tructures. Today, these features are common to most wireless infrastructures.

Page 29: Mobile Computing

8 CHAPTER 1

GSM is gaining increased popularity in North America. Figure 1.8 quantifiesGSM penetration in terms of number of states with GSM services in the US.

Although all of the wireless architecture’s are unique in some respects, they allshare many similar system components. The use of base stations for commu-nication with the mobile computers, the centralized exchange systems whichswitch communications between the wireless domain and the wired infrastruc-ture, and the use of location registers (HLR and VLR) so the system “knows”where the mobile computer is currently located and from where it came, are afew examples of the similarities of these systems.

Another variable, which one might consider a service provider issue, is theincorporation of Advanced Intelligent Networking (AIN) [29]. AIN was a jointeffort between Bellcore (now Telcordia) and the RBOCs in the late 1980’s,with standards completion around 1991. AIN deployment has been slow, butcould play an important role in the realization of third generation wirelessnetworks. Figure 1.9 depicts how the Personal Communications System (PCS)may be incorporated with AIN in an overlay network architecture. The exact

Page 30: Mobile Computing

Introduction to Mobile Computing 9

functionality and nature of the system components is left up to the serviceproviders. Telcordia has only defined the high level attributes of the functionalcomponents and the interconnect between functional planes (layers) to aid theservice providers in product selection and deployment.

Page 31: Mobile Computing

10 CHAPTER 1

The PCS system with AIN services outlined in Figure 1.9 is comparable tothe overlay internetworking system described by Katz and Brewer [69]. ThePCS/AIN system shown above is comprised of many different forms of commu-nications (e.g. cellular, PCS, wired, POTS (Plain Old Telephone Service), etc.)with a centralized management scheme as defined by the Telcordia AIN stan-dards. There exists interconnection across planes and between overlay planes toestablish service attributes. One of the many issues to be addressed is how dowireless service providers and application developers create, deploy, and controlapplications support services given the systems described above.

1.3 LIMITATIONS OF THE WIRELESS AND MOBILE

ENVIRONMENT

For the most part, the existing research and development on mobile computingis driven by the particularities and limitations of the mobile environment. Suchlimitations include:

Frequent disconnection caused by one of the following events:

– handoff blank out in cellular networks; the problem is worse in micro-cellular networks

– long down time of the mobile computer due to limited battery lifetime

– voluntary disconnection by the mobile user

– disconnection due to hostile events such as theft and destruction

– roaming-off outside the geographical coverage area of the wirelessservice

Limited communication bandwidth impacting the following:

– quality of service (QoS) and performance guarantees

– throughput and response time and their variances

– efficient use of battery due to long communication delays (wirelessinterface requires battery energy during the slow send and receive)

Heterogeneous and fragmented wireless network infrastructure leading tothe following problems:

– rapid and large fluctuations in the network QoS

Team-Fly®

Page 32: Mobile Computing

Introduction to Mobile Computing 11

– mobility transparent applications perform poorly without some sortof mobility middleware or proxy.

– poor end-to-end performance of different transport protocols acrossnetworks of different parameters and transmission characteristics.

Other problems include:

– security and anonymity

– service relocation

– support for location-sensitive applications

There are other limitations related to platform and application developmentmethodologies and languages. Operating systems for portable devices (otherthan laptops) are yet to reach maturity. Palm-OS, Windows-CE, EPOCH, andGeOS are the most significant operating systems developed for mobile comput-ing. A version of Linux for hand-held devices is also being developed. Theseoperating systems are light weight with simplified, single-address space mem-ory management. Application portability across these operating systems iscurrently a major problem. The use of Java is currently limited due to the in-adequate performance of JVM on most of these platforms. Development of mo-bile applications on these platforms is typically done through platform-specificSDKs supplied by the operating system vendors. Windows-CE developmentcan also be done using Microsoft Visual C++. A unified and truly portableenvironment is most needed by application developers and inventors of futurekiller apps.

Page 33: Mobile Computing

This page intentionally left blank.

Page 34: Mobile Computing

2WIRELESS

TELECOMMUNICATIONNETWORKS

Today, person to person voice communications, enabled by the telephone, isstill perhaps the most powerful technology available to the average person.The benefit to cost ratio of this technology for the individual is enormous. Anindividual can use a telephone to conduct commerce, earn a paycheck in count-less ways, call for medical assistance, consult experts worldwide on any topic,and essentially obtain almost any critical information imaginable. The mostsophisticated part of this technology is not in the telephone handset itself butin the enormous worldwide communications network to which the handset isattached. The introduction of cellular telephones has certainly improved theindividuals ability to access (or be accessed by) this voice network in any lo-cation. But the global network is now providing more than person to personvoice communications. Data, images, and live video are now routinely trans-ferred to the individual desktop computer. It is expected that these expandingcapabilities will soon be available within some type of portable informationappliance.

There are several well-established cellular infrastructures available today indifferent parts of the world (see Appendix–B for a list of major cellular systems).The European community has standardized largely on GSM. North Americahas broad AMPS coverage with a number of other standards competing in thePCS frequencies. Japan deployed the PHS infrastructure everywhere. A briefcomparison of these predominant standards is shown in Table 2.1.

Page 35: Mobile Computing

14 CHAPTER 2

Page 36: Mobile Computing

Wireless Telecommunication Networks 15

2.1 DIGITAL CELLULAR SYSTEMS

Analog cellular systems such as North America’s AMPS have the disadvantagethat they are very expensive to expand and grow. Each mobile phone requiresa dedicated channel to communicate in a cell site. The only way to expand inAMPS is to build additional cell sites which cost in the range of $500,000 to$1,000,000.

In 1988, the Cellular Telecommunications Industry Association (CTIA) com-missioned a subcommittee called Advanced Radio Technology to define alter-native technologies that allows the cost effective cellular expansion in the US[54]. Proposed technologies focused on Multiple Access network technologies.The first digital system accepted by CTIA is the TDMA system, which standsfor Time Division Multiple Access and which allows users to share the radiochannel through time division. The second digital system accepted by CTIA isCDMA, which stands for Code Division Multiple Access, and which allows usersto share the entire radio spectrum through different, uniquely assigned codesfor transmission and reception. In the next subsections, we briefly describe theTDMA and CDMA cellular systems.

2.1.1 Time-Division Multiple Access (TDMA)

TDMA is a digital transmission technology that allows a number of users to ac-cess a single radio frequency channel without interference, by allocating uniquetime slots to each user within each channel. Currently, a single channel is di-vided into six time slots, with each signal using two slots. This provides a 3to 1 gain in capacity of AMPS. In dispatch systems (e.g. Motorola iDEN), adispatch signal uses one time slot, thus providing a 6 to 1 gain in capacity.D-AMPS, GSM, iDEN and several PCS systems currently use TDMA. TheTelecommunications Industry Association (TIA) provided an early standardfor TDMA over AMPS, known as IS-54, which required digitizing the voice sig-nal, compressing it and transmitting it in regular series of bursts, interspersedwith other users’ conversations. Second generation standard for TDMA by TIAis the IS-136 which uses TDMA on the control channel. TDMA is expected tobe called TIA / EIA-136 once it becomes an ANSI standard.

One problem with TDMA is the wasted bandwidth of unused slots. Timeslots are allocated to specific users whether or not they are using the slots(talking or transmitting data). Hughes Systems Network has contributed anenhancement of TDMA known as Enhanced TDMA (ETDMA) that attempts

Page 37: Mobile Computing

16 CHAPTER 2

to correct this problem. Instead of waiting to determine whether a subscriberis transmitting, ETDMA assigns subscribers dynamically based on whether auser has voice/data to transmit. A phone conversation with long pauses will,therefore, not cause a loss of bandwidth, and will increase the spectral efficiencyof TDMA.

Today, TDMA is becoming a very popular air interface. Over 8 million digitalsubscribers worldwide utilize the IS-54 and IS-136 today. In the US alone, threeof the top four carriers are deploying TDMA IS-136.

2.1.2 Code-Division Multiple Access (CDMA)

In frequency and time division multiplex systems, several hundred channels areavailable within the spectrum allocation of a carrier service. One channel ofone base station is used for each conversation. Upon handoff, the subscriberstation is directed via messaging to discontinue use of the old channel and tuneto the new one. Without reusing the frequency assigned in the spectrum, thetotal number of cells that can be deployed can not exceed the available numberof channels. Frequency reuse is very essential to the design of cellular systemsthat are based on frequency division multiplex.

Frequency reuse utilizes the fact that the attenuation of electromagnetic fieldstends to increase with distance. Therefore, to reuse the frequency without in-curring significant interference, only non-adjacent cells are assigned the samefrequencies. Ideally, cellular frequency reuse is achieved by imposing a hexag-onal array of cells in a service area. A seven cells hexagonal array is shown inFigure 2.1. Seven frequency channels represented by different gray levels areused, one for each cell. The hexagonal array can be replicated and connected,providing a larger coverage area, without using any but the seven frequencychannels. Systems that use frequency reuse includes AMPS in North America,NMT in Scandinavia, and TACS in the United Kingdom.

In reality, cell coverage areas are highly irregular, and do not compare to theideal hexagons shown in Figure 2.1. And even if ideal hexagons are possible, thefrequency division approach offers limited capacity. Take AMPS as an example.Each AMPS operator in North America is allocated 416 channels (30KHz each).In a seven-way reuse hexagon, each cell will be allocated 416/7 = 59 channels.In this example, the capacity of cellular systems can not grow beyond thebandwidth offered by 59 channels, which is 1.8MHz.

Page 38: Mobile Computing

Wireless Telecommunication Networks 17

Code-Division Multiple Access (CDMA) offers a solution to the capacity lim-itation problem. It allows all mobile stations to concurrently use the entirespectrum (all channels) with much less interference. Instead of partitioning ei-ther spectrum or time into disjoint “slots”, each subscriber is assigned a uniqueinstance of a pseudo-noise digital signal. The transmission signal is “spread”over the entire spectrum, using the noise signal. CDMA is, therefore, known asa spread spectrum modulation scheme. The spreading technique is also knownas Direct Sequence scheme. Frequency Hopping is another spreading technique,where the different segments of the subscriber conversation (or data) known asframes are transmitted on a sequence of randomly chosen frequencies withinthe spectrum. In either direct sequence or frequency hopping, the subscriberunit must communicate with the base station to agree on the direct sequence(the pseudo random digital code) or the sequence of frequencies to hop through.Signal interference in CDMA (between neighboring cells) is much less sensitiveto most of the system parameters and is confined within a predictable average.This is one reason CDMA is attractive since it is easier to predict the achievedbandwidth based on the acceptable Noise to Signal Ratio (NSR) and the gainof signal spreading.

Originally, CDMA was invented by Claude Shannon, who suggested that throughnoise-like carrier waves, bandwidth can be increased. Versions of CDMA hasbeen in use for quite sometime by the military for the different reason of se-curity. Transmitted signal is difficult to decode by an intercepting party dueto the spreading and the unknown spreading noise signal. It is known by themilitary to be a Low Probability of Intercept (LPI) and Low Probability ofDetection (LPD) air interface scheme. Since late 1980s, CDMA has been mi-grating into civilian applications and is now reaching maturity and impressivemarket penetration. Future wireless networks known as third and fourth gener-ation wireless networks (based on where you are in the globe) are mostly basedon CDMA and are covered in Chapter 5.

Page 39: Mobile Computing

18 CHAPTER 2

2.2 THE WIRELESS NETWORK TECHNOLOGY

Wireless technologies can be grouped into at least six major categories: (1) in-room, point to point infrared, (2) in-room radio, (3) in-building radio frequency,(4) campus or metropolitan area packet networks, (5) wide-area packet/circuitswitched data networks, and (6) regional-area Satellite Data Networks. Thesesix classes of networks have unique technologies which constrain the nature ofthe applications which can be supported by each of them. A similar taxonomyis provided in [69]. Typically, an overlay of two or more network categories isused to provide continuous coverage in a mixed nomadic/mobile environment.Figure 2.2 shows an overlay of several network technologies. In the followingsubsections, we briefly summarize the characteristics and differences of thesenetworks.

2.2.1 In-room Infrared

The in-room infrared class of networks generally has a network diameter ofabout 40–50m and supports bandwidths of about 1 Mbps. Applications sup-

Page 40: Mobile Computing

Wireless Telecommunication Networks 19

ported by this type of infrastructure are limited to E-mail and collaborative-work applications due to the limited range of the system. The Infrared DataAssociation (IrDA) provides the most common standard used today for thisnetwork technology.

2.2.2 In-room Radio Frequency

The in-room radio frequency class of networks emerged in 1998 with the or-ganized effort of the BlueTooth Special Interest Group [3]. BlueTooth is alow-cost, short range radio that connects mobile PCs with other BlueToohtdevices within a radius of about 10m. Very low energy consumption and about1Mbps transmission speed makes this type of network attractive and suitablefor inter-office device communication.

Hospital intensive care units, bank tellers, and desktop component intercon-nect may be example applications that could utilize in-room RF wireless tech-nologies. The proliferation of portable devices such as 3COM’s Palm Pilot,Windows-CE hand-held computers, and highly portable and powerful laptopssuch as the IBM ThinkPads may incorporate BlueTooth transceivers to bridgethe in-room wireless technology with fixed network infrastructures. The chal-lenge laying ahead is to identify a suitable API for applications that will runatop this specific technology. Such API will allow for the design of “”infras-tructure literate” applications that can accommodate the user expected per-formance levels while maintaining consistency across the infrastructure. Blue-Tooth is further discussed in section 4.2.6 of Chapter 4.

2.2.3 In-building Radio Frequency

This type of network, which is also known as Wireless LAN, expands the rangeof the infrared and the BlueTooth technologies by increasing the network diam-eter to about 200m. Unlike infrared and BlueTooth, in-building radio frequencyis a cellular network, where mobile computers are allowed to roam within andacross cells. Several standards are available today for this type of networksincluding the IEEE 802.11 and the OpenAir interface. Examples of WirelessLANs include Lucent/NCR WaveLAN and Proxim RangeLAN. Both ISA andPC Card interfaces are available with support for Windows and Linux. Proximalso provides additional support to a variety of Windows-CE devices. WirelessLANs can be used in both Infrastructure and Ad-Hoc Modes. In the former,Access Points are used and are connected to the fixed network through a ded-

Page 41: Mobile Computing

20 CHAPTER 2

icated router port. Wireless or nomadic devices with Wireless LAN interfacesaccess the network through the access point in the coverage area (cell). In thismode, the wireless LAN is used as a wireless extension of a fixed, high-speednetwork infrastructure (hence the name). In the ad-hoc mode, several portabledevices with wireless LAN interfaces are placed in the transmission range ofeach others. Each device is capable of communicating with any other devicedirectly, without the help of any networking infrastructure. A private networkis used to configure the network software (TCP/IP) among the ad-hoc groupof devices. Ad-hoc networks is becoming increasingly important technology.

This technology, even though highly mature at this point in time, faces a fewchallenges. First, the IEEE 802.11 standard does not seem to be universallyaccepted (at least not yet). The OpenAir interface consortium, for instance,provides a competing proposal that is gaining popularity. Also, there is a lackof consensus on which air interface to use (direct sequence or the frequencyhopping). Another challenge lies in the fact that wireless LANs are MAC-levelnetworks that do not understand important features of IPv6 such as Multicast,RSVP, among other features. Unless, somehow, these features are implementedfor wireless LANs, certain applications will be difficult to implement.

2.2.4 Campus/Metropolitan Area Packet Networks

This network type encompasses the more traditional “cellular” networkingparadigm. It is typified by a “poletop infrastructure” supporting network di-ameters of 0.2 to 5 miles with data rates of 20-128 kbps. Relay (or router)nodes are strategically placed to support the wider network diameter with asmall price for increased latency. For example, typical latency between a mo-bile device and the first relay node is about 40ms (assuming an uncongestednetwork), and about 20ms between relay nodes.

2.2.5 Wide-Area Packet/Circuit Switched Data Networks

This network is comprised of a more familiar set of technologies and RegionalBell Operating Company (RBOC) services. One such offering is the CellularDigital Packet Data (CDPD) service which is a packetized wireless transportthat utilizes the unused channels of a cellular infrastructure. Motorola’s ARDISand iDEN systems, Ericsson’s RAM (now called MobiTex), and the EuropeanGSM system are contained in this taxonomy. The iDEN network (Integrated

Team-Fly®

Page 42: Mobile Computing

Wireless Telecommunication Networks 21

Digital Enhanced Network) is a packet based voice/data network that uses theMobile-IP networking protocol to route data packets.

Not only is this technology capable of supporting larger diameter networks,but they also tend to have lower bandwidths and higher latency effects thando the in-building networks. This tends to present a unique set of problems inapplication development. Significant body of research on network and systemadaptation through infrastructure awareness components has been or is beingconducted. This research is described in Chapter 7. However, the transforma-tion of this research into commercially available “mobility middleware” is yetto occur.

2.2.6 Satellite Networks

Satellite technology is still emerging. It is a downlink technology where mobilecomputers can only receive direct broadcast from a satellite. Outbound com-munication is initiated by the mobile computer through a modem DIAL-UP orother wireless technology. Hughes Network Systems pioneered the DirecPC net-work which uses the Galaxy satellite and which delivers 400 kbps downlink rate.DirecPC also transmits continuous streams of multimedia information rangingfrom CNN broadcasts, to news, sports, and financial news feeds. Other LowEarth Orbit (LEO) systems are in planning and deployment phases includingthe Internet in the Sky project.

2.3 MOBILITY-BANDWIDTH TRADEOFFS

Another classification of the current wireless networking technology can bebased on the “degree of mobility” offered by these networks. Multi-cellularwireless infrastructures range from in-building cells, to micro-cells (urban cov-erage), to macro-cells (suburban coverage), to satellite (global coverage). In-building cellular offers the highest bandwidth (bi-directional), but very limitedmobility. Micro-cellular offers lower bandwidth but allows for limited-speedmobility; macro-cellular offers much lower bandwidth but allows for the high-est degrees of mobility. As can be noticed, in these networks, the larger thecoverage area (the cell size), the higher the degree of mobility. Satellite net-works are an exception and do not follow this trend. They offer the highestdownlink bandwidth (no uplink possible with satellite networks), but they do

Page 43: Mobile Computing

22 CHAPTER 2

not offer any mobility. Instead, they require a satellite dish to be stationedaiming at the satellite.

Figure 2.3 shows a mapping of the mobility/bandwidth classification onto in-dividual wireless networking technologies. In this mapping, mobility is furtherclassified into indoor and outdoor, with outdoor mobility ranging from station-ary, walking (pedestrian pace), and vehicular speed.

The current mapping of wireless technology to the mobility/bandwidth clas-sification is bound to change. At least this is ITU’s and ETSI’s vision andexpectation of the third and fourth generation networks. For example, wirelessLANs (an in-building technology) is expected to evolve into a network thatallows for limited-speed mobility. Also, macro-cell networks are expected toimprove on the bandwidth they offer. Figure 2.4 depicts this expected evolu-tion.

2.4 SYSTEMS ISSUES

The rapid expansion of wireless Wide Area Network (WAN) services, wirelessLocal Area Networks (LANs), satellite services such as Hughes’ DirecPC andthe planned Low Earth Orbit (LEO) systems have created a large and frag-mented wireless infrastructure. Given such a diverse set of technologies, the

Page 44: Mobile Computing

Wireless Telecommunication Networks 23

need to support mobile applications remains critical and even strategic to manyindustries.

The ability to scale performance and latency while accommodating an increas-ing user density is of paramount importance when designing and/or selecting awireless infrastructure for a particular application. The choice of a wireless in-frastructure must take into consideration the attributes of the application andthe applications class of service requirements including bandwidth, network la-tency, service coverage, and general performance issues. Table 2.2 summarizesapplication classes as stringently defined by ITU-T Recommendation I.211.These classifications have some loose definitions. For example, “interactive”usually means conversational, implying a person on either end of the applica-tion connection. The term “messaging” generally refers to a person talking to amachine. An example would include leaving voice mail or sending a FAX. Theterm ” retrieval” is generally thought of as a machine transferring informationto a person. Also, the term “distribution” is typically thought of as a machinesending to people or machines who listen passively. The Client/Server architec-ture is a primary example of this application class. Application updates mayinclude human intervention, but could be automated. The last five applicationclasses listed in Table 2.2 are considered machine-to-machine interactions, al-though they may have to be “user” activated, while the actual transaction isbetween machines.

Page 45: Mobile Computing

24 CHAPTER 2

Page 46: Mobile Computing

Wireless Telecommunication Networks 25

2.4.1 Multimedia Applications

As of today, there are limitations which prevent the effective exploitation ofwireless networks by portable information appliances beyond the area of voicecommunications, text messaging, and limited data. While it is technically pos-sible to transmit multimedia information such as a motion video clip from theinternet into a portable wireless device, the standards, infrastructure bandwidthlimitations, service costs, data compression technology, and power consumptionconsiderations make this impractical at this time (1999). This limitations canbe attributed to existing standards, which are limited in the level of service theycan provide to the user of a portable information appliance. While they areeffective for voice and text messaging, these standards do not support graphicsintensive internet browsing or real time video at a high enough speed to makethem practical.

In the case of video, the MPEG 1 standard provides for 352 X 240 pixel resolu-tion, comparable to VCR quality video, and requires 1.14 Mbps data rate. Thisis well beyond any of the deployed wireless network standards. Today, consumerexpectation is set by MPEG 2, which supports high resolution video of 1920 X1080 pixels, which requires up to 80 Mbps of bandwidth (typical applicationsof this standard, however, may only require 6 to 8 Mbps). To achieve wirelessmotion video data rates for portable devices, new wireless infrastructure stan-dards will have to be deployed. One such standard is the Wideband CDMAapproach proposed by Ericsson has the specifications shown in Table 2.3.

This standard has been adopted by the European community for the nextgeneration of cellular service and could be implemented globally by 2002. Themotion video quality enabled by such a service would, however, be less thanMPEG 1 in terms of resolution and/or frame rate.

Page 47: Mobile Computing

This page intentionally left blank.

Page 48: Mobile Computing

3PORTABLE INFORMATION

APPLIANCES

3.1 HISTORICAL EVOLUTION

The first portable information appliance was probably a piece of stone or claywith markings on it, used to record numeric information. This informationwas probably very important to the user of this appliance and in some waydirectly affected his livelihood. It may have actually provided the “function”of counting by allowing the user to create a mark corresponding to a piece oflivestock. This would have been very useful to an individual whose society hadnot yet invented a system of numbers. Given the lifestyle of such an individual,ease of use, portability, durability, and reliability were all essential. Ease ofuse probably meant that the individual marks had to be deep enough in theappliance so as to be detectable by touching. This would have been necessitatedby the need to count goats in a heads-up mode while incrementing through themarks with the thumb. Once utilized, this appliance would have to be stowedin an extremely portable fashion so that it did not interfere with other activitiessuch as attempting to frighten away predators, throwing sharpened sticks atpredators and most importantly, running away from predators. Durabilitywould have been important since the user did not have the means to protectthe device from temperature variations, moisture, abrasion, and shock. Tothe user, this device may have played a very important role in establishinghis credibility, accountability, and responsibility with respect to the rest of hiscommunity.

As the technology of mathematics and writing developed, human civilizationprogressed onward to the papyrus scroll (Figure 3.1) and ink pen. This ap-pliance was highly portable and could convey very complex information. Theuser interface took a while to learn (reading and writing), and until relatively

Page 49: Mobile Computing

28 CHAPTER 3

recently, only a limited number of individuals were able to use the technology.Still pen and paper persisted for several thousand years and is still the preferredportable information technology for most of the worlds population.

Two other portable information appliances, the pocket watch and the printedbook (Figures 3.2 and 3.3) are relatively recent inventions which have trans-

Page 50: Mobile Computing

Portable Information Appliances 29

formed human society. The pocket watch enabled the level of logistical syn-chronization between individuals required for industrialization . Printed books,while not as interactive as paper and pencil, have also evolved as the preferredmethod for accessing standardized information in a portable format. Thus, pa-per and pencil, the printed book, and the pocket watch have been the dominantportable information appliances since the dawn of the industrial revolution.

The invention of the semiconductor technology in the Early 1960s began atransformation in portable information appliances, the full impact of whichhas yet to be realized.

The first widely adopted electronic portable information appliance appearedin the early 1970s in the form of electronic calculators. Development work onthese products began in the mid 60s and these designs exploited state of theart discrete transistor technology. By late 1960s, however, companies such asTexas Instruments, Rockwell and Intel had identified handheld calculators asa way to grow the market for Integrated Circuit technology.

In 1970 there were several bulky hand-held calculators on the market at pricepoints of around $300 and above. By 1975, calculators had shrunk to pocketsize and had fallen below the $20 price point. The age of portable electronicdevices, enabled by the integrated circuit, was upon us.

About this time, digital watches also began to replace mechanical watcheswhich had been in place for hundreds of years.

Page 51: Mobile Computing

30 CHAPTER 3

By the early 1980s portable video camcorders had sold over 1 million unitsworldwide and penetration of portable electronics to the consumer had begunin earnest. This rapid penetration was driven by the compelling application ofacquiring and storing motion video images. This trend was further acceleratedby the introduction of 8mm format models which were highly miniaturized.

Personal organizers, such as the Sharp Wizard, were also introduced in this timeframe and were most successful in Japan, where the use of personal computerswas somewhat lagging that of North America. In North America, they werepopular among technophiles but in general, these products tended to be adisappointment to individuals that had experienced desktop computing andfound little compatibility between organizers and desktops.

Cellular phones have seen remarkable penetration worldwide . By the late1980s over 10 million units had been sold worldwide and the cell phone becamea necessity for many and a status symbol for many others.

By the early 1990s, over one million Notebook computers had been sold world-wide as these products demonstrated their usefulness by turning spreadsheetsand word processing into portable capabilities. Early models, in the late 1980s,

Team-Fly®

Page 52: Mobile Computing

Portable Information Appliances 31

from companies like Toshiba and Compaq, featured Monochromatic reflectiveLCDs. These systems were quite adequate for word processing and spread-sheets and were quickly adopted by traveling professionals. Transfer of data inand out of the notebook was achieved through magnetic disk. Prices remainedrelatively high ($2,000 +) due to two factors. First of all, manufacturers wantto maintain high margins so the focus of the Notebook industry was on satu-ration of the business market, in effect, competing with desktop products. Thesecond factor was the desire on the part of the user to have high performancewhich matched as nearly as possible that of a desktop system.

By the early to mid 1990s, several manufacturers were experimenting with thePersonal Digital Assistant (PDA) product concept. These products attemptedto span a gap between the personal organizer products and the notebook com-puter products. These products tended to compromise the miniaturization oforganizers and lacked the full functionality of notebooks. Furthermore, theywere typically crippled with an over sold and poorly performing handwritingrecognition capability. Most importantly, these early products tended to com-pete with, rather than complement the desktop or notebook computer. Severalmanufacturers attempted to add wireless communications to their PDA prod-ucts to make them more appealing. Still, the lack of integration with the

Page 53: Mobile Computing

32 CHAPTER 3

desktop PC and the bandwidth limitations of the wireless telecommunicationsinfrastructure caused these products to fail. The telecommunications infras-tructure in the mid 1990s offered only wireline and cellular modem capabilitieswith fairly low bandwidth (about 14.4 Kbs) for portable products.

3.2 THE ADVENT OF THE PDA

PDAs burst onto the scene in 1993 and mounted a headlong assault into thecommercial market-place only to be quickly repulsed. When the initial exuber-ance subsided, the resulting carnage throughout the industry was both severeand widespread. Now, with forces re-marshaled and armed with a new gener-ation of products, this same industry is attempting another assault, this timetargeting the application specific vertical marketplace.

PDAs emerged in 1993 amid claims of single-point data organization, ubiqui-tous and instantaneous communications, and new operating paradigms usingglitzy graphical user interfaces (GUI) and handwriting recognition. Most if notall of these claims fell short of consumer expectations. The reasons, while ob-vious in hindsight, lay hidden at the time. They were: high customer expecta-tions, immature applications, and incompatible and unrealized infrastructures.

By 1993, the PC industry had introduced its most recent line of laptop com-puters which included computational and storage capacities that rivaled theirmost powerful desktop companions, even though computational and storagecapacity had been doubling every 12 months in recent years. Grazing on thesefertile fields had fattened the software industry and had bred a generation ofsoftware developers with inefficient development skills and tools. This in turnlead to unwieldy applications whose weaknesses were masked only by the rawcomputational and storage capacities of the hardware they ran on. The re-sult was that few wiry developers, and even fewer wiry applications existedthat were capable of operating in the computational, power and storage barrenenvironment of the PDA. Coincidentally, when the first PDAs appeared practi-cally none were supported by third party software and embedded applicationsbeyond the basic notepad, calendar, and calculator were virtually nonexistent.

Early on it was clear the success of the PDA rested heavily upon a varietyof component and service infrastructures with the most critical of these en-ablers being wireless communications. In 1993, riding a sustained boom of 40%growth per year and giddy about recent cooperative initiatives, the cellular

Page 54: Mobile Computing

Portable Information Appliances 33

service providers boasted claims of a complete domestic wireless data infras-tructure (CDPD) by the end of 1994. This effort, seemingly coordinated in itsannouncement, was enthusiastically received by the PDA industry but withina year the initiative would stall and lose much of its support. The breakdowncame in the radio module that provides the link between the PDA and thewireless network. Initially predicted to be PCMCIA sized, it was soon realizedthat the requisite data radios would be both larger and more power hungrythat anyone predicted. When they emerged, larger than some of the PDAsthey were supposed to support, both industries recoiled under the letdown.

As if this was not enough, the whole industry was elevated to a high stateof excitement, by a barrage of hype filled announcements, using phrases like“Imagine if” and “Have you ever ... you will”. Every technology announcementfrom new processor architectures to handwriting recognition techniques addedfuel to the flames. Claims like “desktop performance in your palm”, “timesaving user interfaces”, “ubiquitous communications”, “transportable applica-tions”, “laptop functionality”, and more were touted loud and long. Marketanalysts and prognosticators joined in the frenzy, seemingly unable to separatefuture dreams from first article hardware. As such, market expectations wereset high, and high they stayed, as one product after another fell short andslammed into the reality wall. In fact, the ring of these claims still echoed inthe ears of customers as they tried to use products that were expensive, bulky,fragile, unsupported, incompatible, uncooperative and unstable.

To make matters worse, costs were high and sales were low. The average pricefor a PDA in 1993 exceeded $750, some like the AT&T EO had prices that wentas high as $2000–well outside the reach of many of the target customers. Con-sequently, in the first two years there were just 350,000 units sold. The volumeswere so low in fact, that unlike most consumer electronics, they never crestedthe cost-experience wave which along with competition has the unrelentingability to drive prices asymptotically toward the cost of the raw materials.

There were other problems as well with this initial surge of PDAs, but theyserved only to add to the mass confusion. The industry backlash, however, wasboth clear and severe. With hundreds of millions of dollars invested, two ofthe major players (AT&T EO, and IBM Simon) dropped out completely. Theothers fell back and re-grouped trying to understand what went wrong. Whatwent wrong was equally as clear. Consumers were demanding usefulness andthe first round of PDAs with limited applications and practically no commu-nications simply did not fit the bill. Only a small percentage of the devicessold were ever really used. The vast majority were simply discarded amid thedisappointment and frustration of the once excited user.

Page 55: Mobile Computing

34 CHAPTER 3

Today, the landscape has changed significantly. Unlike the excitement of thepast, PDAs are now met with suspicion and skepticism. Regardless, a newbattalion of products is moving into the fray. This time, however, there isan attempt to reduce the type, and in some cases manufacturers are tryingto distance themselves from the past by avoiding the name PDA altogether,choosing instead names like pocket organizer and personal information manager(PIM).

Despite the reduction in hype, demands on and customer expectations of thePDA have continued to rise. One reason for this is that the laptop computerperformance has continued to double every year. This, along with even higherresolution displays, improved ergonomics increased multimedia functions andmore powerful applications has helped set a new standard for PDAs to meet.While PDAs have made progress in their operating systems, applications andthird party support, they still fall woefully short of customer expectations.

Meanwhile, internet usage has erupted. Reliance on data stored in the myriadof html web sites, not to mention email services, has made wide area commu-nications even more critical to the PDA paradigm than ever. Unfortunately,however, 18 months after ubiquitous wireless data services were promised by thecellular carriers, CDPD is in serious trouble. Southwestern Bell and Airtouchhave essentially stopped their CDPD deployment, leaving major holes like LosAngles, New Orleans and Atlanta in domestic coverage. This lack of clarity inthe wireless infrastructure has caused confusion throughout the industry andcontinues to threaten the viability of the PDA. This problem is compoundedby the fact that the PDA manufacturers seem to rely on third parties to supplywireless modules for their products. This architectural approach results in avariety of inefficiencies and is due to a lack of expertise in wireless implemen-tation. The integration of digital and RF circuitry at the semiconductor levelwill solve this problem in the future, but today, vendors that do not excel inboth computing and communications design suffer a handicap.

Not surprisingly, a detailed look at the current offering of PDAs reveals thatthey are an outgrowth of PC concepts, utilizing the same worldwide componentsand manufacturing infrastructure that has been optimized to support desktopand laptop products. The silicon integration, displays, component size, softwareapplications and substrate densities of this infrastructure has driven the PDAinto one of 2 directions: either toward a fully functional product that is too largeto be practical or toward a product that meets the ergonomics requirements ofthe paradigm, but that severely limits functionality and performance to fit.

Page 56: Mobile Computing

Portable Information Appliances 35

The result is that while the average price of the new PDAs has dropped to$575.00, not much else has changed. Still starved for applications this new gen-eration will not likely outsell its predecessors in the consumer marketplace, eventhough many industry projections say otherwise. Almost in recognition of thisfact the strategists now say that the vertical market is the new focus of theirattention but this is no panacea. The demands of the vertical market are manytimes more stringent than that of the consumer market and the procurementmotivation is much less of an impulse. What is worse, the entrenched competi-tors like Symbol, Norand, and Telxon understand the operating environmentand applications of the vertical market better, and will prove tenacious in theirdesire to maintain market share.

Prom the Early 1960s through the Mid 1990s, the advances in portable infor-mation appliances were impressive. Within the confines of a portable notebook,continuously increasing levels of computing power and display quality had beenachieved. Ubiquitous, wireless voice communications via cellular phone had be-come common place. Consumers recorded hundreds of millions of hours of videodata every year using hand-held camcorders. These products drove the devel-opment of important technologies. Silicon integration evolved from discretetransistor devices to single chips containing over 6 million transistors. Portabledisplays had evolved from simple numeric segment displays with less than 100pixel elements into full color displays with over half a million pixels. Electronicand mechanical packaging technology was capable of connecting thousands ofcomponents in a compact volume compared to only a few tens of parts at thestart of this period. Batteries in the early 60s could store no more than 100watt-hours per litre. Their capacity today is up to 200 watt-hours per litre.

For all of these advances in hardware technology, however, many of theseportable information appliances still seemed as static as the printed book. Ifthey were connected to the outside world at all, it was through a low band-width wireless voice channel which was often unreliable for data transfer. Fur-thermore, the Internet appeared and created heightened expectations aboutinformation access. Without mobile access to the growing global informationnetwork, these portable devices would not live up to their potential.

3.3 PALMTOP COMPUTERS

It is likely that wireless network connectivity will trail wired connectivity interms of performance for the foreseeable future. The best strategy for the de-

Page 57: Mobile Computing

36 CHAPTER 3

velopers of portable information appliance is to design products which eitherprovide useful standalone functions such as an electronic still camera, or whichcomplement wired network platforms. The emerging market of Palmtop Com-puters is a breakthrough in terms of the ability of the Palmtop to complementthe desktop computer.

3.3.1 The Palm Pilot

The Pilot is a highly portable appliance which is the first truly viable substitutefor traditional pencil and paper technology. With desktop synchronization, thisdevice allows the desktop user to augment the networked desktop computingexperience with a portable time management interface. While the Pilot isunlikely to provide services like high quality real-time video in the near future,this product concept has made important inroads into sensibly merging theinteractions of portable and stationary information appliances.

Many other contemporary product designers have failed to take this approachby attempting to combine and therefore replace other devices. One examplewould be a smart phone that combines the functions of a cellular phone anda notebook computer. Such product concepts often end-up compromising thefeatures which make the individual products appealing. For instance, may

Page 58: Mobile Computing

Portable Information Appliances 37

smart phones have poor display quality, unusable keypads, poor battery life,poor performance, and are much bulkier than most cellular phones. The resultis a product that does not effectively replace either of the products that it iscompeting with.

The Palm Pilot V specifications are shown in Table 3.1. The Palm Pilot Vdevice is depicted in Figure 3.7.

3.4 HAND-HELD COMPUTERS

The hand-held computer is another device that attempts to complement thedesktop. It is much more capable than a Palm Computer, larger in size andweight, but can not be fitted in a pocket. Since their first emergence, hand-heldcomputers have been competing with the Palm Computer market.

3.4.1 Sharp Power Zaurus

Page 59: Mobile Computing

38 CHAPTER 3

The Sharp Power Zaurus is a popular hand-held computer that competes withthe Palm Computer market. The Zaurus which is depicted in Figure 3.8 is bestdescribed in terms of its specifications listed in Table 3.2.

Page 60: Mobile Computing

Portable Information Appliances 39

3.4.2 VADEM Clio

Clio is a Windows CE based hand-held PC with a swing-top design that pro-vides three modes of interaction: keyboard, pen and tablet, and presentationmodes. The three modes are achieved by swinging and/or folding the displayaround the keyboard base. The specifications of the Clio, which is shown inFigure 3.9 are listed in Table 3.3.

3.5 COMMUNICATORS

The Communicator is a PDA concept that combines the benefits, portabilityand functionality of digital cellular phones and palmtop computers. The idea

Page 61: Mobile Computing

40 CHAPTER 3

is to stick a palmtop computer to a cell phone with data capabilities to provideremote access, in addition to the stand-alone form factor applications thatcan be found on palmtop computers. Internet access, telnet, email, and webbrowsing are all applications offered by communicators.

3.5.1 Nokia 9000

The Nokia 9000 is the most popular communicator, not only because of itsappearance in the hands of Agent 007 in one of his recent movies (1997), butbecause of the unprecedented unique features and capabilities. The Nokia 9000combined a compact personal organizer with Internet access and a versatilevoice and text messaging system. The organizer includes: an address book,note editor, calendar with to-do list, calculator, and world clock. A built-inbrowser, Telnet, and a VT100 Terminal emulation are built-in applications thatbring the Internet to the mobile user anywhere GSM coverage is available. Amulti-protocol email client, Short Message System (SMS) and a Fax applicationare also bundled to provide a wide spectrum of communication alternative, ofcourse, in addition to the digital voice phone interface.

The specifications of the Nokia 9000 are listed in Table 3.4. Figure 3.10 depictstwo pictures of the communicator. The picture to the right shows the commu-nicator on a recharge base station and reveals the cell phone side of the device.The picture to the left shows an open communicator with a Web page on thebacklit display.

Team-Fly®

Page 62: Mobile Computing

Portable Information Appliances 41

3.5.2 Motorola Marco

The Marco wireless communicator was introduced to the market one year be-fore the Nokia 9000 communicator (in 1995). It featured a built-in two-waywireless packet data modem allowing users to send and receive messages. TheMarco Wireless Communicator, depicted in Figure 3.11, also included a faxand data modem, allowing information to be communicated through any tele-phone network. To augment its functionality, the Marco was equipped withtwo PCMCIA Type II slots to allow users to simultaneously operate third-party software applications and add memory to store more data. The Marco

Page 63: Mobile Computing

42 CHAPTER 3

weighs 1.8 pounds and is 7.5 inches high, 5.8 inches wide, and 1.4 inches deep.The device features a bright portrait screen that allows easy reading in manylighting conditions.

At the time the Marco was introduced, Motorola had the vision of creatingthe first “wireless Newton”. Newton OS 1.3 was therefore used. A similarproduct based on the Magic Cap operating system (from General Magic) wasintroduced in parallel. That was the Envoy depicted in Figure 3.12.

Unfortunately, the Apple Newton did not make it and despite all the softwareand personal information management tools loaded in the Marco, Motorola hadonly sold several thousand units before the device production was discontinued.

3.6 SUB-NOTEBOOKS (MICRO-NOTEBOOKS)

As mobile users continue to demand lightweight, long battery life, and ruggedportable computers, advances have been made in a number of diverse prod-uct concepts including what is now known as higher performance “micro-notebooks”, or sub-notebooks.

Table 3.5 shows the specifications of the Sony PCG-707C sub-notebook that isdepicted in Figure 3.13.

Page 64: Mobile Computing

Portable Information Appliances 43

Page 65: Mobile Computing

44 CHAPTER 3

3.7 NOTEBOOKS

The notebook computer has enjoyed great success as the portable extension ofthe desktop computing environment. Notebooks are now starting to replacedesktops for many users. Today the notebook market provides a most wantedportability by an increasing majority of users. We provide one example ofnotebooks which is the HP Soujourn. It weighs 3.2 pounds and is less than0.71in thick. It uses an Intel Tillamook 233-MHz processor and comes with a2.1 GB hard disk and a 64 MB of memory. Its display is limited though to only12.1in SVGA. The HP Soujourn is shown in Figure 3.14.

3.8 LAPTOPS

Laptops are designed to replace the desktop. They can also be envisioned asnomadic desktops that can be easily moved from one place to another. Theusers of laptops require high performance, large high quality displays, andoccasional portability. Such laptops may have maximum capabilities (as of1999) such as up to 15.0in Color TFT (1024x768), integrated AC adapter,two battery support, up to 14GB disk storage, and 256MB memory. Thesecapabilities come at the price of limited portability with these laptops weighingup to 8 Ibs.

Page 66: Mobile Computing

Portable Information Appliances 45

We provide only one example of a laptop since almost all mobile users arefamiliar with laptops and their capabilities. The specifications of the FujitsuLifebook 900 laptop is summarized in Table 3.6. The laptop is depicted inFigure 3.15.

Page 67: Mobile Computing

46 CHAPTER 3

3.9 OTHER INFORMATION APPLIANCES

3.9.1 HP CapShare

HP’s Capshare 910 is a hand-held portable device that allows mobile users tocapture, store, communicate and print documents, The(inches) device shown in Figure 3.16 weighs 12.5 oz and uses two AA NiMHrechargeable batteries that last for 100 document capture followed by a down-load. Typically, a mobile user capture documents from a newspaper or a mag-azine and then stores the document into his laptop or other portable device.Both PDF and TIFF formats are supported. The device has 4MB of memoryand can capture from business cards and small receipts up to legal-size doc-uments or 25in. newspaper columns. Maximum capture area of 119 squareinches. A standard letter-size page takes about 6 seconds to capture. Moreinformation about this device can be found at www.capshare.hp.com.

Page 68: Mobile Computing

Portable Information Appliances 47

3.9.2 Clarion AutoPC

The Clarion AutoPC (depicted in Figure 3.17) is the first in-dash personaldigital assistant. It integrates cellular telephony, Internet email, navigationsoftware, GPS satellite tracking, contacts information and calendar, real-timeinformation feeds (e.g. stock quotes and traffic information) in a single device.Hands-free interaction is possible through a speech recognition interface. Moreinformation can be obtained from www.autopc.com.

Page 69: Mobile Computing

This page intentionally left blank.

Page 70: Mobile Computing

4FUTURE INFORMATION

APPLIANCES

By 1995, cellular phones, at least in North America, became less of a statussymbol and more of a annoying necessity. Individuals, and those around them,were no longer enamored with cellular phone technology and began to resent theintrusion that a cellular phone imposed on the personal time management andgroup interactions. Today, many people leave their phones off, except whenthey are making an outbound call or are expecting an important call. Thedemand for additional services such as text messaging and integrated voice mailare growing rapidly and are a key marketing feature of the new PCS services.As we move into the future, people will continue to demand the benefits ofany time, anywhere communications and access to information. They will alsodemand more sophisticated methods of managing the way in which the restof the world has access to their personal time. Ultimately, technologists haveyet to deliver a portable information appliance that has the level of impact onindividual lifestyles as did the wired telephone or the wristwatch.

This chapter addresses the requirements for future Portable Information Appli-ances (PIA) that will be able handle multimedia information. It also presentsseveral emerging PIAs and PIA prototypes and concepts.

4.1 NEW CHALLENGES

There are several requirements for the development and usability of an idealmobile computing device. They can be summarized as follows:

Page 71: Mobile Computing

50 CHAPTER 4

Development of everywhere, low-cost communication infrastructure per-mitting ubiquitous access. One of the larger challenges will be coverage,connectivity and the ability to hand-off between private and public wirelessnetworks.

Intelligent energy management. In our information society, content isking. The two largest power-consuming components to a mobile deviceare communications and display. Unfortunately, battery technology is notexpected to improve significantly in the next five to ten years. Therefore,energy-efficient design and operation of processors, storage, memory anddisplay will play an important factor in the usability and widespread useof this technology.

Interoperability across platforms. Creating a computing environment thatis not interoperable is a problem, not a solution. The mobile environmentneeds to be interoperable. The mobile computer should interoperate withmultiple heterogeneous networks. Within the same communication systemthat uses different frequency bands, the mobile computer should be ableto adapt to different frequency bands.

Portability. Applications designed to work on a single platform will notlive for long. Applications should be portable to different platforms. Theuser is not sure which mobile device he will use from one day to the next.An application should be usable and accessible on a communicator, a palmcomputer, or a laptop. Equivalently, the user should be portable to thedifferent platforms and devices. Regardless of the device chosen, the userenvironment should be portable.

Ubiquitous Computing. Related to the issue of portability and interoper-ability, users today may own several computers including desktops, lap-tops, palm computers, handhelds, and communicators among other smalldevices (even a wristwatch). The proliferation of small devices has alreadycreated an unmanageable environment for the confused users who spendsignificant time attempting to unify their information through synchroniza-tion and manual mediation. This confirms Mark Weiser’s views on how weare evolving from one computer one man, into many computers, one man.Ubiquitous computing will require that several computers collaborate toserve the same owner, not the other way around.

Usability. Smaller portable devices that maintain functionality and user-friendliness of larger devices are highly desired. Miniaturization makesgetting information in and out of mobile computing devices difficult, par-ticularly in Personal Digital Assistants (PDAs) and Hand-held PersonalComputers (HPCs). Making the interface intuitive and simple with as

Team-Fly®

Page 72: Mobile Computing

Future Information Appliances 51

few steps as possible to accomplishing a task is critical because input andoutput of information using these devices is painful enough.

Just like the evolution of network-based computing (from mainframe terminalnetworks to client server networks), we will witness the empowering of themobile user and the mobile device. The mobile device will increase in capabilitynot only because the hardware engineers are pushing the edge, but also thecommunications engineers are cramming more data bits down a circuit. Theincrease in end-terminal capabilities is a trend that happens with all networks;the intelligence of the network is constantly pushed to the outer edges.

Even after the aforementioned challenges are overcome, economy of scale condi-tions must be met in order for mobile computing to be pervasive. The growth ofnetwork terminal intelligence will be driven by money. Computers and networksexist from a business perspective to make money either by sales transactions,advertising, or increasing efficiency. The service provider can push processingcycles off of its equipment and on to the consumer end terminal where theservice is delivered. Since the consumer has excess processing capacity and iswilling to give up the processing cycles, the service provider saves money onequipment and bandwidth. A programming language that accommodates thisparadigm is Java. Wireless networks and mobile computing have traditionallybeen an expensive luxury and like their wired counterparts will not becomepervasive until they become economical, they help in making money, and theyfulfill specific consumer needs.

4.2 EMERGING PORTABLE INFORMATION APPLIANCESAND TELESERVICES

Future portable information appliances and services are still open to the imag-ination. However, a surprising trend is observed, where portables and wirelessnetworks are becoming increasingly tightly coupled to the mobile user. In thissection, we illustrate some of the ambitious research, developments, productsand standards of future portable devices and teleservices.

Page 73: Mobile Computing

52 CHAPTER 4

4.2.1 Wearable Computing (MIT)

This project personalizes computing in a fashion that enables computers tobe worn much like eyeglasses or clothing. The systems include some of thefollowing components: heads-up displays, unobtrusive input devices, personalwireless local area networks communication and context sensing tools. Inter-action with the mobile computing system is based on the context of the situa-tion. Applications and services offered might fall into the following categories:intelligent assistant, remembrance agent, augmented reality, or intellectual col-lectives. Figure 4.1 shows a wearable, head-mounted display unit that includesboth an output display and an input video camera as well. A video imageof the camera’s input is continuously projected into each eye. Two CRTs aredriven by one video camera who’s focal length is adjusted to avoid angularmodification of the user’s expected visual field [9].

4.2.2 Wearable Computer Systems (CMU)

At Carnegie Mellon University (CMU), wearable computers are tools that inte-grate the user’s mobile processing, information space, and work space providingautomatic, portable access to information. Wearable Computers is a CMU re-search project being funded by DARPA ETO, with additional support fromDaimler-Benz, Intel, and DEC. Its intended applications and services are insuch areas like maintenance and plant operations and customization in manu-facturing. The main vision of wearable computer systems is that they are tools

Page 74: Mobile Computing

Future Information Appliances 53

in the user’s environment much like a pencil or a reference book instead of asstand alone, remote computers running programs. With the use of wirelessnetworks, the wearable computer can provide automatic, portable access toinformation. Furthermore, the information can be automatically accumulatedby the system as the user interacts with or modifies the environment [5].

4.2.3 IBM Wearable PC

IBM Japan has unveiled a wearable personal computer (PC) prototype devel-oped by IBM Japan Yamato Laboratory. Roughly the size of a portable stereo,with a stand-alone headset and a miniature one-handed controller, the wearablePC can be worn as a piece of light clothing and performs like a full-functionnotebook computer. The prototype was demonstrated at the IBM Fair, Tokyo,Japan, 1998. The wearable PC is composed of three elements connected bycables:

the main unit, which is the size of a portable stereo unit.

the miniature controller, with microphone, TrackPoint and click button,all of which fit in the palm of a hand, and

the headset with micro-display and headphones.

The main unit weighs only 299g, and incorporates a 233Mhz processor, an IBMmicrodrive (a one-inch hard disk drive), and battery pack. It uses MicrosoftWindows 98 as its operating system and can be loaded with all regular appli-cation programs. The total weight of the wearable PC, including main unit,controller, headset and connector cables, (shown in Figure 4.2) is 449g, makingit far lighter than any other high-performance personal computer to date.

4.2.4 Body LAN: A Wearable RF Communications System

In 1996, BBN recognized how the increase in human-worn electronics wouldlead to interconnection and wiring problems. BBN’s solution is the BodyLAN,a radio communications network over the area of a person’s body allowing freedistribution of wearable components and a gateway for access to the Internetor an Intranet [2].

Page 75: Mobile Computing

54 CHAPTER 4

Page 76: Mobile Computing

Future Information Appliances 55

4.2.5 Toshiba Desk Area Network (DAN)

Toshiba has developed a network software that manages either radio or infraredtransmissions to link personal computers, surpassing the need for a server. Thistype of network is a wireless desk area network, referred to as DAN. It allows onegroup or multiple groups of users to quickly form shared data (text or graph-ics) without needing the magic of a networking guru. The implementation ofwireless DANs brings nomadic computing, the paperless office, and groupwareone step closer to the business office. Toshiba DAN discovers neighboring de-vices and autonomously creates a self-organized network of terminals or laptopswithout a server. When machines enter or leave the area, it automatically anddynamically rebuilds an ad hoc network environment to reconstruct groupingsand maintain communication routes. DAN technology has been commercializedfor wireless RF and infrared [16].

4.2.6 BlueTooth

BlueTooth is a recent development similar to Toshiba’s DAN. It is a technologyspecification for low-cost, short range radio links between mobile PCs, mobilephones, and other portable devices. The idea is to standardize a cheaper andshorter range version of existing RF technology. By doing so, it will be feasibleand cost-effective to equip different portable devices with a wireless exten-sion. The goal is to enable users to connect a wide range of computing andtelecommunications devices easily and simply, without the need to buy, carry,or connect cables. BlueTooth can also be used to quickly setup ad-hoc net-works by allowing automatic, unconscious, connections between devices. Oneexpected benefit to BlueTooth is that it virtually eliminates the need to pur-chase additional or proprietary cabling to connect individual devices. More onthe ongoing developments in BlueTooth can be found in [3].

4.2.7 Seiko Wristwatch PC

Seiko Instruments Inc. successfully commercialized the world’s first wristwatchPC in Japan on June, 1998. The watch, which is called the Ruputer, is thecheapest wearable PC on the market today with models pricing in the range of$300. Hardware specifications include a 16-bit CPU, 128 KB of main memory,512 KB of ROM, 512 KB of flash memory, 102x64 dots backlit display, 19,200bps infrared port, 19,200 bps serial port, cursor pointer (only left and rightmovements), and four buttons. Its software includes several applications, some

Page 77: Mobile Computing

56 CHAPTER 4

of which run under Microsoft’s Windows-95 operating system such as Schedule,AddressBook, MemoBook, FamilyBook, data entry editors, and viewers (text,images, and sound). The watch can also be used to play games (of course!).Both text and images can be downloaded to the watch from PCs. The watchcan exchange data, via infrared signal, with other Ruputer watches, or can beconnected to a computer or a laptop through a serial line [15]. Figures 4.3 and4.4 show pictures of a ruputer and its docking station.

4.2.8 NTT PHS Wristwatch Phone

Nippon Telegraph and Telephone Corp. (NTT) developed a prototype wrist-watch telephone which weighs 1.58 ounces (45 grams) [13]. The wristwatchphone is powered by a high density lithium ion battery, and is based on anewly developed LSI (large scale integration) chip with special power-savingfeatures. The wristwatch phone operates on the personal handy phone system

Page 78: Mobile Computing

Future Information Appliances 57

(PHS), which was developed in Japan. PHS is similar to a regular cellularphone, but has a shorter range and is usually less expensive. The wristwatchoperates on public or home modes, includes an 11/2 character display, and pro-vides a sophisticated, user-dependent voice recognition. NTT field-tested thewristwatch phone during the Winter Olympics in Nagano (Japan). NTT plansto make its PHS wristwatch phone commercially available by 2000. Figure 4.5shows the PHS wristwatch phone.

4.2.9 NTT Ring Keyboard

Because the size of portable devices is most often limited by the keyboard anddisplay, advances improving portability of these two components are heraldevents. One unique idea developed out of Nippon Telephone and Telegraph(NTT), Japans telecommunications giant, is the Ring-Keyboard. Shock-sensorrings allow a person to type on any surface and translates the different finger

Page 79: Mobile Computing

58 CHAPTER 4

combinations into syllables. Data is transmitted to the main computer throughthe watch.

4.2.10 Display Pad: The Next Generation TV

The consumer electronics industry is on the verge of the biggest revolution sincethe introduction of the personal VCR. This next revolution will not be drivenby a single new product but rather by multiple new innovations which are beingintroduced simultaneously. Of these new innovations, the most important arethe large flat panel display and broadband data services to the home.

Plasma displays with diagonal dimensions greater than 40 inches are beingintroduced into the industrial and consumer markets at an aggressive pace.Retail price points for these devices may drop below $2000 before 2002. Ourexperience with these devices indicates that consumers will find these largehigh-resolution displays to be a very compelling enhancement to the traditionalpastime of watching TV. Furthermore, the resolution afforded by these displayssurpasses the critical cognitive threshold required to blend WEB style networkcontent with traditional broadcast and pay per view TV services. The resultwill be a massive and accelerated acceptance of network centric services by thegeneral public.

The impact that this revolution will have on the nature of portable devicesshould not be underestimated. While the preferred interface to the network

Page 80: Mobile Computing

Future Information Appliances 59

will be the large flat panel with a high bandwidth connection, there will bea tremendous demand for portable devices that support network interaction.Many of these devices will be involved in interacting with the large flat panelitself; others will provide an alternate interface to the network when the flatpanel is not available. Whereas today’s successful portable devices are viewed asbeing useful in and of themselves, future devices will be judged on their abilityto leverage the network. Consumers will require these devices to be highlyergonomic and compatible with all systems that they are likely to encounter.

The Display Pad is a future concept of an enhanced version of the executiveportfolio (pad of paper). Pen and paper provide a proven platform for er-gonomic data capture. The surface hardness and friction characteristics of penand paper cannot be matched by schemes, which require the user to use a mockpen on the surface of a flat panel display. The Display Pad digitizes this famil-iar form of user interface. Information on the network can be viewed privatelyor shared. Display Pad has a built in browser for viewing information and en-tering information into network-based applications. The Display Pad, which isshown in Figure 4.7, is not a general-purpose notebook PC, although it couldsuffice the needs of many users that today require a notebook.

4.2.11 The Ear Phone

Page 81: Mobile Computing

60 CHAPTER 4

The Ear Phone is a future concept of a low power wireless voice communica-tion device that allows the user to access voice based services over the network.These services can include telephony, and access to information services thatutilize voice recognition for input and voice synthesis for output. The mostpowerful attribute of this device will be an embedded IP address. This IP Ad-dress will be continually broadcast to local servers allowing them to downloadinstructions from a particular network location. These instructions will con-tain a user profile that will allow the local server to interpret the signals comingfrom the Ear Phone in a customized fashion. This will allow the user to havevoice control over appropriate devices connected to the local server. The userinterface device stays with the user. The information needed to interpret thatusers signals follows the user around the network. TCP/IP protocols alreadyprovide the basic capability to make this a reality. The Ear Phone device isshown in Figure 4.8.

4.2.12 Power Ring and the Magic Wand

The Power Ring and the Magic Wand are concept devices that will be madepossible by utilizing Micro-Electro Mechanical Systems (MEMS) technology.These devices sense a user’s hand motions and transmit telemetry data to alocal server. The server will be able to interpret these motions because it hasdownloaded a user profile for the device as with the Ear Phone. The PowerRing and Magic Wand are personal devices that can be used to provide agesture interface to various network devices. For example, a person may use acombination of voice (Ear Phone) and gesture (Power Ring) control to interactwith a large flat panel display. Figure 4.9 depicts the Power Ring and the MagicWand.

Team-Fly®

Page 82: Mobile Computing

Future Information Appliances 61

4.3 CONCLUDING REMARKS

Even if we discover the most powerful and ideal portable device, availabilityof bandwidth will remain a major factor in determining what the subscribercan do or access. To achieve the ultimate goal of wireless and mobile multi-media, developments in the pico-cell wireless technology must catch up withdevelopments in the portable information appliance technology. Increasing thefrequency of the radio transmission increases the rate of signal change. Further,the more the signal rate changes, the more bits you can send down that pipe.The management of interference will also need to be improved. One methodof managing interference is to decrease the distance between mobile and basestations. In short, higher frequencies and shorter transmission distances requirethe wide development of pico cell technology.

Portable devices will not be very useful without wireless information serviceproviders offering a variety of teleservices. Accessing such teleservices ubiqui-tously will rely on the constant availability of an access point to the Internet.Lack of access to the network equates to services not being delivered and lossof money. One of the greater challenges to ubiquitous access is financial, nottechnical. As previously stated, computers and networks are driven by money.The service provider that develops the successful financial model to providingubiquitous access to their services will be the winner.

As far as energy consumption, the emerging packaging technology known as“system-on-chip” will help eliminate the bottleneck between the memory and

Page 83: Mobile Computing

62 CHAPTER 4

the processor by building both memory circuits and logic circuits on the samechip. System-on-chip also improves power management by decreasing the phys-ical distance between the memory and processor components, and by dynam-ically turning off and on the logic of parts of system whose functions are notneeded all the time.

Finally, corporations that are selling devices in the portable computing marketare each in search of the Holy Grail and it looks like this: a device that has thetechnical prowess for ubiquitous wireless access, intelligent power management,cross platform interoperability, portability, and an intuitive human interface; adevice that makes money for business consumers and accommodates the needsof the home user. Mobile computing will bring services directly to the consumer,not to the consumers home or to the television set, but to the actual person.By arriving at the ideal portable device, mobile computing will be enabled tohandle multimedia and will open up an untethered world of new services andmarkets beyond our imagination.

Page 84: Mobile Computing

5FUTURE WIRELESS

COMMUNICATION NETWORKS

As we move into the 21st century, air is proving to be the ultimate medium de-sired for most of our voice and data communication needs. Consumer demandsand expectations are driving wireless technology to be the fastest growing seg-ment of the telecommunication industry. In response, researchers and standardorganizations are stretching the physical capacity of radio to prepare and ac-commodate for future wideband services that the consumer will demand.

This chapter provides a roadmap for future directions in wireless cellular net-works. It focuses on future wireless teleservices that will be supported by thirdgeneration networks. It also describes some of the research activities associatedwith fourth generation wireless networks.

Understanding future wireless teleservices permits us to get a glimpse into fu-ture network design and deployment. Telecommunications service evolutionfrom wireline network services to expected wireless consumer services is dis-cussed. Requirements for third-generation wireless networks and wireless tele-services, as defined by various standard organizations, are also covered.

The role of ITU and ETSI in the development and deployment of third and forthgeneration networks is explained. Emerging technologies and associated re-search activities in the wireless communication industry are covered, includingTD-CDMA (time division - code division multiple access), W-CDMA (wide-band code division multiple access), SDMA (space division multiple access),and W-ATM (wireless asynchronous transfer mode). Each technology is brieflydescribed and its impact on future cellular systems is examined.

Page 85: Mobile Computing

64 CHAPTER 5

5.1 FUTURE WIRELESS TELESERVICES

Consumer-driven wireless teleservices will define future wireless network andterminal requirements. However, consumer expectations for future teleservicesare primarily driven by the existing wireline network service capabilities. Stan-dards bodies like the ITU (International Telecommunications Union) and ETSI(European Telecommunications Standards Institute) have invested years of ex-perience in defining and orchestrating the development of future networks, alsobased on existing wireline services.

5.1.1 Wireline Network Services

The three wireline networks in the US that consumers are most likely to relateto are the PSTN (public switched telephone network), the cable TV network,and the Internet. The wireline teleservices offered through these three networkswill be the key driver and the benchmark for consumers when comparing orasking for similar future wireless teleservices.

Page 86: Mobile Computing

Future Wireless Communication Networks 65

However, some of the existing wireline services are broadband services, whichrequire network bandwidth and other resources beyond the capabilities of ex-isting wireless networks. Broadband services, therefore, present a huge obstaclein the evolution of future teleservices. Table 5.1 outlines the present uplink anddownlink requirements for some wireline broadband services.

Wireless networks hope to offer services by exploiting the same intrinsic ser-vice characteristics built into wireline networks. For example, the JapanesePHS system (Personal Handy Phone System) is based on ISDN [18]. Bothstandards use common channel signaling and similar set of services, includingsub-addressing. The GSM standard (Global System for Mobile Communica-tion) also incorporates compatibility with ISDN [53].

To give a deeper insight into the direction of future wireless services, wirelineprotocols can be researched for their service capabilities. ISDN, ADSL, SDSL,VDSL (Asymmetric, Symmetric, and Very high rate Digital Subscriber Line)are wireline protocols that will be delivering essential services to home sub-scribers [56]. When these services become widely used, wireline subscriberswill become so accustomed to them and will start demanding the same servicesin an untethered and perhaps mobile fashion.

5.1.2 Wireless Service Evolution

As wireless networks extend deeper into our lives, wireless teleservices are evolv-ing from general purpose services into specific services satisfying personal needs.The teleservices of tomorrow will span WAN, LAN, DAN, BAN, (Wide, Local,Desk and Body Area Networks), and will serve such industries as police/publicsecurity, banking, utility, transportation, health care, marketing, restaurantand retail.

The first teleservices evolved from voice services delivered via wireline terminals(handsets), to cordless phones, to cellular phones. Wireless teleservices willevolve with network deployment. Voice services already evolved to build intobasic data services such as CDPD (Cellular Digital Packet Data). When basicdata services become ubiquitous, they will build to higher rate data services,and then evolve to bandwidth on demand. Figure 5.1 characterizes wirelessnetwork evolution by charting different wireless systems by mobility and datarate. For example, cordless phones evolved into PHS offering higher data bitrate and greater mobility. Another example is cellular now evolving to IMTS-2000.

Page 87: Mobile Computing

66 CHAPTER 5

5.1.3 Market Evolution

As wireless networks are becoming more common in our daily lives, we are wit-nessing a rapid, world-wide expansion of the wireless network market. In fact,some observers in the Asian telecommunication industry go as far as predictingthat by the year 2000, there will be as many wireless subscribers as wirelinesubscribers. More developing countries are deploying wireless infrastructuresuch as GSM, and even in well established markets, such as in Japan, new cus-tomer groups are emerging. For example, the Japanese PHS system is recentlybeing catered towards children.

But the largest impact on future wireless teleservices is anticipated to be dueto the Internet. In fact, a convergence of the wireless and Internet industrieshas recently occurred, leading to the emergence of the mobile Internet. Thisconvergence will have a profound impact on our methods of communication,electronic commerce, and other future wireless teleservices. The ubiquitous,untethered access to the Internet is setting expectations for consumers as wire-less data communication networks and wireless Internet service providers areon the rise.

Page 88: Mobile Computing

Future Wireless Communication Networks 67

5.2 EMERGING WIRELESS NETWORK STANDARDS

ITU (International Telecommunication Union) is a United Nation affiliatedorganization that oversees global telecommunication systems and standards.ETSI (European Telecommunications Standards Institute) is Europe’s premiertelecom standards organization well known for its development of the GSMstandards. Both organizations are currently leading efforts to promote coop-eration in the definition and development of future wireless networks. Onegoal common to both organizations is achieving seamless communication forthe global consumer through cooperation on technical developments. Decisionsmade within these two organizations will have a dramatic effect on the futuredirections of wireless networks and services.

5.2.1 IMT-2000

The ITU (International Telecommunication Union), headquartered in Geneva,Switzerland, is an international organization within which governments and theprivate sector coordinate global telecom networks and services [20]. IMT 2000(International Mobile Telecommunications by the year 2000 project) is a projectunder the ITU that plans to facilitate cooperation in deciding global wirelessaccess for the 21st century. Until recently, IMT 2000 was known as FPLMTSor Future Public Land Mobile Telephony System.

IMT 2000’s vision is to “provide direction to the many related technologicaldevelopments in the wireless industry to assist the convergence of these essen-tially competing wireless access technologies.” IMT 2000 is expected to unifymany different wireless systems, leading to the global offering of a wide rangeof portable services. It is expected that the IMT 2000 project will enable themerging of wireless services and Internet services, leading to the creation of amobile multimedia technology and new modes of communication.

IMT 2000 has the following goals:

Incorporation of a variety of systems

Achieve a high degree of commonality of design world wide

Compatibility of services within IMT 2000 and with the fixed network

High quality and integrity, comparable to the fixed network

Page 89: Mobile Computing

68 CHAPTER 5

Accommodation of a variety of types of terminals including the pocket sizeterminal

The ability to use a small pocket terminal world wide

Connection of mobile users to other mobile users or fixed users

Provision of services by more than one network in any coverage area

Availability of a range of voice and data services to the mobile user

Service portability–no difference between the services, transport capabili-ties, source coding, customer service or human-machine interface, regard-less of where the call was placed [30]

Provision of these services over a wide range of user densities and coverageareas

Efficient use of the radio spectrum consistent with providing service atacceptable cost

Provision of a framework for the continuing expansion of mobile networkservices and access to services and facilities of the fixed network

An open architecture which will permit easy introduction of advances intechnology and of different applications

A modular structure which will allow the system to start from as smalland simple a configuration as possible and grow as needed, in size andcomplexity [20]

IMT 2000’s vision of future wireless teleservices can be summarized as twosets of recommendations or requirements. The first (shown in the list below)is future, third-generation network requirements, and the second (shown inTable 5.2) is future wireless teleservices requirements.

Third generation IMT 2000 network requirements include:

Operation in a multi-cell environment (satellite, macro, micro and pico)

Operation in a multi-operator environment

Near-wireline quality voice service

Near-universal geographical coverage

Page 90: Mobile Computing

Future Wireless Communication Networks 69

Low equipment cost, both subscriber stations and fixed plant

Minimum number of fixed radio sites

Seamless inter-frequency hand-off

Mobile speed data rate of 144 kbps

Portable speed data rate of 384 kbps

In-building fixed wireless data rate of 2 Mbps

BER (bit error rate) of

Creation of direct satellite access

Transmission scheme suited for high speeds (e.g. fast running trains 100km/h)

Flexibility for evolution from pre-third-generation and for post third gen-eration systems [74, 79, 83, 67].

IMT 2000’s vision of future wireless teleservices is specified in terms of informa-tion rate, user delay sensitivity, and bit error rate requirements. Eight classesof services are specified as listed in Table 5.2.

Initial assignments of the IMT-2000 spectrum for Europe, US and Japan areshown in Figure 5.2. It shows how the dream of global roaming might beachieved in the future.

5.2.2 UMTS

UMTS, which stands for Universal Mobile Telecommunications System, is cur-rently a project under the SMG (Special Mobile Group), a committee in ETSI.

Decisions made in early 1998 by ETSI has given Europe a clear direction to-wards the realization of its third generation wireless communication system.ETSI agreed in January 1998 on two different UTRA methods: W-CDMA inthe paired portion of the radio spectrum, and TD-CDMA in the unpaired por-tion. Both of these access schemes are explained in greater details later in thischapter.

The main goals of the UMTS system can be summarized as follows:

Page 91: Mobile Computing

70 CHAPTER 5

Team-Fly®

Page 92: Mobile Computing

Future Wireless Communication Networks 71

The accommodation of high speed, multimedia interfaces to support In-ternet applications at speeds of up to 2 Mbps, through a quantum leap intechnology

At least a 3-fold increase in spectral efficiency

Support from an evolved GSM core network

Compliance in meeting or exceeding ITU’s Family Concept IMT 2000 Sys-tem [21]

The UMTS project schedule and milestones are shown in Figure 5.3. The firstUMTS deployment is shown to be planned for the year 2002.

5.2.3 ACTS

Another organization that is greatly influencing the direction of wireless com-munications, particularly W-ATM, are the projects funded out of ACTS (theAdvanced Communications Technologies and Services). ACTS is a group of Eu-ropean research projects with budget 50% funded by the European EconomicCommission (EEC). The remaining 50% of the research funding is providedby those industry organizations involved in the research. ACTS broad objec-tive is to develop advanced communications systems and services for economic

Page 93: Mobile Computing

72 CHAPTER 5

development and social cohesion in Europe [1]. Research projects by ACTS in-clude: multimedia, photonics, high speed networking, and mobile and portablecommunications.

5.3 THIRD GENERATION WIRELESS NETWORKS

IMT 2000’s original vision for third generation wireless networks was to create asingle global communication system common to all countries and regions. Thisvision was too revolutionary to second generation wireless network providerswho have invested heavily in current technology. To protect their investments,carriers requested ITU to consider a more evolutionary approach to third gen-eration network standards. ITU, in turn, modified its vision into creating a“family of systems” that would converge and comply with a common set ofrequirements for third generation networks.

Following IMT 2000’s vision, current research, development, and global stan-dardization efforts are focused on upgrading second generation systems includ-ing GSM, CDMA, and TDMA. A major goal of this conversion is to upgradethese system evolutionary over time while maintaining the operation and prof-itability of the existing second generation network infrastructure.

Page 94: Mobile Computing

Future Wireless Communication Networks 73

TD-CDMA (Time Division, Code Division Multiple Access) and W-CDMA(Wideband Code Division Multiple Access) are the two major evolutionarynetwork schemes currently under consideration by ITU. SDMA (Space DivisionMultiple Access) is also receiving attention as a network scheme with similarevolutionary nature.

Before summarizing the details of TD-CDMA, W-CDMA, and SDMA, we firstdescribe the evolution of the wireless network technology in Japan, Europe,and the USA. Figure 5.4 helps clarify the alphabet soup in which the wirelessindustry swims. The figure defines the technology and generation in which aparticular wireless system operates (or once operated). The data rates at thebottom of the chart apply only to the cellular technologies, not the cordlessor WAN packet data rates. It is also important to note that at the time ofpublication, the third generation wireless technologies for the US had not beenselected. Europe (presented in ETSI) just selected a combination of W-CDMAand TD-CDMA. Japan had chosen W-CDMA for their third generation wirelesstechnology.

Page 95: Mobile Computing

74 CHAPTER 5

5.3.1 Time Division/Code Division Multiple Access

TD-CDMA is a proposed radio interface standard that uses CDMA signalspreading techniques to enhance the capacity offered by conventional TDMAsystem. Digitized voice and data would be transmitted on a 1.6 MHz widechannel using time-segmented TDMA technology. Each time slot of the TDMAchannel would be individually coded using CDMA technology, thus supportingmultiple users per time slot.

One design goal of TD-CDMA is to allow the CDMA technology to be smoothlyintegrated into the existing, second generation GSM TDMA structure world-wide. This will allow GSM operators to compete for wideband multimediaservices while protecting their current and future investments.

An important feature of TD-CDMA is its ability to adjust the ratio of spectrumallocated for the uplink and the downlink. The air interface can therefore betuned to enhance the performance of certain applications such as Internet accessand voice applications.

TD-CDMA uses the same frame structure as GSM 5.5. It has eight time slotswith a burst duration of 577 micro seconds and a frame length of 4.616 msecas shown in Table 5.3. The TD-CDMA carrier bandwidth is eight times thatof the 200 kHz GSM carrier equaling 1.6 MHz. The compatibility between theTD-CDMA and GSM time bursts and frame structure permits the evolutionarystep to third-generation systems.

As many as 8 simultaneous CDMA codes are allowed in one time slot in TD-CDMA. This permits 8 users per time slot, or a larger combination of voice anddata users to communicate without interference. For example, TD-CDMA canaccommodate 11 voice users and 5 data users and still maintains the appropriateBER [83] ( for voice and for data). Eight users per time slot appearsto be selected because it offers a happy medium between the number of voiceand data calls that the system can accommodate.

Because TD-CDMA has the ability to assign multiple codes to one user, itpermits broadband (or bandwidth on demand) transmission capabilities. As-suming a bandwidth of 1.6 MHz, a time slot with an information rate of 16 kbpsusing QPSK data modulation, eight possible users per time slot (eight CDMAcodes per time slot) gives you an information rate of 128 kbps. If all eight timeslots were allocated to a single subscriber in a pico cell environment or wheremobility is restricted, 1024 kbps can be achieved. Changing to a different data

Page 96: Mobile Computing

Future Wireless Communication Networks 75

Page 97: Mobile Computing

76 CHAPTER 5

modulation scheme like 16QAM instead of QPSK, an information rate of 2048kbps is conceivable. Requirements for third-generation cellular systems are metby the TD-CDMA system [67].

Another advantage of TD-CDMA is the fact that intra-cell interference is or-thogonal by time. This enables multiple subscriber signals to be received atdiffering power levels thereby eliminating the near-far effect and the need fora soft hand-off. The hand-off is conducted through a separate TD-CDMA orGSM carrier simplifying dual mode, dual band handsets. This is a divergencefrom GSM that conducts the voice and control channels on the same 200 kHzradio band. The sources studied on the TD-CDMA were not clear if mo-bile assisted hand-offs (MAHO), where the subscriber unit returns radio signalstrength information back to the base station, is a feature in the TD-CDMAsystem.

The strategic importance of TD-CDMA can be summarized as follows:

1. The networks that TD-CDMA is catered towards are significantly deployedinfrastructures, including:

GSM: deployed in 74 countries, 200+ networks, and 20+ million sub-scribers, and

AMPS: deployed in 110 countries, 40+ million AMPS subscribers,1.5+ million D-AMPS subscribers.

2. The cost of changing the air interface in a cellular system is significant.System design and setup with regards to the MSC (mobile switching cen-ter), the BSC (base station controller), the cell location, and frequencyreuse are all based upon the characteristics of the access scheme. Selectinga revolutionary different access scheme is therefore more than just changingthe air interface; it is a costly operation. The TD-CDMA system is de-signed to be an evolutionary –not a revolutionary– step from GSM secondgeneration infrastructure to third generation infrastructure.

The benefits to taking a revolutionary step include the absence of a legacysystem and a quantum leap in abilities. The risk, however, is shorteningthe return on investment on second generation infrastructure [30]. Butregardless of the philosophical underpinnings, keeping deployed infrastruc-ture profitable is a concept well-embedded in the telecommunication in-dustry.

3. TD-CDMA promises to be future proof:

Page 98: Mobile Computing

Future Wireless Communication Networks 77

Spectral efficiency twice that of GSM

Reuse of existing GSM network structure and principles: cell sites,planning, hierarchical cell structures

Efficient interworking with GSM

Inherent TDD (time division duplex) support for cordless operation

Data rate up to 2 Mbps indoor, 1 Mbit in all environments

No soft hand-off and fast power control

TD-CDMA has recently been agreed upon by ETSI as a third-generation solu-tion for GSM service providers. Support from most major telecommunicationsequipment providers in Europe has played a role in ETSI’s decision to adaptTD-CDMA.

5.3.2 Wideband Code Division Multiple Access

W-CDMA [53, 48] is a spread spectrum technology in which the entire band-width is shared by multiple subscribers for transmission. A subscriber’s datais modulated with PN codes; the signal is then spread and transmitted acrossa wideband. The receiver is responsible for despreading the desired signal fromthe wideband transmission and contending with interference. The despread-ing process at the receiver shrinks the spread signal back down to the originalsignal and at the same time decreases the power spectral density of the in-terference [50]. This inherent ability to manage interference is at the heart ofW-CDMA.

In W-CDMA, there are four control channels: the pilot, synchronization, pag-ing and access channels. The channels are identified in the transmission byusing a specific PN code, either a Walsh or Hadamard function (see [48] forfurther explanation). In the 5 MHz W-CDMA forward link, there are two desig-nated codes for possible assignment to two possible pilot channels, two codes fortwo possible synchronization channels, a maximum of seven inclusive sequen-tial codes for paging channels, and the remainder codes are not assigned andare used for forward traffic channels (see Table 5.4). The traffic channels areassigned n channel code numbers based upon desired the data rate, n = 0 … 64for 64 kbps, n = 0…127 for 32 kbps, and n = 0… 255 for 16 kbps. The re-verse link channels are the access and the reverse traffic channels. Codes remainunassigned on the reverse channels so that channel assignment can be done dy-namically and in response to paging channels and to interference. The forward

Page 99: Mobile Computing

78 CHAPTER 5

link and the reverse link are FDD (frequency division duplexed). Frequencyseparation depends on the countries frequency allocation scheme. The channelsin both the forward and reverse links are frequency division multiplexed.

W-CDMA’s channel responsibilities can be described as follows:

Pilot channel (Forward link)The BTS (base transceiver station) transmits one or two pilot channelscarrying a reference clock necessary for demodulation and the hand-offprocess. The pilot channel also carries information used in estimatingBTS signal strength therein indicating the best communication link for thesubscriber terminal. After deciding on the best pilot signal, the subscriberterminal demodulates the synchronization channel.

Synchronization channel (Forward link)The synchronization channel contains system parameters, offset time, ac-cess parameters, channel lists and neighboring radio channel lists, all nec-essary in synchronizing with the paging, access and voice channels. Thesynchronization channel always operates at 1200 bps.

Paging channel (Forward link)System parameters and paging information to groups or a single subscriberare continually sent on the paging channel. Pages are combined into groupspermitting a sleep mode to be built into the subscriber terminal, extendingbattery life. Subscriber terminals can monitor multiple paging channels.Thus, when another cell’s paging channel has a better signal, a hand-off isrequested. The paging channel has a data rate of 9600 bps or 4800 bps.

Access channel (Reverse link)When a page is detected the terminal attempts to access the systemthrough the access channel. The terminal increases signal strength sent

Page 100: Mobile Computing

Future Wireless Communication Networks 79

to the BTS until the system responds, a random time limit has expired,or maximum power levels have been exceeded.

Traffic channel (Both forward and reverse links)Within the traffic channel, there are two types of in-band signaling used.Blank-and-burst in-band signaling where an entire 20 msec frame is re-placed with control information. Dim-and-burst is also in-band signalingbut the control information is distributed throughout a variable numberof 20 msec frames.

The forward and reverse channels are modulated differently, QPSK for theforward channel, and O-QPSK for the reverse channel. There are five steps tothe modulation process:

1. A PN multiplier multiplies the user data by the Walsh or Hadamard func-tion, uniquely identifying that information to a specific subscriber ter-minal. The functions are time-shifted so that the set of functions areorthogonal

2. The output of the multiplier is a code rate of 4.096 Mcps using 5 MHzbandwidth (see Table 5.5) that is split into two signals: inphase (I) signaland quadrature (Q) signal

3. The pulse shapes for the I and Q signals are smoothed, minimizing rapidsignal transition that results in radio frequency emissions outside the allo-cated bandwidth

4. The balanced modulator multiplies the I and Q signals by two signals thatare 90 degree phase-shifted. The number of bits per chip depends on thedata rate supplied to the balanced modulator

5. The output of the balanced modulator is then fed to a RF (radio frequency)amplifier

W-CDMA’s 5 MHz bandwidth provides robust frequency diversity. Selectivefrequency fading usually affects only a 200 - 300 kHz range of the signal [53].Time diversity happens because of multipath fading channels. W-CDMA solvesthe problem in a couple of different ways. One way is the selection of onlythe strongest signal, a process similar to antenna diversity. Rake reception isanother technique where weak signals are added together to build a strongsignal. Inherent time diversity receiver provides robustness against fading.

Page 101: Mobile Computing

80 CHAPTER 5

Exact time alignment in W-CDMA is not necessary. Time offset or variableoffset is why W-CDMA is considered to be quasi-orthogonal by time. Thischaracteristic is used to decrease the interference by shifting the time alignmentby 1.25 msec.

W-CDMA employs ADPCM (adaptive differential pulse code modulation) asa speech coding method at a coding rate of 32 kbps. Voice is sampled anddigitized at 64 kbps then supplied to the speech coder that characterizes andcompresses the data at a rate of 16 - 32 kbps, depending on speech activity.The acceptable BER and the speech encoding are variable, which improves thecapacity in a W-CDMA environment. Data is sent when speech is detected;there is no reservation of time slots or frequencies, thus you are achievingmaximum utilization of the available bandwidth. To achieve the same effectin a TDMA or FDMA environment, statistical multiplexing must be employedwith speech detection. Statistical multiplexing requires frequencies and timeslots to be reassigned, thus complicating the system and raising the cost.

Power control in a W-CDMA environment is an open and closed loops. Anopen loop is a coarse adjustment of the signal strength. This means the sub-

Team-Fly®

Page 102: Mobile Computing

Future Wireless Communication Networks 81

scriber terminal continually receives from the BTS Radio Frequency amplifieradjustments measuring signal strength loss and the terminal reacts accordingly.A closed loop is fine adjustment of the signal strength, meaning in every 1.25msec time slot from the BTS there is a power control bit indicating to the sub-scriber unit to increase or decrease transmission power. The end result is thesignal received at the BTS is always at approximately the same power level.

There are two fundamental types of W-CDMA systems–synchronous and asyn-chronous [50]. In synchronous operations, all symbol/chip transmissions ofall subscribers are orthogonal by time eliminating co-subscriber interference.This decreases interference and increases channel capacity, but increases systemcomplexity. Asynchronous operation, on the other hand, permits co-subscriberinterference and allows more flexibility in system design, but lowers channelcapacity. Synchronization on a system level is coordinated through the use ofthe synchronization channel. Synchronization on the subscriber level is coor-dinated via the pilot channel reference clock and is used in demodulating thereceived signal.

Hand-offs in a W-CDMA system are soft hand-offs. They are initiated by thesubscriber terminal finding a better paging channel from a different BTS. Thesubscriber terminal communicates with both BTSs while the MSC coordinatesthe simultaneous communication. The call is then handed off from one BTSto the other, completing what is known as a soft hand-off. The hand-off isreferred to as soft because the terminal is always in communication with aBTS, a condition that results in fewer dropped calls.

Japan’s jump into W-CDMA is encouraged by a lack of capacity in the presentlydeployed system. The population density is such that a third-generation sys-tem is needed immediately. NTT predicts that wireless subscribers will equalwireline subscribers in the year 2000 at 60 million.

The Japanese W-CDMA system will be connected to an advanced broadbanddigital wireline network. The wireline connections are to be as follows: ATMadaptation layer 2 (AAL2) to be used between the BTS and the MSC via theBSC, PSTN and ISDN from MSC to the central office, and TCP/IP for In-ternet connections. The experimental prototype includes three cell sites, sevenmobile stations, and up to 2 Mbps transmission rate. Other W-CDMA systemparameters for the NTT DoCoMo/Ericsson testbed are outlined in Table 5.7.

The following are the features in the NTT DoCoMo/Ericsson W-CDMA ex-perimental system:

Page 103: Mobile Computing

82 CHAPTER 5

Page 104: Mobile Computing

Future Wireless Communication Networks 83

1. Subscriber unit can receive multiple channels resulting in multimedia band-width. The NTT DoCoMo W-CDMA system can accommodate up to six64 kbps channels simultaneously for a total bandwidth of 384 kbps persubscriber, enabling six different teleservices at the one time. This bit rateachieves the NTT DoCoMo phase one testbed goal of 384 kbps per sub-scriber; phase two has the goal of achieving 2 Mbps per subscriber. Formore details about W-CDMA channel information see Tables 5.5 and 5.6.

2. The system allows for future expansion with the aid of adaptive antennas.Adaptive antennas use SDMA techniques. As explained in section (5.3.3),SDMA manages interference and thus increases the network capacity, im-proves link quality, increases signal range, reduces transmission power, andextends the life and profitability of the deployed infrastructure.

3. New random access procedure with fast synchronization that provides flex-ibility in user data rates

4. Protocol structure that is similar to the GSM protocol structure

5. Inter-Frequency Hand-off (IFHO )

6. Hierarchical Cell Structure (HCS ), permitting hand-offs between differentwireless systems, (i.e. a hand-off between PHS infrastructure to the W-CDMA infrastructure)

7. VOX - Voice activation silence suppression, does not send data when theaudio level is below a threshold. VOX is also noted in the PHS ARIBstandard as a low power consumption operation for the private system [18]

8. Speech coding Orthogonal Variable Spreading Factor codes (OVSF ). Uti-lization of a speech detection tool and orthogonal speech codes providesmaximum bandwidth utilization in the W-CDMA environment. The speechdetection tool, as explained earlier, assists in transmitting only the neces-sary data by transmitting less when speech activity is low. The orthogonalspeech codes prevent interference with other channels decreasing interfer-ence and increasing capacity

5.3.3 Space Division Multiple Access

SDMA is a technology which enhances the quality and coverage of wirelesscommunication systems. It uses a technique wherein the subscriber’s access isvia a narrow focused radio beam and the location of the subscriber is trackedadaptively by an intelligent antenna array system (see Figure 5.6). The name

Page 105: Mobile Computing

84 CHAPTER 5

SDMA is derived from the physical spatial characteristics between the focusedradio beams. Spatial processing is not a new concept; it is used in presentlydeployed cellular infrastructures. For example, some cell sites are sectoredat 120 degree. Also, most base station sites use two antennas for diversityreception regardless of whether they are sectored or not [46].

The most distinguishing aspect of SDMA is its management of interference.Reducing interference increases the effective network capacity, link quality andsignal range. It also reduces the transmission power. Collectively, all thebenefits brought by SDMA are expected to extend the life and profitabilityof second-generation network infrastructure.

SDMA is applied to the TDMA and CDMA systems differently because ofthe systems’ basic differences. TDMA co-cell subscribers are orthogonal bytime. Increasing the capacity in a TDMA environment by employing SDMAtechniques requires multiple users on different radio beams to be assigned tothe same carrier frequency and time slot. If the spatial component becomesinsufficient between subscribers then an intra-sector hand-off is required to beinitiated. The TDMA protocol needs to be expanded to permit these intra-sector hand-offs.

CDMA subscribers use the same frequency and are quasi-orthogonal by time.Subscriber signals are distinguished by code filtering, not by time slots. Because

Page 106: Mobile Computing

Future Wireless Communication Networks 85

of these CDMA characteristics, no intra-sector hand-offs are needed, keepingprotocol overhead to a minimum [46].

Four types of interference concern cellular systems [50]:

1. Background noise

2. External interference

3. Other cell interference, and

4. Other user noise

All systems deal with interference types 1 and 2. Interference types 3 and4 are dealt with differently depending on the type of access method. In aTDMA system, interference types 3 and 4 are orthogonal either by frequencyor time and do improve with frequency reuse planning. An interference signalfrom a neighboring cell base station is orthogonal by frequency to the desiredsignal. Also an interference signal from a co-subscriber within the same cell isorthogonal by time to the desired signal.

In a CDMA system, interference types 3 and 4 are spread across the samefrequency and not necessarily orthogonal by time. In CDMA, all subscribers usethe same frequency. The access method distinguishes between the desired signaland interference types 1 through 4 (which includes co-subscriber interference)by using a Pseudo Noise (PN) code. The PN code is known by both the basestation and the subscriber unit for spreading and despreading the desired signalin the bandwidth. In a SD/CDMA environment, the spreading code acts likea direction estimator. The spreading code has the responsibility of locatingthe signal within the interference so the antenna array just has to establishan antenna beam in the direction of the user. In the TDMA environment,the antenna array has to distinguish between the interferer and the user whosesignal structures are the same. The interferer signals have to be “nulled” beforeestablishing a radio beam; the consequence of which might be the cancellationof all but one subscriber accessing that antenna array [46].

The employment of SDMA in a CDMA environment provides an easy increasein capacity. There are no additional protocols or controls that need to be im-plemented; only the deployment of an intelligent antenna array is required.Employment of SDMA in a TDMA en vironment, however, requires new fre-quency planning and alteration of protocols [46].

Page 107: Mobile Computing

86 CHAPTER 5

5.4 FOURTH GENERATION WIRELESS RESEARCH

Beyond third generation wireless infrastructure is W-ATM (Wireless Asyn-chronous Transfer Mode). As previously explained in this chapter, wirelessinfrastructure is greatly influenced by the wireline infrastructure. Future direc-tions in wireline infrastructure are towards ATM, and the wireless infrastruc-ture will align itself appropriately. W-ATM is being researched because of thepossibilities of providing high speed data transmissions with a low BER andhigh QoS in densely populated areas.

ATM is a protocol designed to accommodate multiple network services. Theintent is to have one type of network for all types of data thereby increasingefficiency, services, and throughput, while decreasing costs and network com-plexity. ATM is an end-to-end communication system accommodating networkservices requirements for lossy or lossless data, bursty traffic with real-timerequirements, or data with no time requirements.

W-ATM is a communication system for a hybrid tethered/tetherless environ-ments. It has challenges and obstacles at layers one, two, and three of theOSI (Open Systems Interconnection) model. Comments in this section will befocused on layer one issues and solutions. Many W-ATM layer one issues aremainly comprised of characteristics found in a mobile radio environment. Thisincludes:

Fading

Multi-path propagation

Signal attenuation, and

Interference types, including inter-cell subscriber, intra-cell subscriber, back-ground noise, and external or other noise.

Some significant research toward resolving W-ATM layer one issues is in thearea of diversity reception like antenna arrays and SDMA. Diversity receptiontechniques solve some of the issues that W-ATM is facing, like fading and multi-path propagation. Multiple received signals caused by multipath propagation,received at antennas spaced at a distance of a fraction of the wavelength, allowsthe received signals to be treated as independent rays. Statistically, one of thereceived signals at a given point will not have faded, then by using diversitycombining, the strongest portion of the two independent signals are used to

Page 108: Mobile Computing

Future Wireless Communication Networks 87

create a third signal, partially eliminating the effects of multipath propagationand fading. Diversity combining also allows the mobile terminal to reduce thetransmit power–something battery researchers like to hear.

Higher frequencies and wide bands are capable of delivering the needed through-put rates expected of an ATM network. Interference and other layer one issuesat higher frequencies are minimized by decreasing the transmission distancebetween the mobile terminal and the base station, thereby improving BER andQoS. Given the bandwidth, BER, and QoS requirements of an ATM network,W-ATM will be deployed first in the private and public pico-cell infrastructureor in highly populated areas, i.e. indoor wireless LANs or systems like Japan’sPHS infrastructure.

Japan has the advantages of already having a publicly-deployed pico cell net-work and the population density to make a pico cell network financially possible.The PHS network is deployed on the island of Japan, where 125 million peopleare packed into an area slightly smaller than the size of California. With apublicly-deployed pico-cell network, the evolution to W-ATM will be swift anddecisive. In the year 2002, Japan will be fulfilling the mobile multimedia dreamwith 10 Mbps wireless ATM links.

5.5 CONCLUDING REMARKS

Air is the ultimate communication medium, making the untethered domain anattractive and booming market for teleservices. Future directions in wirelessinfrastructure will be driven by consumer expectations, wireline network ser-vice capabilities, and pre-defined requirements developed by standards-makingbodies.

Significant research in wireless networks conducted in laboratories around theworld today is centered around the wireless air interfaces which are responsiblefor a significant fraction of the total wireless infrastructure cost. TD-CDMAis being developed for the safety net that it extends to the GSM network. W-CDMA, on the other hand, is being developed because of its many strengthsincluding: decreasing the power spectral density of the interference, coping withmultipath fading channels, and requiring no frequency planning. W-CDMAalso has both variable capacity and multimedia bandwidth capabilities. SDMAis being researched and developed to extend the profitability quality of second

Page 109: Mobile Computing

88 CHAPTER 5

generation networks. It uses diversity reception techniques that increase cellcapacity and significantly reduce interference.

Japan’s NTT DoCoMo is building flexibility into the W-CDMA system per-mitting compatibility with W-ATM. Japan’s fourth generation networks willtherefore develop faster because of the extensive, already-deployed public pico-cell infrastructure. This equates to higher mobile terminal requirements, earlierthan other markets, i.e. 10 Mbps in a pico cell network by the year 2002.

In addition to air interface issues, there are a number of third generation issuesthat need to be given equitable importance: (1) a new generation of multimode,multiband terminals is needed, (2) standardization over multiple autonomouswireless systems is required; this will lead to hierarchical cell structured net-works that will require the capability of handing-off between different and some-times competing systems, and finally (3) the convergence of the Internet andthe wireless world is inevitable; to truly realize a nomadic multimedia comput-ing environment, efficient wireless data services and nomadic computing modelsneed to be developed.

Page 110: Mobile Computing

6STATE OF INDUSTRY: MOBILITY

SUPPORT SOFTWARE

The suite of products that provide some support for mobile computing spans thetechnology space from end-user client applications, such as spreadsheets, Webbrowsers, through middleware, down to products implemented in hardware thatprovides cellular or other radio transmission-based communications services.This chapter addresses primarily the top layers of this technology stack: end-user client applications and, to a limited extent, that portion of middlewarethat appears as options to users as they choose the features to be includedwith their portable computer.

End user applications such as Pocket Quicken, Farcast, Wyndmail, and Pact,that are narrowly focused on email, personal organizers, two-way paging andother similar services are not covered here. Instead, we focus on those end-userapplications which involve remote access, client/server, and/or data synchro-nization. In those applications, we further focus on the top layers of the mid-dleware communications software that provide support for mobile computingand data access.

As a caveat, we should point out that the marketplace for mobility-supportsoftware is currently undergoing rapid changes both in the products available,and in the companies that provide them.

6.1 COMPETING PHILOSOPHIES

The mobile client-application architectures which are emerging in commercialproducts can be roughly divided into three overlapping classes: Remote-Node,

Page 111: Mobile Computing

90 CHAPTER 6

Client Proxy, and Replication. The basis for this subdivision is the need toaddress the problems associated with wireless bandwidth and battery limita-tions and the alternatives, that are commercially available today, for managingthose problems. This classification is, therefore, different from the “research”classification given in Chapter 7. A brief description of each class follows.

Remote-Node. This approach attempts to create a facsimile of a fixednetwork client node by hiding all artifacts introduced by wireless com-munications. Under this model, all client software which run on a wirednetwork platform would function without change on a mobile platform thatincludes a compatible OS and other library services. Accordingly, it placesthe most stringent demands on the middleware and other software (whichsupports the client application) to mediate the problems that arise as wire-less artifacts. As a result, this approach is most susceptible to failures inthe wireless infrastructure. Software packages which adopt this approachmay recognize some of the wireless limitations and adapt their behavioraccordingly. For example, when response time is of concern, the limitedbandwidth of wireless communications encourages the system to deliverrecords one at a time as they are retrieved from a database server ratherthan sending all record hits for some query. However, the ultimate goal isto provide an opaque overlay for the underlying ensemble of networks thatshields the user from any concern for their interoperability. Remote-nodeapplications can be realized by porting full clients (as used in the wirelinenetwork) to a mobile computer with compatible communication middle-ware. Shiva PPP is a famous middleware that supports most TCP/IPclients.

Client Proxy. This approach, characterized by products like Oracle Mo-bile Agents, attempts to minimize transmission costs and the impact ofdisconnects by buffering a client’s requests, and/or the servers responses,and by resorting to batch transmissions. In this way, a user may select avariety of record types from several different tables, and then save batterypower by disconnecting while the server processes the request. At somelater time, the client can reconnect and receive a batch of records that sat-isfies all of the requests. The underlying assumption is that the end-userrecognizes that periods of disconnect will occur, and that these periodswill not impact the user’s ability to perform useful work.

Replication. Clients which will be disconnected for extended periods oftime, but which require immediate access to important data can satisfythose requests from locally cached replicas of key subsets of the databases

Team-Fly®

Page 112: Mobile Computing

State of Industry: Mobility Support Software 91

which are stored at some server site. Changes to the data that occur ei-ther at the client or the server must be reconciled through periodic clientconnects which may be initiated manually by the user, or automaticallyby the replication software. Some update conflicts may occur when mul-tiple disconnected clients alter the same records. These collisions must bereconciled in some way.

6.2 END-USER CLIENT APPLICATIONS

A flurry of activity appeared in the trade press in late 1995 describing the rushby vendors, both large and small, to market mobile client software packages.Some of those products are discussed in this section. Recent literature searchsuggests that many of these products never materialized, were re-targeted towired networks, or in some cases, are still struggling with weak sales. However,there are some big players with deep enough pockets to continue to pursue thismarketplace. The discussions here are restricted to those products and servicesthat still appear to have a current or promised market presence.

6.2.1 Oracle Mobile Agents

This product is a buffering and communications package for wireless platforms.A software agent that runs on the mobile client platform intercepts requestsmade by the client to the Oracle server and buffers them for a later transmissionto the server. A companion Oracle agent runs on the Oracle server platform.That agent receives the buffered requests, submits them to the Oracle server,and buffers the responses for later transmission to the client. The server agentis capable of serving any number of mobile agents simultaneously. Conversely,a client agent can access any server agent that it knows about and for which itholds the appropriate DBA access privileges. Oracle agents can run on mobileplatforms equipped with NT, Unix, or Windows and can communicate overTCP/IP using Shiva’s PPP communications middleware. This product doesnot automatically support transactions or queries that span multiple Oracleservers.

6.2.2 Oracle Lite

Page 113: Mobile Computing

92 CHAPTER 6

This product is a cut-down version of the Oracle server that can run in a smallportable system (or a desktop workstation). It can be used as a companiontechnology for the Oracle Agent Software to store local copies of subsets ofcorporate databases and can accumulate updates to the data that are generatedlocally at the mobile client. Oracle may provide the Oracle Lite server with“two way” replication which could automatically propagate updates either tothe client from the central server site, or vice versa. Recently, Oracle andPalm Computing (a 3Com company) announced an alliance to integrate theOracle Lite client database and the 3Com Palm III and PalmPilot organizers,allowing new and existing Palm Computing platform applications and data tobe replicated, synchronized, and shared with an Oracle 8 database server.

6.2.3 Oracle Software Manager

This product is intended for a database administrator who needs to propagatesoftware updates to remote copies of the Oracle server. It is capable of perform-ing the distribution via hardwired networks or through wireless connections. Itis not clear whether this package is versatile enough to accomplish a distributedsoftware update to a collection of mobile devices as though the entire operationwere a distributed transaction. For example, if the DBA needs to update themobile Oracle-Lite server software for entire sales staff, the updates may haveto be performed individually by the DBA.

6.2.4 Oracle Replication Manager

Oracle has announced a version of its Replication Manager which will even-tually support bi-directional replication among a collection of distributed andcentralized server databases. The Oracle approach is based on a peer-to-peermodel, much like Lotus Notes, in which a collection of distributed processesmanage replication collectively.

6.2.5 Sybase SQL Remote

Unlike the Oracle Replication Manager, the Sybase product called SQL Remotehas adopted a centralized model for managing replication. This product is amember of the Sybase SQL Any Where suite of tools (formerly called WatcomSQL). Also, Sybase has optimized its replication server to accommodate users

Page 114: Mobile Computing

State of Industry: Mobility Support Software 93

that are only occasionally connected. So while this product has been developedwith wired network users as a primary target, the software does include acomponent that recognizes the frequent disconnects that typify mobile users.

6.3 MOBILITY MIDDLEWARE

The majority of products targeted for the middleware market rely on TCP/IPand socket-like connections for the client server interface whether they are in-tended to be deployed in the wireline network arena or the wireless domain.Variants of TCP have been proposed to circumvent the problems that plagueTCP for some wireless applications. By choosing to adopt this defacto stan-dard transport protocol, vendors are positioning their products for deploymentin a large existing infrastructure. As a result, it is already possible to surf theInternet using a Netscape interface on many wireless platforms and a simplecellular phone connection.

Two key players in the wired-network middleware market that provide sup-port for distributed users are Novell’s Netware and Microsoft’s Remote Access.Neither of these products will be discussed further since neither has yet an-nounced plans (that we have seen) for moving into the wireless middlewaredomain. However, Microsoft Exchange has been integrated with Shiva’s PPPsoftware that allows communication of clients to servers through the cellularphone network.

6.3.1 MobileWare Office Server

This suite of products was introduced in 1995 as a solution to managing mobileaccess to corporate data. The basic strategy that underlies MobileWare is tominimize mobile platform connect time by executing data transfers in a burstmode. The intent of this software is to make the mobile platform appear to theuser as though it were actually a node connected into the wired network. Theinitial customer target focused on large sales staffs that were primarily mobileand who needed access on demand to sales support information that was toobulky and/or volatile to carry on extended trips.

The current flagship product, MobileWare Office Server, includes a native Lo-tus Notes mail and database replication support. MobileWare Office Server isan agent-based middleware for wireless or wireline access to application data.

Page 115: Mobile Computing

94 CHAPTER 6

Services supported by Mobile Ware Office Server includes Lotus Notes, Webbrowsing, e-mail, and file transfer. A core component of the MobileWare Of-fice Server is the Intelligent Transport Engine. The transport engine providesseveral features including:

Connection Profiles. The user chooses from a collection of profiles based oncurrent working environment (LAN, Dial-up, or wireless connections, andTCP/IP, NetBios, etc.). Each profile contains a set of tuned parametersthat optimize the communication between the clients and the servers.

Data checkpointing. This ensures efficient recovery and fast reconnectionafter failures and involuntary disconnections.

Automatic reconnection in response to involuntary lost connections.

Data compression.

Dynamic Packet-Scaling. Based on the current connection quality andcapacity, data packets are dynamically re-sized to minimize connectiontime.

Encryption and authentication. Uses DES encryption for per-connectionauthentication.

Security. Forces re-authentication from the client upon receipt of anyunregistered packet.

Queuing. Application data is stored on both the client and the server in aclient assigned outbox until a connection is made to transfer the data.

Follow-Me Server. Uses a notification and delivery mechanism for eventssuch as arrival of data to the client’s outbox on the server. The user’smobile computer is notified if connected, and alternative notification pro-cedures (such as paging) are allowed.

MobileWare Corporation (http://www.mobileware.com) was founded in 1991,and is now a private subsidiary of Itochu Japan.

6.3.2 Shiva PPP

Shiva’s remote access client (known as PPP for Point-to-Point Protocol) enablesmobile users to access servers embedded in either wireline or mobile servers

Page 116: Mobile Computing

State of Industry: Mobility Support Software 95

almost seamlessly. For example, a client application that uses transaction pro-cessing services from BEA’s Tuxedo can now access those services from a mobileplatform using PPP. This software suite provides some limited security featuressuch as limiting the number of login tries, or disconnecting a session and call-ing the user back at a pre-established number. However, it does not providethe rich collection of services available from Mobile Ware’s Intelligent TransportEngine described above.

6.4 INTEROPERABILITY AND STANDARDIZATION

The wired infrastructure has been designed and deployed around a rich set ofinternational standards. For example, the legacy local area network (LAN)consists of such technologies as Ethernet, Token-Ring, and Token-Bus whichwere defined in precise details by the IEEE 802 committees. Moreover, newernetwork technologies like FDDI, HIPPI, and Fibre Channel have been definedby the ANSI X3T working groups with a mature set of approved specifica-tions. Both the IEEE and the ANSI bodies add further credibility to theirwork by helping international organizations like ISO and ITU to easily migratethe specifications into international standards bodies for worldwide acceptance.Similar efforts are underway in the ATM Forum to create a set of implementa-tion agreements which should permit interoperability between different vendorimplementations and products.

The wireless industry currently embraces a small number of standards. Theclosest effort is within the IEEE 802 working group which recently completedthe IEEE 802.11 Wireless MAC (media access control) standard. The primaryobjective of the IEEE 802.11 effort is to permit wireless LANs from differentvendors to interoperate.

IEEE 802.11 does not, however, address the needs of the wide area wirelessnetworking industry which currently deploys various packetized protocols (e.g.CDPD, GPRS) across unused cellular channels. Each network type is based onits own set of assumptions about the kinds of service the customers are willingto purchase. Service providers for each of these types of networks have differentgoals and strategies and do not seem likely to provide interoperability amongthe other classes of service.

Other mobile infrastructures are also lacking in internationally recognized stan-dards. This is evident in the cellular telephone industry: a PHS telephone

Page 117: Mobile Computing

96 CHAPTER 6

will not function in a cell serviced by a GSM or PCS infrastructure. Thesame is true for any combination of the aforementioned technologies. More-over, cordless telephones, infrared transmission, satellite channels, and mostmobile communication systems are either based on proprietary data interfaces,or have implemented selected parts of existing and/or emerging deploymentagreements. Such key attributes as Quality of Service, Location Register con-tents, Database formats, update policies, and data exchange rates are left tothe equipment providers and service providers which may be based more ondeployment schedules than on availability of standards and interoperabilityguarantees. The emerging UMTS system standard (discussed in Chapter 5),which is expected to be deployed by the year 2002, will provide a golden op-portunity for interoperability of data links interfaces, digital voice, and wirelessdata and services.

Many of the client-application products, or the communications substrate thatthey rely on, are recognizing that several competing wireless transmission pro-tocols exist with each network type. They also recognize that the number ofsuch protocols may grow or shrink. As a result, these client-level packages areadapted to use the popular underlying protocols. This limited form of inter-operability appears to meet the needs for developers of client software. Asan example, the Oracle Mobile Agents product discussed previously supportsboth CDPD and Shiva PPP. However, no client software we have seen claimsto migrate seamlessly among the different wireless network classes.

At some level, interoperability among the various network classes can be pro-vided by adopting popular communications standards. For example, thoseclient applications developed to exploit TCP/IP in wired networks can inter-operate without change in the wireless domain if some variant of TCP/IP isoffered as a service, e.g. lETF’s Mobile IP. However, the quality of service thatis provided by this approach may not be transparent, or even acceptable. Inaddition, it remains the client’s responsibility to transfer among the variouscompeting network services.

The heterogeneity of the existing and emerging wireless network protocols posesnot only a need for interoperability, but also a stringent quality of service re-quirements. This is because the inherent unreliability and bandwidth limitationlargely varies from one network to the other, leading to rapid fluctuations in thequality of the provided services. Recent research efforts proposed extensions toformal open systems standards.

The wireless application protocol (WAP ) standard currently being developedby the WAP forum group offers an OSI-like protocol stack for interoperability

Page 118: Mobile Computing

State of Industry: Mobility Support Software 97

of different wireless networks. The WAP stack allows applications to registerinterest in quality of service events and thresholds (QoS). This, in turn, allowsthe application to be mobility-aware and adaptable to changes in the environ-ment. The WAP stack also provides negotiation protocols between producersand consumers of data to optimize the necessary level of data presentationbased on the nature of data, the current wireless network between the sourceand the destination, and the capabilities of the destination device. Contentnegotiation should play a major role in maintaining QoS across heterogeneousnetworks.

Another proposal in [36, 47] extends the ISO Reference Model for Open Dis-tributed Processing (RM-ODP) so that clients are able to adapt to the variationof the network service they encounter. Under this proposal, applications willhave to be mobility-aware, but nonetheless will be able to maintain the requiredQoS.

6.5 SHORTCOMINGS AND LIMITATIONS

Mobility-support software and products that are commercially available todayleaves much to be desired in terms of functionality, performance, portability,and interoperability. The continuous decline in wireless communication costand the recent convergence towards a truly global and standard communicationsystem will help reduce the business risk associated with investments in thissoftware market. This will be true for giant software vendors as well as smallstartup companies. Several limitations and shortcomings of existing productsare summarized below.

Translucent Overlays. None of the product offerings or announced plansfor products that we have seen have included the vision of a translucentclient context for exploiting nomadic applications. Each client or middle-ware offering is tailored to a specific kind of network service and assumesthe client will manage its transition from one network class to another asthe mobile platform roams about.

Multi-Database Access. DBMS vendors such as Oracle and Sybasewhich are announcing mobile client products are providing connection ser-vices only to their proprietary DBMS product. This trend mirrors the paththat these vendors have adopted for the fixed network environment. Prod-ucts which have emerged for the wired environment which make it possible

Page 119: Mobile Computing

98 CHAPTER 6

for a client to interact with a variety of vendor DBMS, e.g. ODBC inter-faces, will not automatically extend service into the wireless domain. Thisfailure occurs because current solutions that couple mobile clients withserver DBMS include insertion of vendor-specific agents on both ends ofthe wireless connection to mediate wireless artifacts. Thus, a Mobile Or-acle client can only talk to those servers that are serviced by the MobileOracle server agents.

Mobile Transactions. We have seen no product that addresses issues oftransaction management in the mobile environment. The concept of a mo-bile transaction, where the locus of control of the transaction is maintainedby the mobile user, remains as a research idea needy of commercialization.BEA Systems markets a variety of products that couple wired clients witha variety of TP monitors, and implicitly to any of the DBMS productsthose TP monitors serve. If mobile transaction products will ever be madeavailable, they will be offered by companies such as BEA Systems andTransarc.

Workflows. Workflow products lag database access products in theirmigration to wireless and mobile environments. We have seen no workflowproducts targeted for the mobile domain in our literature searches.

Location Dependent Services. None of the server or client softwarepackages targeted for the wireless domain claim to offer location dependentservices. This important class of services will be essential at both ends ofthe wireless client-server communications link.

Page 120: Mobile Computing

7RESEARCH IN WIRELESS AND

MOBILE COMPUTING

An important requirement of the evolving information infrastructure is theseamless, ubiquitous, wireline or wireless connectivity that engenders contin-uous interactions between people and interconnected computers. This newrequirement, supported by amazing technological advances in small-size com-puters and wireless communication networks, has led to a proliferation of re-search in this area. This chapter describes some of the achieved and ongoingresearch in wireless and mobile computing. The following topics are covered inthis chapter:

Mobile networking

Quality of service in mobile networks

Mobile access to the World Wide Web

Mobile transactions

Mobile computing models

A complete and thorough coverage of research in wireless and mobile comput-ing is beyond the scope of this book. For more details, the reader is referred tothe Imielinski and Korth edited book, “Mobile Computing” [58] and the bookby James Solomon on Mobile-IP, “Mobile IP: The Internet Unplugged” [88].The reader is also encouraged to examine the proceedings of the ACM/IEEEInternational Conference on Mobile Computing and Networking (MobiCom 95,96, 97, 98 and 99), and the proceedings of the IEEE Workshop on Mobile Com-puting Systems and Applications (WMCSA 94 and 99). The ACM SIGMobileis also an important forum for research dissemination.

Page 121: Mobile Computing

100 CHAPTER 7

7.1 MOBILE NETWORKING

Internetworking mobile computers with the fixed-network raises the additionalrequirements of mobility transparency and mobility and location management.The mobility behavior of a node should be transparent to a peer node. Apeer node should be able to communicate with a mobile node using some fixedIP address irrespective of the current point of attachment. The mobile net-working protocol should also be transparent to the hosts and routers whichdo not understand or support mobility. Thus, the mobility unaware routersshould be able to route packets destined to a mobile host as normal IP datapackets. Security is another important concern in internetworking. In mo-bile networking it is more so, since the mobile nodes will be visiting foreignnetworks, requesting services, and accessing data. Thus, it is important thatthe security of the visiting network is not breached due to the presence of aforeign node in its network. Authentication of the mobile nodes and foreignnetworks is also important. Thus, at a minimum, mobile networking protocolsshould provide authentication and security features comparable to those foundin fixed-network IP protocols such as IPv4 and IPv6.

In this section, various approaches and protocols for mobile internetworkingare examined, including:

Early approaches: virtual IP mechanisms

Loose source routing protocol

The Mobile Internet Protocol (Mobile-IP)

Cellular Digital Packet Data (CDPD)

The General Packet Radio Service protocol (GPRS)

Emphasis is placed on protocol mechanisms, leaving out the details which canbe obtained by following cited work and web resources.

7.1.1 Early Approaches: Virtual IP Protocols

In this approach, a mobile network is a virtual network with a virtual addressspace. A mapping is maintained between the physical or actual IP addressesand the Virtual IP addresses. This mapping is performed by the mobile host

Team-Fly®

Page 122: Mobile Computing

Research in Wireless and Mobile Computing 101

which obtains a care of address from the local network being visited usingeither the Dynamic Host Configuration Protocol (DHCP) [38] or the BOOTP[51] protocols or by any of the link layer protocols. Below, we describe twoVirtual IP protocols.

7.1.1.1 Sunshine And Postel

The earliest solution for managing mobile hosts was proposed by Sunshine andPostel [89] in 1980. They proposed that the mobile hosts be assigned a virtualIP address which can be used to identify them. A mobile host in the foreignnetwork is required to obtain a care-of-address, and to update its location ina mapping database. When a packet has to be routed to the mobile host, itscurrent location is looked up in the database and the packet is transmitted tothat location.

7.1.1.2 The SONY Protocol

This protocol [34] was proposed in 1992 by F. Teraoka et al. of Sony Labora-tories. In this scheme, a mobile host has two IP addresses associated with it.A virtual address, which is immutable and by which it is known to the outsideworld, and a physical address, which is acquired from the local network. Twosublayers are introduced in the network layer and are used to map the physicaladdress to the virtual address. The transport layer interfaces with the networklayer through the virtual layer interface and addresses its packets to the virtualaddress of a mobile host. A cache called the Address Mapping Table (AMT )is used for fast address resolution. A copy of this cache is maintained at eachhost/router. The VIP (Virtual IP ) is implemented as an IP option. A set ofpacket types is also defined for host communication.

On entering a foreign network, the mobile host obtains an IP address andinforms its home network of its current location. The home network broadcaststhis information so the AMT cache gets updated. A stationary host, whenrequired to communicate with a mobile host, looks up its cache. If the mappingis available, the packet is transmitted in the normal fashion by appending theVIP header. If the cache entry is not available, the packet is addressed to theVIP address. A set of connection gateways are required for the co-existence ofmobility aware and mobility unaware hosts on the network.

7.1.2 Loose Source Routing Protocol

Page 123: Mobile Computing

102 CHAPTER 7

This approach was proposed by David Johnson [63] of CMU in 1993. It usesthe Loose Source Route option available in the IPv4 for routing packet data.The option allows the source to specify the intermediate gateways in the IPpacket. Thus, the source can control the route the IP packet takes. At eachdestination, the gateway picks up the next IP address from the IP packet, setsit as the destination, and advances a pointer stored in the IP packet header.

The home network maintains a database of all mobile hosts native to its net-work. When a mobile host changes location, it informs its home network ofits new location. When an IP packet destined to the mobile host arrives atthe home network, the packet is forwarded to the mobile host at the currentlocation address, and the corresponding source host is informed of the currentlocation of the mobile host. The corresponding host can use this information tocache the location, thus avoiding communication with the home network untilthe mobile host changes its location again. Source route set up is done by thecorresponding host.

7.1.3 The Mobile Internet Protocol (Mobile-IP)

The Mobile Internet Protocol (Mobile IP) [88, 80, 19, 7, 59, 76] defines en-hancements to the Internet Protocol to allow routing of IP packets to mobilenodes in the internet. The IP version 4 assumes that the Internet Protocol ofa node uniquely identifies the point of attachment of the node to the internet-work. Packets are routed based on the IP address. In a mobile environment,the point of attachment of the mobile node will be different from time to time,and the mobile nodes could be attached to different networks. For IPv4 towork correctly in the mobile environment, the mobile node will either haveto be assigned a new IP address every time it changes its point of attach-ment, or the host specific routing information has to be supplied throughoutthe network. Both of these alternatives result in scalability and connectionmanagement problems. The mobile IP protocol describes a mechanism whichallows nodes to change their point of attachment on the Internet.

The major architecture components of the mobile IP protocol are:

Mobile Node (MN): is a host or a router that changes its point of attach-ment to the network from one subnetwork to another. The MN is knownthroughout the network by an IP address assigned to it in the home net-work. The mobile node can communicate from any location as long as thelink layer connectivity to the internetwork is established.

Page 124: Mobile Computing

Research in Wireless and Mobile Computing 103

Home Agent (HA): is a mobile-IP capable router on the mobile node’shome network. The HA maintains the location information for the mobilenode. It also acts as the tunneling agent for packets destined to the mobilenode. The HA manages the registration and authorization information ofall the mobile modes belonging to its network.

Foreign Agent (FA): is a mobile-IP capable router that the mobile nodehas visited. After attaching to the foreign network, the mobile node isrequired to register itself with the FA. The FA detunnels and routes thepackets destined to the mobile node. The FA may also act as a defaultrouter for mobile nodes registered with it.

The mobile IP protocol can be summarized as follows:

1. The mobility agents (HA and FA) in the network broadcast their avail-ability through agent advertisement packets.

2. The mobile node, after connecting to a network, receives information aboutthe mobility agents through the agent advertisement broadcasts. Alterna-tively, the mobile nodes can solicit the agent information if no broadcastshave been received.

3. The mobile node determines the network it is attached to. If it is con-nected to the home network, it operates without mobility services. If itis returning back to the home network, the mobile node deregisters itselfwith the HA and operates without mobility services.

4. If the mobile node is attached to a foreign network, a care-of-address isobtained from the FA.

5. The mobile node operating from a foreign network registers itself with itshome agent. The foreign agent then acts as a relay in this registrationprocess.

6. When the mobile node is away from its home network, datagrams destinedto the mobile node are intercepted by the home agent, which then tunnelsthese datagrams to the mobile node’s care-of-address. The tunneled pack-ets destined to the mobile node are detunneled either by the foreign agentor by the mobile node itself. In the latter case, the mobile node obtainsa temporary IP address on the foreign agent network to be used for for-warding. This can be done using the IETF Dynamic Host ConfigurationProtocol (DHCP).

Page 125: Mobile Computing

104 CHAPTER 7

7. The datagrams originating from the mobile node are routed in the normalfashion. The foreign agent may act as a default router in this case.

The routing path of a datagram sent from a fixed host to a mobile node is asfollows: (1) the datagram is sent from the fixed host to the home agent usingstandard IP routing; (2) the home agent encapsulates the received datagraminside another datagram and sends it to the foreign agent (IP-in-IP tunneling[8]); (3) the encapsulated IP packet is received by the foreign agent, decapsu-lated, and forwarded to the mobile node; (4) the mobile node replies by sendinga datagram to the fixed host through the foreign agent. The Mobile IP protocolstack on the fixed network and on the mobile unit is depicted in Figure 7.1.

The Mobile Host Protocol, known as Mobile-IP, is an evolving standard beingdeveloped by the IETF Working Group on IP Routing for Wireless/MobileHosts. Standards for both IPv4 and IPv6 have been proposed and are beingreviewed for enhancements in scalability and performance. In particular, thetriangular routing between the mobile node, the home agent, and the foreignagent (that must be performed every time the mobile node switches over toanother communication cell) is a bottleneck that is being removed in IPv6 [81].Packets addressed to the mobile node’s home address are transparently routedto its care-of address. The optimized protocol enables IPv6 nodes to cachethe binding of a mobile node’s home address with its care-of address, and to

Page 126: Mobile Computing

Research in Wireless and Mobile Computing 105

then send any packets destined for the mobile node directly to it at this care-ofaddress.

The MosquitoNet project at Stanford [24] aimed at relaxing the requirementof foreign agent availability. MosquitoNet follows the IETF specification ofMobile-IP to support host mobility, but does not require FA support in foreignnetworks visited by the mobile node.

More details on achieved and ongoing efforts in Mobile IP and its routingoptimization can be found in [62, 60, 14, 11, 88, 80].

7.1.3.1 Support for Ad-Hoc Mobility

An ad-hoc mobile network is a collection of wireless mobile nodes forming atemporary network without the aid of any established infrastructure or central-ized administration. Examples of ad-hoc networks include wireless portabledevices of a group of collaborator, such as an emergency team in a disasterarea. No routing is needed between ad-hoc nodes which are within transmis-sion range of each others. Otherwise, additional nodes must be used to form asequence of hops from the source to the destination. Routing algorithms in thead-hoc environment are therefore a necessary support for this mode of mobileconnection.

Traditional routing algorithms used in wireline networks use distance vector orlink state routing algorithms, which rely on periodically broadcasting routingadvertisements by each router node. The distance vector algorithm [55] broad-casts its view of the distance from a router node to each host. The link staterouting algorithm [17] broadcasts its view of the adjacent network links. Neitheralgorithms is suitable for the ad-hoc environment because periodic broadcastswill drain battery power quickly.

Research in ad-hoc routing is dedicated to finding algorithms that avoid theneedless battery consumption and the inefficient use of the wireless bandwidth.Dynamic source routing is one such algorithms due to Johnson and Maltz [61].It allows for route discovery, route maintenance, and the use of route caches.To discover an available route, a source node sends out a route request packetindicating the source, the target nodes, and a request identifier. When a mo-bile node receives a route request packet, it checks a list of recently processedrequests. If a request is found for the same source and request id, the requestis dropped and no further action is taken. Otherwise, the address of the nodeservicing the request is added to the route request packet before the packet is

Page 127: Mobile Computing

106 CHAPTER 7

re-broadcasted. However, if the address of the node servicing the request isidentical to the target node address, the requested route is discovered, and areply is sent to the source node.

Due to unpredictable node mobility, cached routes may become incorrect.Route maintenance is therefore necessary in this environment. This is achievedby requiring nodes routing packets to acknowledge successful forwarding andto send error messages to the source node if a route ceases to exist. Activemonitoring such as MAC-level acknowledgements, as well as passive monitor-ing (listening to nearby broadcast, in a promiscuous mode), can be used inroute maintenance.

Other recent ad-hoc routing protocols that can be found in the literature includethe on-demand distance vector routing [82], the Location-Aided Routing (LAR )algorithm [71], and the Distance Routing Effect Algorithm [28].

7.1.4 Cellular Digital Packet Data (CDPD)

CDPD is a connectionless multi-network protocol, proposed originally by theCDPD Forum (now called the WDF Forum). It is based on the early versionsof Mobile-IP . The idea behind CDPD is to share unused channels in existingAdvanced Mobile Phone Systems (AMPS) to provide up to 19.2 kbps datachannel.

Even though CDPD and Mobile-IP are similar, their terminologies are differ-ent. CDPD follows the OSI model terminology. For example, the mobile nodeis called a Mobile End-System (M-ES); the home and foreign agents are calledMobile Home and Mobile Serving Functions (MHF and SF respectively) and re-side in a mobile data intermediate system (MD-IS ). A Mobile Database Station(MDBS) is also defined which deals with the air link communications and actsas a data link layer relay between the M-ES and the serving MD-IS . Two pro-tocols, the Mobile Node Registration Protocol (MNRP ) and the Mobile NodeLocation Protocol (MNLP ), are responsible for registration of the M-ES withits home MD-IS and the proper routing of packets destined for the M-ES.

The main resemblance between CDPD and Mobile-IP is in the triangular rout-ing approach between the mobile node and the home and foreign agents. Themain differences can be summarized as follows [6]:

Page 128: Mobile Computing

Research in Wireless and Mobile Computing 107

The user’s IP address must be assigned by the CDPD service provider.Mobile IP makes no such assumptions.

Mobile IP allows the mobile node to also be a foreign agent. Combiningthe M-ES and the Serving MD-IS was not considered and is not practicalin CDPD.

CDPD’s mobility tunnelling is based on CLNP. Mobile IP’s mobility tun-nelling is based on the IP-in-IP protocol, which is IP-based.

Mobile IP operates completely above the data link layer. CDPD mobility,on the other hand, is mostly above the data link layer.

Since the infrastructure of the CDPD network is closed there are less se-curity considerations for CDPD.

While the standardization process of Mobile IP has been progressing ratherslowly, CDPD has been deployed for a few years now, and is receiving thesupport of major AMPS carriers. However, due to its lack of openness, thefuture of CDPD deployment and/or acceptance can only be guessed.

7.1.5 The GSM General Packet Radio Service (GPRS)

GPRS is a GSM packet data service developed by the European Telecommu-nication Standards Institute (ETSI) as part of GSM phase developments.The goal of GPRS was to support data transfer rates higher than the 9.6 kbpsachieved through GSM’s circuit switching technology. Unlike Mobile-IP, GPRSis not restricted to IP packet data protocols, and offers connection to standardprotocols (such as TCP/IP, X.25, and CLNP) as well as specialized data packetprotocols. Mobile-IP, however, influenced the design of Mobility managementin GPRS.

Figure 7.2 shows the architecture of a GSM system that uses GPRS. In additionto the Base Transceiver Station (BTS), Base Station Controller (BSC), and theMobile Switching Center (MSC), a new logical network node called the GPRSsupport node (GSN) was introduced in order to create an end-to-end packettransfer mode. Physically, the GSN can be integrated with the mobile switchingcenter (MSC), or it can be a separate network element based on the architectureof data network routers. GSN is a mobility router that provides connectionand interoperability with various data networks, mobility management with

Page 129: Mobile Computing

108 CHAPTER 7

the GPRS registers, and delivery of data packets to MSs, independently oftheir locations.

One GSN is designated the Gateway GSN (GGSN) and acts as a logical inter-face to external packet data networks. The GGSN is similar to the home agentin Mobile-IP . It updates the location directory of the mobile station (MS) usingrouting information supplied by the Serving GSN node (SGSN). The latter issimilar to the foreign agent in Mobile-IP. GGSN also routes the external datanetwork protocol packet encapsulated over the GPRS backbone to the SGSNcurrently serving the MS. It also decapsulates and forwards external data net-work packets to the appropriate data network and handles the billing of datatraffic.

The SGSN is responsible for the delivery of packets to the mobile stationswithin its service area. The main functions of the SGSN are to detect newGPRS MSs in its service area, handle the process of registering the new MSsalong with the GPRS registers, send/receive data packets to/from the GPRSMS, and keep a record of the location of MSs inside of its service area. TheGPRS register acts as a database from which the SGSNs can ask whether anew MS in its area is allowed to join the GPRS network. For the coordinationof circuit and packet switched services, an association between the GSM MSCand the GSN is created. This association is used to keep routing and locationarea information up-to-date in both entities.

7.1.6 Security and Authentication Issues in Mobile Networks

In a mobile computing environment, it is desirable to protect information aboutthe movements and activities of mobile users from onlookers. In addition tothe basic security concerns in wireline systems (authentication, confidentiality,and key distribution), a new issue is the privacy and anonymity of the user’smovement and identity. In fact, a typical situation arises when a mobile userregisters in one domain (home domain) and appears in a different foreign do-main; the user must be authenticated and his solvency must be confirmed.Usually during this process the user has to provide a non-ambiguous identityto his home domain and has to verify it. If no care is taken, this identity can betapped on the air interface in a cellular environment or through the signalingprotocols exchanged on the registered wired network.

In CDPD , all the mobility management, as well as security-related activity, areconcentrated in the Massage-Data Intermediate System (MD-IS) . Each MD-

Page 130: Mobile Computing

Research in Wireless and Mobile Computing 109

IS controls an area covered by a number of base stations. Upon arrival to anew area, the mobile unit engages in a Diffie-Hellamn key exchange protocolwith the local MD-IS. As a result, both parties obtain a shared secret key.Subsequently, the mobile unit encrypts its real identity (Network EquipmentIdentifier) and transmits it to the local MD-IS. This approach allows the localMD-IS to discover the real identity of the mobile unit. Unfortunately, the keyexchanging protocol itself is not secure. This means that an active attackermasquerading as the local domain authority can engage in the key exchangeprotocol with the mobile unit and obtain a shared key.

7.2 QUALITY OF SERVICE IN MOBILE NETWORKS

Mobile network protocols such as Mobile-IP and GPRS provide mobility trans-parency at the network layer level. This allows the higher layers of the protocolstack to be used unchanged. Unfortunately, there are ill consequences to thistransparency that are mostly attributed to the constraints of the wireless andmobile environment. For example, transport layer protocols that rely heavilyon timeout mechanisms for re-transmission, if used unchanged, will performpoorly under variable delays and limited bandwidth. This is especially true forapplications that require continuous-media streams. Existing session protocolsare not of much use under frequent disconnections and reconnections of the

Page 131: Mobile Computing

110 CHAPTER 7

same mobile computation. Similarly, existing presentation layer protocols areinappropriate to use unchanged, in the wireless and mobile environment. Forexample, a user with a limited display and limited battery PDA will not be ableto browse the Web unless the presentation of the downloaded data is changedto suite her PDA’s capabilities. Regardless of which particular upper layer inthe protocol stack suffers the consequences of transparency, the effect on theend-user will always be felt as unacceptable fluctuations in the perceived QoS.

In this section, we describe the following three research efforts that addressQoS concerns in the wireless and mobile environment.

Optimizing TCP/IP for Mobile Networks. Transport or network layer so-lutions to get TCP/IP to work despite the fluctuations in the underlyingnetwork QoS. Solutions are not application-sensitive and do not addressan overlay of heterogeneous networks.

QoS driven, high-level communication protocols. Session and/or applica-tion layer protocols directly addressing QoS parameters. Solutions aresensitive to applications, but do not address network heterogeneity issues.

QoS driven, full protocol stacks. All layers are aware of either QoS orthe limitations introduced by mobility and by the wireless networks. Tworesearch efforts will be discussed including, the BARWAN project and theWireless Application Protocol (WAP) standard.

7.2.1 Optimizing TCP/IP for Mobile Networks

Since mobile users will need connection-oriented communication to obtain re-mote services, they will have to use transport protocols developed for the fixednetwork. Unfortunately, such protocols like TCP perform poorly when used un-modified in the mobile network. For example, TCP acknowledgment timeoutis in the range of tens of milliseconds. A mobile unit crossing cell boundariesblanks out during a hand-off procedure that could last up to 1,000 milliseconds.This leads to sender timeouts and repeated re-transmissions. Another sourceof re-transmission is the high error rate inherent in the wireless transmissioncharacteristics. Another problem that can lead to performance degradation un-der standard TCP is bandwidth allocation under unpredictable mobility. Anunpredicted number of mobile users can move into the same cell, thus compet-ing on sharing the limited wireless link. Under this scenario, it is difficult tobuild applications or services that provide performance guarantees or quality

Team-Fly®

Page 132: Mobile Computing

Research in Wireless and Mobile Computing 111

of service. A few approaches have been proposed to optimize and extend thestandard TCP protocol so that it can be used efficiently under a mobile networkprotocol such as Mobile IP.

7.2.1.1 Yavatkar et al

Yavatkar et al [92] proposed an approach whereby the communication path be-tween the mobile end and the fixed end is split into two separate connections:one over the wireless link and another over the wired links. The connection overthe wireless link may either use regular TCP or a specialized transport protocoloptimized for better performance. The splitting of a connection is transparentto an application and no changes are necessary to protocol software on the sta-tionary hosts. A new session layer protocol called Mobile Host Protocol (MHP)is introduced atop standard TCP. MHP compensates for wireless link charac-teristics and for host migration. It is located at both the base station and themobile host. An advantage of this approach is that performance degradation inTCP is limited to a “short” connection over the wireless hop, while traffic overthe “long” connection over the wired network can be protected from the impactof erratic behavior over the wireless link. A second alternative is proposed inthe same work which is similar to the MHP alternative except that MHP usesa specialized protocol instead of TCP over the wireless hop. The specializedprotocol differs from standard TCP in that the former uses selective acknowl-edgement by the receiver, in which a bitmask is used to indicate all missingsegments of the connection stream. This way, the recovery of all losses canbe performed via a single round trip message, resulting in a better throughputperformance.

Another approach similar to Yavatkar’s is the I-TCP protocol (Indirect Trans-port Layer Protocol) [25], which also splits the communication path betweenthe mobile host and the fixed network host into two connections; the first be-tween the mobile host and the base station, over the wireless link, using theI-TCP protocol; and the second between the base station and the fixed networkhost using standard TCP.

7.2.1.2 Balakrishnan et al

Balakrishnan et al [26] took a slightly different approach to improve the per-formance of TCP in the mobile network. They focused on the re-transmissionbehavior of TCP due to hand-off. They redesigned the network layer so that itcaches packets at the base stations. Retransmission can therefore be performed

Page 133: Mobile Computing

112 CHAPTER 7

locally between the base station and the mobile unit. The gain is that the er-ratic transmission characteristics of the wireless link are dealt with in isolationof the rest of the fixed network. Experimental evaluation showed a throughputincrease of up to 20 times over standard TCP. Their results are based on theLucent/NCR Wavelan network.

7.2.1.3 Caceres et al

Similar research by Caceres and Iftode [31] addressed the problem of communi-cation pauses due to hand-off. They observed that such pauses are interpretedby standard TCP (Tahoe in their experiment) as packet losses due to conges-tion, which consequently causes retransmissions that get further timed out dur-ing the hand-off. They proposed using the fast re-transmission option availablein TCP-Tahoe immediately after hand-off is completed. Their experimentalverification shows clear smoothening of TCP performance during hand-off.

7.2.2 QoS Driven, High-Level Communication Protocols

Optimizing the behavior and performance of transport protocols is not suffi-cient to maintain the QoS required by applications. For example, most Webbrowsers use multiple TCP connections to access a multimedia page. While thisparallelism achieves speedup in the fixed network, it is slow and inappropriatein the wireless and mobile environment. In addition to transport optimiza-tions, what was found needed are application-aware (or application-specific)mechanisms to monitor, request, and maintain QoS from the application oruser point of view. This section describes high-level, above-transport protocolsthat understands application QoS requirements and resource limitations.

7.2.2.1 The Loss Profile Approach

Seal and Singh [87] considered the problem of unpredictable mobility and itseffect on the degradation of the wireless communication performance. Theyaddressed the case where the aggregate bandwidth required by all mobile unitsin an overloaded cell exceeds the cell’s available bandwidth. Their mechanismis simple and relies on policies and measures for discarding parts of the data ofthe mobile users. Instead of discarding data in an arbitrary manner, guidelinesare proposed to avoid discarding critical portions of the data. A Loss Profile isproposed and is defined to be a description, provided by the application, of an“acceptable” manner in which data for its connection may be discarded. The

Page 134: Mobile Computing

Research in Wireless and Mobile Computing 113

loss profile is used in the event of bandwidth reduction at the wireless end of theconnection. An elaborate example of a loss profile is given on viewer perceptionof a video clip under data loss. The loss profile is used by a specialized sessionlayer which is transparent to the application.

7.2.2.2 QEX: The QoS Driven Remote Execution Protocol

In [36, 47], the problem of fluctuations in the quality of service (QoS) in a fed-eration of heterogeneous networks is addressed. The work describes a designof a distributed system platform that supports the development of adaptableservices. The design allows services to tolerate the heterogeneity of the envi-ronment by dynamically adapting to changes in the available communicationQoS. The implementation of the distributed system is based on APM Ltd.’sANSAware software suite, which is based on the ANSA architecture that hashad some influence on the ISO Reference Model for Open Distributed Process-ing (RM-ODP). The purpose of this effort is to propose extensions to emergingdistributed systems standards in order to support mobile services. The basicANSAware platform is extended to support operation in the mobile environ-ment by introducing the notion of explicit bindings, which is a QoS-aware RPCprotocol for objects called QEX.

Explicit bindings allow application programmers to specify QoS constraintson bindings between objects, and to detect violations of these constraints atrun time. To support explicit bindings, a new remote procedure call protocolhas been developed for ANSAware. The new RPC is able to maintain QoSinformation on the underlying communications infrastructure and to adapt tochanges in the perceived QoS. Moreover, it is able, via explicit bindings, topass on relevant QoS information to interested applications. This allows theapplications themselves to adapt to changes in the QoS. Binding parametersinclude specification of parameters such as the desired throughput, latency,and jitter associated with the binding. Clients are returned a binding controlinterface as a result of an explicit bind operation. To control the QoS of theflow once the binding has been established, the control interface includes apair of operations setQoS() and getQoS(). These operations take as argumentsa set of QoS parameters which can then be passed by the stream binding tothe underlying transport protocol. A call-back mechanism is also provided toinform client objects of QoS degradations reported by the underlying transportservice.

The work is being put to test using an adaptive collaborative mobile applicationdesigned to support field engineers in the U.K. power distribution industry.

Page 135: Mobile Computing

114 CHAPTER 7

7.2.3 QoS Driven, Full Protocol Stacks

Future mobile services will be built upon federations of heterogeneous networksmaintained and administered by different providers. The mobility of users willforce an application to migrate along overlays of networks that vary in theirbandwidth, latency, range, and transmission characteristics. Unless the appli-cation adapts to variations in the network overlay, the application performanceis bound to suffer. A network overlay can include a cellular network, a personalcommunication system (PCS), a wireless LAN, an Internet connection, and/ora satellite communication loop, among other networks. In addition to the het-erogeneity of networks, the heterogeneity of the mobile platforms imposes agreat impediment to mobile application portability. Unless applications adaptto the capabilities and limitations of the mobile computer with respect to thetype and media of communicated data, applications will remain proprietary tothe specific mobile computer platforms they were originally designed for. Thissection describes a research project that proposes a full stack solution as anoverlay network stack atop a heterogeneous collection of wireless subnets. Thissection also describes an ongoing standardization effort called WAP that aimsat proposing a specification of a full ISO/OSI-like network stack that is wirelessand mobile aware.

7.2.3.1 BARWAN: The Wireless Overlay Network Architecture

The BARWAN project [68] at the University of California at Berkeley de-veloped an architecture that supports applications’ graceful adaptation to theavailable bandwidth and latency of the wireless network. The architectureassumes an overlay of various wireless networks ranging from regional-area,wide-area, metropolitan-area, campus-area, in-building, and in-room wirelessnetworks. A testbed of wireless overlay network management that supportsmedia-intensive applications has been used to demonstrate the adaptability fea-tures of BARWAN. The testbed that has been developed in the San FranciscoBay Area includes the participation of over six local carriers including Nexteland Metricom. The testbed integrates the participants’ networks and allowsfull coverage of the greater Bay Area. The BARWAN architecture is gateway-centric, meaning it provides gateway connections from the mobile host to eachparticipating wireless networks. Medical imaging applications have been devel-oped to drive the testbed.

The layered architecture of BARWAN is shown in Figure 7.3. It shows all layersdesigned for wireless overlay network integration and for providing applicationsupport. The lowest layer is the wireless overlay subnets, which are the car-

Page 136: Mobile Computing

Research in Wireless and Mobile Computing 115

Page 137: Mobile Computing

116 CHAPTER 7

rier networks including data link interface, and possibly carrier network rout-ing. The details of this layer depends on the specific subnets being integrated.Next is a layer called the Overlay Network Management Layer which includesnetwork and transport functionalities including location tracking, QoS-basedhand-off management, other QoS services, and connection-oriented transportmechanisms. The next higher up layer is the Session Management Layer whichprovides a “transactional” transport (called message-oriented interface). Thelayer attempts to optimize transport connections related to the same appli-cation by session sharing whenever possible. On top of the session layer isthe Application Support Services including support for various data types andcontinuous media such as audio and video. Finally, the mobile multimedia ap-plication is on top of the stack. Figure 7.3 also shows how the quality of servicesneeds pass down the layers from applications towards the network managementlayers, while information about network capabilities propagates up the layers.

7.2.3.2 The Wireless Application Protocol (WAP)

In June 1997, Ericsson, Nokia, Motorola, and Phone.Com (previously UnwiredPlanet) formed a consortium for the standardization of an open middlewarearchitecture for wireless application. The objective was to create the specifica-tion of a wireless application environment and a wireless ISO/OSI-like protocolstack. The goal was to provide the needed interoperability to connect dif-ferent portable devices, via heterogeneous wireless networks, into the internetand corporate intranets. The focus was to bring the internet content and ad-vanced services to digital cellular phones and other hand-held devices such assmart communicators and PDAs. In January 1998, the consortium created anonprofit company named the WAP Forum with the mission of enabling: (1)interoperability across heterogeneous portable devices, wireless networks, andinternet contents, and (2) portability of third party wireless software and appli-cations across different portable devices that are WAP-compliant. Currently,the WAP Forum is creating a set of specifications for the Wireless ApplicationEnvironment and for each layer in the WAP protocol stack.

The architectural infrastructure of WAP is depicted in Figure 7.4 and consistsof: (1) hand-held devices ranging from digital cellular phones, to smart com-municators such as the Nokia 9000, to palmtop computers. Only devices thatwill be WAP-compliant (implement the WAP stack and wireless applicationenvironment) are part of the WAP infrastructure, (2) Wap-compliant wirelessnetworks, which are carrier networks augmented with the WAP stack on top ofthe air link interfaces, (3) WAP-compliant internet information providers suchas Web servers, that must conform to levels of presentations of information

Page 138: Mobile Computing

Research in Wireless and Mobile Computing 117

suitable to the capabilities of the hand-held device requesting the information,and (4) WAP-compliant TeleVAS providers.

In the heart of the WAP standard is the WAP protocol stack shown in Fig-ure 7.5. The stack is similar to the ISO/OSI stack and consists of a lowestlayer containing air link interfaces such as GSM’s GPRS, CDPD, D-AMPS,among others. This lowest layer corresponds to both the physical and the datalink layers combined in the OSI stack. On top of the air link is the transportlayer, in which datagram and connection-oriented streams are supported. Also,transactional connections are supported to enable electronic commerce appli-cations. This layer corresponds to both the network and the transport layersof the OSI stack combined. On top of the transport, WAP dedicates a layer forsecurity. This includes encryption, authentication, and capabilities. On top ofsecurity is the session layer which is responsible for enabling multi-tasking onthe hand-held device. This is because multiple connections can be maintainedas multiple sessions managed by the session layer. The session layer, which isthe most elaborate layer, also contains critical QoS features including:

exception mechanisms to allow applications to register interest in QoS-related network events and parameter thresholds. This allows the applica-

Page 139: Mobile Computing

118 CHAPTER 7

tion to be mobility-aware, by using QoS API to program how to adapt tochanges in the environment.

mechanisms for capability and content negotiation. This will enable theWAP stack itself to partner through its pieces (on the fixed network, on thewireless network gateways, and on the hand-held device) to perceive andadapt to the mobility and the changes in network characteristics. Whencertain information content is being delivered, the WAP stack negotiateswith the device the capability to receive and display the contents. Thenegotiation decides for the feasibility of the transfer and for the level offiltering that might be needed to deliver the the information while main-taining QoS.

The first capability provides applications with the environment awareness neededto initiate QoS adaptations. The second capability, on the other hand, providesthe system with automated awareness mechanisms not only of the environment,but also of the device capabilities and the characteristics of the informationcontent.

Another standardization effort similar to WAP is the Mobile Network ComputerReference Profile (MNCRF), which is based on the NCRF standard developed

Page 140: Mobile Computing

Research in Wireless and Mobile Computing 119

by the Open Group [12]. The first draft of the standard has been releasedaddressing the unique requirements of mobile network computing. Details ofthis initiative are available in a white paper and a reference specification doc-ument [10].

7.3 MOBILE ACCESS TO THE WORLD WIDE WEB

More and more users are becoming increasingly dependent on information theyobtain from the World Wide Web. Users are also demanding ubiquitous access,any time, anywhere, to the information they rely on. Several research efforts ex-plored the problems associated with wireless access to the Web. Most solutionsused a Web proxy that enabled Web browsing applications to function overwireless links without imposing changes on browsers and servers. Web proxiesare also used to prefetch and cache Web pages to the mobile client’s machine,to compress and transform image pages for transmission over low-bandwidthlinks, and to support disconnected and asynchronous browsing operations.

7.3.1 The Wireless WWW (W4)

In [27], a prototype consisting of commercially available PDAs and a wirelessLAN has been used to provide a “proof of concept” for the Wireless WorldWide Web (W4). A simplified version of Mosaic was ported to the PDA forthe purpose of experimenting with response time performance and to sort outdesign choices. A PDA cache was used to improve the performance.

7.3.2 Dynamic Documents

The concept of dynamic documents was introduced in [66] as an approach toextending and customizing the WWW for mobile computing platforms. Dy-namic documents are programs executed on a mobile platform to generate adocument; they are implemented as Tcl scripts as part of the browser client.A modified version of the NCSA Mosaic browser was used to run the dynamicdocuments it retrieves through a modified Tcl interpreter. The interpreter isdesigned to execute only commands that do not violate safety. By using dy-namic documents, an adaptive e-mail browser that employs application-specificcaching and prefetching is built. Both the browser and the displayed e-mailmessages are dynamically customized to the mobile computing environment in

Page 141: Mobile Computing

120 CHAPTER 7

which they run. Dynamic documents can solve the problem of limited resourcesin the mobile host. For example, the Tcl script could be a filter that reducesan incoming image so that it fits the screen size or resolution. Unfortunately,dynamic documents being placed at the client side are not wireless-media sen-sitive. This is because filtering occurs after all transmitted information is re-ceived by the client. Although caching and prefetching can alleviate some of thecommunication overhead, excess data (that would be reduced by the dynamicdocument) is, however, communicated, leading to inefficient utilization of thewireless bandwidth.

7.3.3 Dynamic URLs

The Mobisaic project [90] at the University of Washington extends standardclient browsers to support dynamic URLs and active documents. The MosaicWeb client and the URL syntax are modified so that when the user traversesa dynamic URL, the client resolves any references to dynamic information itmay contain and sends the result back to the server. This is helpful in defininglocation-sensitive resources. Active documents are Web pages that notify theclient browser when dynamic information changes. This feature also supportslocation-sensitive information by keeping the mobile client aware of servicerelocation or of services offered by a mobile server.

7.3.4 Mobile Browser (MOWSER)

In [65], a design of a mobile-aware Web browser is discussed. The design isbased on a mediator server that filters retrieved information according to thelimitations of the mobile unit. Color, resolution, display mode, sound capa-bility, and maximum file size are among the factors considered. The browser,called MOWSER, connects to two servers in the fixed network. The first is thepreference server that maintains the user profile; the second is a proxy serverthat implements all the filtering indicated by the preference server. MOWSERassumes that the user is aware of the mobile unit limitations, which in a waysacrifices transparency. Similar to the dynamic document approach, MOWSERdoes not directly consider the limitations of the wireless media (although themaximum file size indirectly preserves the limited bandwidth).

Team-Fly®

Page 142: Mobile Computing

Research in Wireless and Mobile Computing 121

7.3.5 WebExpress

WebExpress [57] uses the proxy approach to intercept and control communi-cations over the wireless link for the purposes of reducing traffic volume andoptimizing the communication protocol to reduce latency. Two componentsare inserted into the data path between the Web client and the Web server: (1)the Client Side Intercept (CSI) process that runs in the client mobile deviceand (2) the Server Side Intercept (SSI) process that runs within the wired andfixed network (see Figure 7.6).

The CSI intercepts HTTP requests and, together with the SSI, performs opti-mizations to reduce bandwidth consumption and transmission latency over thewireless link. From the viewpoint of the browser, the CSI appears as a localWeb proxy that is co-resident with the Web browser. On the mobile host, theCSI communicates with the Web browser over a local TCP connection (usingthe TCP/IP “loopback” feature) via the HTTP protocol. Therefore, no exter-nal communication occurs over the TCP/IP connection between the browserand the CSI. No changes to the browser are required other than specifying the(local) IP address of the CSI as the browser’s proxy address. The CSI com-municates with an SSI process over a TCP connection using a reduced versionof the HTTP protocol. The SSI reconstitutes the HTML data stream and for-wards it to the designated CSI Web server (or proxy server). Likewise, forresponses returned by a Web server (or a proxy server), the CSI reconstitutesan HTML data stream received from the SSI and sends it to the Web browserover the local TCP connection as though it came directly from the Web server.

Page 143: Mobile Computing

122 CHAPTER 7

The proxy approach implemented in WebExpress offers the transparency ad-vantage to both Web browsers and Web servers (or proxy servers) and, there-fore, can be employed with any Web browser. The CSI/SSI protocols facilitatehighly effective data reduction and protocol optimization without limiting anyof the Web browser functionality or interoperability. WebExpress optimizationmethods are summarized below:

Caching: Both the CSI and SSI cache graphics and HTML objects. Ifthe URL specifies an object that has been stored in the CSI’s cache, it isreturned immediately as the response. The caching functions guaranteecache integrity within a client-specified time interval. The SSI cache ispopulated by responses from the requested Web servers. If a requestedURL received from a CSI is cached in the SSI, it is returned as the responseto the request.

Differencing: CSI requests might result in responses that normally varyfor multiple requests to the same URL (e.g., a stock quote server). Theconcept of differencing is to cache a common base object on both the CSIand SSI. When a response is received, the SSI computes the differencebetween the base object and the response and then sends the difference tothe CSI. The CSI then merges the difference with its base form to create thebrowser response. This same technique is used to determine the differencebetween HTML documents.

Protocol reduction: Each CSI connects to its SSI with a single TCP/IP con-nection. All requests are routed over this connection to avoid the costlyconnection establishment overhead. Requests and responses are multi-plexed over the connection.

Header reduction: The HTTP protocol is stateless, requiring that eachrequest contain the browser’s capabilities. For a given browser, this infor-mation is the same for all requests. When the CSI establishes a connectionwith its SSI, it sends its capabilities only on the first request. This infor-mation is maintained by the SSI for the duration of the connection. TheSSI includes the capabilities as part of the HTTP request that it forwardsto the target server (in the wire line network).

Page 144: Mobile Computing

Research in Wireless and Mobile Computing 123

7.4 MOBILE DATA MANAGEMENT

Mobile data access can be broadly classified into two categories: (1) data accessin mobile client/server, and (2) data access in ad-hoc networks. Several researchprojects from each category are presented in the following subsections.

7.4.1 Mobile Client/Server Data Access

In the first category, mobile data access enables the delivery of server data andthe maintenance of client-server data consistency in a mobile and wireless envi-ronment. Efficient and consistent data access in mobile environments is a chal-lenging research area because of the weak connectivity and resource constraints.The data access strategies in a mobile information system can be characterizedby delivery modes, data organizations, and consistency requirements, amongother factors. The mode for server data delivery can be server-push, client-pull,or a hybrid of both. The server-push delivery is initiated by server functionsthat push data from the server to the clients. The client-pull delivery is initi-ated by client functions which send requests to a server and “pull” data fromthe server in order to provide data to locally running applications. The hybriddelivery uses both server-push and client-pull delivery. The data organizationsinclude mobility-specific data organizations like mobile database fragments inthe server storage and data multiplexing and indexing in the server-push deliv-ery mode. The consistency requirements range from weak consistency to strongconsistency.

7.4.1.1 Broadcast Disks: A Server PUSH Approach

When a server continuously and repeatedly broadcasts data to a client commu-nity, the broadcast channel becomes a “disk” from which clients can retrievedata as it goes by. The broadcasting data can be organized as multiple disksof different sizes and speeds on the broadcast medium [22]. The broadcastis created by multiplexing chunks of data from different disks onto the samebroadcast channel. The chunks of each disk are evenly interspersed with eachother. The chunks of the fast disks are repeated more often than the chunksof the slow disks (see Figure 7.7). The relative speeds of these disks can beadjusted as a parameter to the configuration of the broadcast. This use of thechannel effectively puts the fast disks closer to the client while at the same timepushing the slower disks further away.

Page 145: Mobile Computing

124 CHAPTER 7

7.4.1.2 Odyssey: A Client PULL Approach

Odyssey is a CMU research project led by M. Satyanarayanan [86, 77]. It ad-dresses an application-aware adaptation approach to deal with application di-versity and concurrency in mobile environments. The application-aware adap-tation is implemented with the support of system-coordinated, type-specificoperations. It supports concurrent execution of diverse mobile applicationsthat execute on mobile clients, but read or update remote data on servers. Thedata accessed by an application may be stored in one or more general-purposerepositories such as file servers, SQL servers, or Web servers. It may also bestored in more specialized repositories such as video libraries, query-by-image-content databases, or back–ends of geographical information systems.

Ideally, a data item available on a mobile client should be indistinguishablefrom that available to the accessing application if it were to be executed on theserver storing that item. But this correspondence may be difficult to preserveas mobile resources become scarce; some form of degradation may be inevitable.In Odyssey, fidelity is used to describe the degree to which data presented at aclient matches the reference copy at the server. Fidelity has many dimensions.One well-known, universal dimension is consistency. For Video applications,data has at least two additional dimensions: frame rate and image quality ofindividual frames. Odyssey provides a framework within which diverse notionsof fidelity can be incorporated.

Page 146: Mobile Computing

Research in Wireless and Mobile Computing 125

7.4.1.3 Rover: A Mobile Objects Approach

The Rover project at MIT [64] provides mobility support to client server ap-plications based on two ideas: relocatable dynamic object (RDO) and queuedremote procedure calls (QRPC). An RDO is an object (code and data) witha well-defined interface that can be dynamically loaded into a mobile clientfrom a server computer, or vice versa, to reduce client-server communicationrequirements, or to allow disconnected operation. Queued remote procedurecall is a communication system that permits applications to continue to makenon-blocking remote procedure calls even when a mobile client is disconnected;requests and responses are exchanged upon network reconnection. Rover givesapplications control over the location where the computation is performed. Bymoving RDOs across the network, applications can automate the movement ofdata and/or computation from the client to the server and vice versa.

7.4.2 Mobile Data Access in Ad-hoc Networks

The Bayou project [37] at Xerox PARC developed a system to support datasharing among mobile users. The system is intended to support ad-hoc mobil-ity, where no network infrastructure is assumed to be available. In particular,a user’s mobile computer may experience extended disconnection from othercomputing devices. Bayou allows mobile users to share their appointment calen-dars, bibliographic databases, meeting notes, evolving design documents, newsbulletin boards, and other types of data in spite of their intermittent networkconnectivity. The Bayou architecture supports shared databases that can beread and updated by users who may be disconnected from other users, eitherindividually or as a group. Bayou supports application-specific mechanismsthat detect and resolve the update conflicts, ensures that replicas move to-wards eventual consistency, and defines a protocol by which the resolution ofupdate conflicts stabilizes. Bayou includes consistency management methodsfor conflict detection called dependency checks and per-write conflict resolutionbased on client-provided merge procedures. To guarantee eventual consistency,Bayou servers are able to rollback the effects of previously executed writes andredo them according to a global serialization order. Furthermore, Bayou per-mits clients to observe the results of all writes received by a server, includingtentative writes whose conflicts have not been ultimately resolved.

In the Bayou system, each data collection is replicated in full at a number ofservers. Applications running as clients interact with the servers through theBayou API, which is implemented as a client stub bound with the application.

Page 147: Mobile Computing

126 CHAPTER 7

This API, as well as the underlying client-server RPC protocol, supports twobasic operations: Read and Write. Read operations permit queries over a datacollection, while Write operations can insert, modify, and delete a number ofdata items in a collection.

7.5 MOBILE TRANSACTIONS

A mobile transaction is a long-live transaction whose locus of control movesalong with the mobile user. Mobile transactions may access remote data wire-lessly, through a weak connection, or may access local replicas of data in dis-connected mode. The differences between mobile and distributed transactionmanagement are significant because their goals are different. In distributedtransactions, the main goal is maximizing availability while achieving ACIDproperties. In mobile transactions, maximizing reliability while achieving somesort of consistency is the main goal.

In this section we describe some of the existing approaches to mobile transactionmanagement. The transaction models considered here have been proposed byChrysanthis [33], Dunham et al. [39], Pitoura et al. [43, 44], Satyanarayananet al. [75], Gray et al. [52], Walborn et al. [49], and Nielsen [35]. All themodels described assume the mobile computing reference model presented inFigure 1.6 of chapter 1.

7.5.1 Reporting and Co-Transactions

This model [33] is based on the Open Nested transaction model. A computationin the mobile environment is considered to consist of a set of transactions, someof which may execute on the mobile node and some of which may execute on thefixed host. The model addresses sharing of partial results while in execution,and maintaining computation state in a fixed node so that the communicationcost is minimized while the mobile host relocates.

The model proposes to modify Reporting and Co-Transactions [32, 33] to suitmobile environments. The model defines a mobile transaction to be a set ofrelatively independent transactions which interleave with other mobile transac-tions. A component transaction can be further decomposed into other compo-nent transactions allowing arbitrary levels of nesting. Component transactionsare allowed to commit or abort independently. If a transaction aborts, all com-

Page 148: Mobile Computing

Research in Wireless and Mobile Computing 127

ponents which have not yet committed may abort. Some of the transactionsmay have compensating duals and may be compensated. The model classifiesmobile transactions into the following four types:

Atomic transactions: normal components and may be compensatable withatomic compensating dual steps.

Compensatable transactions: atomic transactions whose effects cannot beundone at all. When ready to commit, the transaction delegates all oper-ations to its parent. The parent has the responsibility to commit or abortthe transaction later on.

Reporting transactions: can make its results available to the parent at anypoint of its execution. It could be a compensating or a non-compensatingtransaction.

Co-transactions: behave in a manner similar to the co-routine construct inprogramming languages. Co-transactions retain their current status acrossexecutions, hence they cannot be executed concurrently.

A reporting transaction reports its results to other transactions by delegatingthe results. A reporting transaction can have only one recipient at any givenpoint of time. The changes made by a reporting transaction is made perma-nent only when the receiving transaction commits. If the receiving transactionaborts, the reporting transaction aborts as well. A co-transaction, on the otherhand, reports its results in a way similar to reporting transactions. But upondelegation, the transaction stops execution and is resumed from the point itleft off. For any pair of co-transactions, either both commit or both abort.

7.5.2 The Kangaroo Transaction Model

This model introduced in [39] is based on the global transactions and the splittransaction models, where transaction relocation is achieved by splitting thetransaction at the point of hand-off. A mobile transaction (called Kangarootransaction) is considered a global transaction in a multidatabase environment.A Kangaroo transaction (KT) is a global transaction that consists of a set ofJoey Transactions (JT). A JT is associated with the base station or the cellin which it executes. When the mobile unit moves to a new cell, the JT inthe previous cell is split, and one of the JTs is moved to the current cell ofthe mobile unit. Each JT may consist of a set of local and global transactions.

Page 149: Mobile Computing

128 CHAPTER 7

The model is built upon the existing databases. The transactions are micro-managed by the individual database transaction managers. A Joey Transactionshould terminate in an abort, commit, or a split. For a KT to be successful,the last JT in the order of execution should end in a commit or abort, whereasall other JTs should be split. Based on the ability to compensate the splittransaction component, a KT can be executed as a whole atomic transaction, orin a relaxed mode where only component transactions are executed atomically.

7.5.3 The Clustering Model

This model described in [44] assumes a fully distributed system. The databaseis divided into clusters. A cluster defines a set of mutually consistent data.Bounded inconsistencies are allowed to exist between clusters. These inconsis-tencies are finally reconciled by merging the clusters. The model is based onthe open nested transaction model, extended for mobile computing. A trans-action submitted from a mobile host is composed of a set of weak and stricttransactions. Transaction proxies are used to mirror the transactions on indi-vidual machines as they are relocated from one machine to another. A clusteris defined as a unit of consistency in that all data items inside a cluster arerequired to be fully consistent, while data items residing in different clustersmay exhibit bounded inconsistency.

Clusters can be defined either statically or dynamically. A wide set of parame-ters can be used for defining clusters. This could include the physical locationof data, data semantics, and user definitions. Consistency between clusters canbe defined by an m-degree relation, and the clusters are said to be m-degreeconsistent. The m-degree relation can be used to define the amount of deviationallowed between clusters. In this model, a mobile transaction is decomposedinto a set of weak and strict transactions. The decomposition is done basedon the consistency requirement. The read and write operations are also classi-fied as weak and strict. The weak operations are allowed to access only dataelements belonging to the same cluster, whereas strict operations are alloweddatabase-wide access. For every data item, two copies can be maintained–oneof them strict and the other weak. As mentioned above, a weak operation canaccess only the local copies of a data item. Weak operations are initially com-mitted in their local clusters. When the clusters are finally merged, they areonce again committed across the clusters.

Page 150: Mobile Computing

Research in Wireless and Mobile Computing 129

7.5.4 Isolation-Only Transactions

The Coda file system [70] at CMU provides an application-transparent file sys-tem for mobile clients by using file hoarding and optimistic concurrency control.A proxy logs all updates to the file system during disconnection and replaysthe log on reconnection. Automatic mechanisms for conflict resolution are pro-vided for directories and files through the proxy and the file server. Hoardingis based on user-provided, prioritized list of files. Periodically, the proxy walksthe cache to ensure that the highest priority files are present and consistentwith the server. Coda provides Isolation-only Transactions (IOT) [75] to au-tomatically detect read/write conflicts that could occur during disconnection.Unlike traditional transactions, it does not guarantee failure atomicity and onlyconditionally guarantees permanence.

The SEER hoarding system [73] developed at UCLA is based on the Codafile system. It operates without user intervention by observing user activitiesand predicting future needs. It defines and uses a measure called “semanticdistance” between files to determine how best to cluster files together in prepa-ration for hoarding. The semantic difference between two files is based on thetime elapsed between the events of opening the files, and on how many refer-ence to other files occurs in between. SEER does not actually hoard files, butinstead interfaces with Coda (and other replicated systems) to do the hoarding.SEER also detects hoard misses during disconnection.

7.5.5 The Two-tier Transaction Model

A two-tier replication scheme has been proposed in [52] whereby mobile discon-nected applications are allowed to propose tentative update transactions. Onconnection, tentative transactions are applied to (re-processed at) the masterdata copy in the fixed network. At the re-processing stage, application seman-tics are used (such as finding commutative operations) to increase concurrency.To reduce re-processing costs that can be high in certain occasions, the workin [78] uses a history-based approach. On reconnection, tentative transactions,which are represented as histories, are merged with base transactions’ histo-ries. The merging process quickly identifies the set of tentative transactionsthat need to be backed out to resolve conflicts.

Page 151: Mobile Computing

130 CHAPTER 7

7.5.6 Semantic-based Nomadic Transaction Processing

The semantics-based mobile transaction processing scheme [49] views mobiletransactions as a concurrency and cache coherence problem. It introduces theconcepts of fragmentable and reorderable objects to maximize concurrency andcache efficiency exploiting semantics of object operations. The model assumesthe mobile transaction to be long-lived with unpredictable disconnections. Tra-ditional definitions of concurrency and serializability is too restrictive for mostoperations [41]. Commutativity of operations is an important property whichallows concurrent operations on an object. If certain operations on an objectis commutative, then the database server can schedule these operations in anarbitrary manner. Recovery also becomes quite simplified. Operations maybe commutative either for all states or part of the states of the objects. TheI/O values of the operations can be used to redefine serial dependencies of theoperations. Though this may improve concurrency, it may require complexrecovery mechanisms than normal schemes. Organization of the object can beused for selective caching of the object fragments, necessary for continuing theoperation during the disconnected state. This approach reduces the demandon the limited wireless bandwidth and provides better utilization of the cachespace available on the mobile host. Application semantics can also be utilizedto define the “degree of inconsistency,” “degree of isolation,” and the “degreeof transaction autonomy” [41, 32]. Techniques like epsilon serializability andquasi copies [23, 84] can be used to specify allowable inconsistencies in thesystem.

This approach utilizes the object organization to split large and complex ob-jects into smaller easily manageable pieces. The semantic information is utilizedto obtain better granularity in caching and concurrency. These fragments arecached and/or operated upon by the mobile hosts and later merged back toform a whole object. A stationary server sends out the fragments of an ob-ject when requested by mobile units. The objects are fragmented by a splitoperation. The split is done using a selection criteria and a set of consistencyconditions. The consistency conditions include the set of allowable operationson the object and the conditions of the possible object states. On completionof the transaction, the mobile hosts return the fragments to the server. Thesefragments are put together again by the merge operation at the server. If thefragments can be recombined in any order, then the objects are termed “re-orderable” objects. Aggregate items, sets, and data structures like stacks andqueues are examples of fragmentable objects.

Team-Fly®

Page 152: Mobile Computing

Research in Wireless and Mobile Computing 131

7.6 MOBILE COMPUTING MODELS

Computing in the mobile environment is different from the conventional fixed-network computing. This is partially due to the movement of the mobile hoststhat require to remain connected from different access points while moving. Thedifference also stems from the nature of the wireless links that are relativelyunreliable and offer low communication bandwidth. Furthermore, mobile hostsequipped with rechargeable batteries suffer from limited operation time con-straints. As a consequence, new models in the mobile environment are neededto support information access for mobile users.

The following models of computing in the mobile environment are currentlybeing researched and investigated:

Client/Server

Client/Proxy/Server

Disconnected Operation

Mobile Agents

The Thin Client Model

In the following, each model is described and its advantages and disadvantagesare discussed.

7.6.1 The Client/Server Model

In this model, neither the client nor the server are aware of the client (orserver) mobility. The conventional client/server model is used without anymodifications made in the application or the transport layer. The wirelessmedia is transparent since it is handled in the data link layer. The mobility,on the other hand, is made transparent by handling the variable client/serverlocation through location-based routing in the network layer. Mobile IP is anexample of a network protocol that hides the C/S mobility. The advantage ofthis model is its portability. The client or the server need not be changed inany way. Simply, the client (server) is ported to the mobile host, with a fixedaddress specified for the server (on the fixed network or on a mobile host). Thedisadvantages of this model are listed below:

Page 153: Mobile Computing

132 CHAPTER 7

The mobile client may suffer from a slow and unpredictable response time,especially when large server replies such as query results are transmittedwithout any considerations to the limited wireless bandwidth.

The server caching strategy may not work properly in this model becausethe majority of the fixed network caching algorithms use call backs to inval-idate the client cache. Most invalidation algorithms rely on the continuousavailability of the client. Since the client could be disconnected or tem-porarily inaccessible (during hand-offs for example), the cache invalidationprocess could fail.

7.6.2 The Client/Proxy/Server Model

To overcome the shortcomings of the conventional client/server model, theclient/proxy/server (C/P/S) model introduces a mobility-aware middle layerto mediate the interactions between the client and the server. The basic ideabehind this model was introduced in the Mowgli architecture [72], which is tosplit the communication path between the client and the server into two partsby using a store-and-forward interceptor.

The main advantage of this model is that the proxy allows the client and theserver to be designed without any built-in mobility assumptions. The proxy as-sumes that the client is mobile and the server is in the fixed network. The resultfrom the server is sent back to the proxy. The proxy filters the results accordingto the limitations of the wireless media and/or the client’s mobile unit. Theproxy may also store the filtered results until the client is connected. Examplesof data filtering include: color and resolution reduction, audio file removal, andfile size reduction. The programming of the proxy involves knowledge of themobile host hardware specifications. For example, a mobile host that does nothave audio capability will benefit from audio file removal filtering. In additionto the mobile host, the mobile user profile can be useful in providing the proxywith user preferences such as no-images and no-colors.

In support of the C/P/S computing model, a C/P/S network architecture forreliable communication is introduced in the Mowgli architecture [72]. In thisarchitecture, the mobile host is provided with a specialized transport service,the Mowgli Data Channel Service (MDCS). It provides prioritized data chan-nels with flow control between the mobile host and the base station. ExistingTCP/IP protocols are used between the base station and a fixed host so thatthe protocol software in the fixed network remains unmodified. Two media-

Page 154: Mobile Computing

Research in Wireless and Mobile Computing 133

tors, the Mowgli Agent and the Mowgli Proxy, which reside at each end of thewireless link, are used to provide functionality similar to that of TCP and UDP.

7.6.3 The Disconnected Operation Model

Mobile clients may face wide variations in network conditions and local resourceavailability when accessing remote data. This is true in the C/S and the C/P/Smodels. Disconnected operations is a variation of the C/S model where, insteadof working under the extreme case of weak-connectivity, the mobile client ef-fectively switches to use a network of zero bandwidth and infinite latency. Theoperations that enable a client to continue accessing critical data during the dis-connection (switch off) period are called disconnected operations. The abilityto operate when disconnected can be useful even when connectivity is available.For example, disconnected operation can extend battery life by avoiding wire-less transmission and reception. It allows radio silence to be maintained, a vitalcapability in military applications. And, it is a viable fallback position whennetwork characteristics degrade beyond usability. Voluntary disconnection canbe treated as planned failures which can be anticipated and prepared.

7.6.4 The Mobile Agent Model

Agents can be classified as static, mobile scripts, or mobile objects [42]. Staticagents are those which execute just on a single site, either as a client or asa server. A static agent could be carrying out some activity like mail filter-ing. Mobile scripts are those that are downloaded from a server and executedon a client. Java applets, perl, or python scripts can be classified as mobilescripts. Mobile agents are mobile scripts with an associated execution stateinformation. A mobile agent could either be relocated along with the user, orit could be relocated during the execution of the agent. The relocation of theagent involves saving the state before initiating relocation and later restartingthe mobile agent at the new location. The mobility of agents raises a largenumber of issues like security, authorization mechanisms, access mechanisms,and relocation mechanisms.

The mobile agent is an emerging new model that provides an alternative tothe C/P/S model. A mobile agent is an active entity that is knowledgeableof both the limitations of the mobile environment and the mobile user. Toaccess remote data, the mobile user sends a mobile agent on his behalf to thedata source in the fixed network. The mobile agent is an execution context

Page 155: Mobile Computing

134 CHAPTER 7

initially loaded with the queries or data access requests. Once the agent movesto the data source (server), it acts as a local client to the server. Once theC/S interactions between the mobile agent and the server are completed, theagent “targets” the resulting data in preparation for transmitting the result tothe mobile user. Such targeting includes filtering and transcoding actions suchas color depth and resolution reduction and compression. The mobile agentparadigm is depicted in Figure 7.8.

7.6.5 The Thin Client Model

The thin client computing model attempts to offload most application logicand functionality from mobile clients to stationary servers. In this model,applications in stationary servers are usually mobile-aware and optimized formobile client devices. This model is especially suitable for dumb terminal orsmall PDA applications.

Page 156: Mobile Computing

Research in Wireless and Mobile Computing 135

The thin client architecture from CITRIX Corporation allows a variety of re-mote computers, regardless of their platform, to connect to a Windows NTterminal server to remotely access a powerful desktop and its applications [4].A server called MetaFrame runs under Windows NT in the desktop machine andcommunicates with the thin clients executing at the remote computers usingthe Independent Computing Architecture protocol (ICA). The ICA client andthe MetaFrame server collaborate to display the virtual desktop on the remotecomputer screen. They also collaborate to process mouse and keyboard eventsand to execute programs and view data stored at the server. All executionsare remote and none take place at the client portable computer. The researchwork described in [40] examines extensions to CITRIX thin client architectureso that it is optimized in the wireless environment. The work pointed out thatbandwidth limitation is not as detrimental to the thin client performance asnetwork latency. This is because the thin clients’ use of bandwidth is limited.

Page 157: Mobile Computing

This page intentionally left blank

Page 158: Mobile Computing

APPENDIX–A: GLOSSARY OFCOMMON ABBREVIATIONS

AAAAAAA The American Association for the Abolishment of AbusedAbbreviations and Asinine Acronyms :-)

ACTS Advanced Communications Technologies and Services. A re-search organization 50% funded by the European EconomicCommission Air Link Interface

ADSL Asymmetric Digital Subscriber Line

AMPS Advanced Mobile Phone System

ARDIS Advanced Radio Data Information Service

ATIS See Appendix–C

ATM Asynchronous Transfer Mode

BER Bit Error Rate

BOOTP Bootstrap Protocol. Used to establish communications be-tween computers and devices at start up time

BS Base Station. Also known as Mobility Support Station or(MSS). Also known as Base Station Controller (in GSM)

BSC Base Station Controller (in GSM)

BTS Base Transceiver Station (in GSM)

CDMA Code Division Multiple Access, An air link interface codingscheme wherein multiple subscribers are granted access to thesame radio frequency source by assigning subscribers transmitand receive signals a spectrum spreading code. CDMA is alsoreferred to as “Cellular, IS-95”, a digital spread-spectrum sys-tem initially developed by QUALCOMM Inc. and standard-ized by the Telecommunications Industry Association (TIA).

Page 159: Mobile Computing

138

CDPD Cellular Digital Packet Data is a connectionless network pro-posed by the CDPD Forum, a trade association of carriersequipment suppliers and application developers. It is basedon early IETF Mobile-IP work. CDPD transmits packet dataover free (unused) channels in existing AMPS cellular radio

CDPD Forum See Appendix–C

CLNP Connectionless Network Protocol. An OSI protocol used forthe delivery of data. It uses datagrams that include address-ing information for routing network messages. Similar to theInternet UDP datagram protocol.

CMRS Commercial Mobile Radio Services spectrum as defined bythe FCC, and its international equivalents

CTIA See Appendix–C

DARPA Defense Advanced Research Projects Agency

DCE Data Communication Equipment (in Motorola iDEN). A de-vice or attachment to a DTE (See below) that is responsiblefor communication.

DCT Digital Cordless Telephone

DECT Digital European Cordless Telecommunications (Standard)

DECT Forum See Appendix–C

DHCP Dynamic Host Configuration Protocol. An IETF protocol

DTE Data Terminal Equipment (in Motorola iDEN). Also knownas Mobile Equipment (in GSM), or Mobile Unit

E-TDMA Extended Time Division Multiple Access. A proposed systemto extend TDMA. It makes use of 30 KHz wide channels,but allows up to six time multiplexed users per channel withvocoder rate of 4 kbits/s

ETSI See Appendix–C

FCC See Appendix–C

Page 160: Mobile Computing

Research in Wireless and Mobile Computing 139

GPRS General Packet Radio Service. A GSM packet data servicedeveloped by ETSI as part of GSM phase 2+ developments.It supports mobile connections to multiple networks includingTCP/IP, X.25, and CLNP. GPRS mobility management usesmobility agents (home and foreign) similar to Mobile-IP

GPS Global Positioning System. A US government satellite systemthat provides location and time information to users

GSM Global System for Mobile communications (Actually: “GroupeSpeciale Mobile”). Similar to TDMA except that it uses200 KHz wide channels with eight users per channel and hasa vocoder rate of 13 kbits/s. It is the first digital cellular sys-tem to be used commercially and has been adopted in Europeand many Pacific rim countries

HLR Home Location Register

HTTP HyperText Transfer Protocol

ICMP Internet Control Message Protocol. Used to ping Internetcomputers to test their reachability across the networks

iDEN Motorola’s Integrated Digital Enhanced Network. A packetbased network for voice and data

IEC See Appendix–C

IEEE 802.11 See W-LAN

IETF See Appendix–C

IF Intermediate Frequency. Ranges from 45 to 210 MHz

IMT 2000 International Mobile Telecommunications by the year 2000project. Under the subgroup IMT (International Mobile Telecommunications) in the ITU that plans to facilitate cooperationin deciding global wireless access for the 21st century

IR Infra-Red communication. Related to IR is IrDA which isprotocol adopted for infrared communication between porta-bles without wires over distances as long as one meter. TheIrDA 1.1 spec, for example, operates at 4 Mbps speed

IrDA See Appendix–C

ISDN Integrated Services Digital Network

Page 161: Mobile Computing

140

ITU See Appendix–C

JDC Japan Digital Cellular

JTACS Japanese Total Access Communication System. A narrowband analog FM follow-on to the TACS system used in Japan

LEO Low Earth Orbit. Any orbit around the earth below the geo-stationary satellite orbit, generally within several hundredKilometers above the earth’s surface and usually inclined tothe equatorial plane.

Macro Cell A unit of communication coverage in cellular networks, usu-ally in the order of 5-15 Kilometer in diameter

MBS Mobile Broadband Systems. Experimental high bandwidthwireless communication with provision for mobility. ACTS iscurrently active in MBS research

ME Mobile Equipment (in GSM). Also known as Mobile Unit(MU)

MEMS Micro-Electro Mechanical System

Micro Cell A unit of communication coverage in cellular networks, usu-ally in the order of 500-1500 meters

MSS Mobile Satellite Service. A service that links mobile earthstations with base stations and with one another via one ormore satellite. The same abbreviation is used to denote Mo-bile Support Stations which is also known as Base Stations(See BS)

NAMPS Narrowband Advanced Mobile Phone Service. An analog cel-lular telephone system currently used in North America

NMT Nordic Mobile Telephone

OSI Open Systems Interconnection. An ISO (see Appendix–C)standard of a Reference Model for how messages should betransmitted between any two points in a telecommunicationnetwork

PCMCIA Personal Computer Memory Card International Association.Today, PCMCIA refers to a credit-card sized, removable mod-ule that has become the expansion vehicle for portable com-puters. This includes memory, I/O and hard disks. Alsoknown as PC card

Team-Fly®

Page 162: Mobile Computing

Research in Wireless and Mobile Computing 141

PCN Personal Communications Network. European equivalent ofNorth America’s PCS (See PCS). Also known as second gen-eration cellular system, in North America

PCS Personal Communication Systems

PDA Personal Digital Assistant. A class of PIAs.

PDC Personal Digital Cellular (Japan)

PHS Personal Handy Phone System (Japan)

PIA Personal Information Appliance

Pico Cell A unit of communication coverage in cellular networks, usu-ally on the order of 5-100 meters

PPP Point to Point Protocol. A serial line version of the TCP/IPprotocol.

PSTN Public Switched Telephone Network.

QoS Quality of Service. A measure of guarantees that can be madein meeting certain performance requirements

RBOC Regional Bell Operating Company. One of the seven “babyBells” caused by the 1984 divestiture of AT&T

RF Radio Frequency. Frequency of the transmitted/received sig-nal. For some cellular telephones in the US this is in the rangefrom 800 to 900 MHz

SDSL Symmetric Digital Subscriber Line

SIM Subscriber Identity Module. A smart card technology thatstores the subscriber identity information. By using SIMs,subscribers can change cell phones and other devices and yetobtain the same services.

SMS Short Messaging Service. An extension of paging services.

SQL Structured Query Language. A standard language for query-ing and manipulating relational databases

T1P1 See Appendix–C

Page 163: Mobile Computing

142

TACS Total Access Communication System. An Analog FM com-munication system used in the United Kingdom and Japan.It uses 25 KHz wide channels and signaling is superaudio

TD-CDMA Time Division, Code Division Multiple Access

TDMA Time Division Multiple Access. Method wherein multiplesubscribers are granted access to the same radio frequencysource by limiting subscribers transmit and receive signals totime slots. TDMA is also referred to as the first digital cellu-lar FM system standardized in North America. It makes useof 30 kHz wide channels, but three users are time-multiplexedon each channel. The vocoder rate is 8 Kbits/s

TeleVAS Telephony Value Added Services. Such as wait calling, callforwarding, and conference calling

TIA See Appendix–C

UMTS Universal Mobile Telecommunications System is a project un-der the SMG (Special Mobile Groupe), a committee in ETSI.UMTS is the first implementation, by the Europeans, of theIMT-2000 standard

USTA See Appendix–C

UTRA UMTS Terrestial Radio Access

VDSL Very high rate Digital Subscriber Line

VLR Visitor Location Register

W-ATM Wireless Asynchronous Transfer Mode network

W-CDMA Wideband Code Division Multiple Access

W-LAN Wireless Local Area Network. A wireless extension to wirelineintranets. W-LANs do not require licenses and must meet reg-ulatory requirements such as maximum transmission power.The IEEE 802.11 is a recent (1998) standard that guaranteesinteroperability of W-LAN products from different vendors

W-TDMA Wideband Time Division Multiple Access

WAP Wireless Application Protocol. A ISO/OSI protocol stack foruse by digital phones and other wireless devices to access theInternet. Developed by the WAP Forum

Page 164: Mobile Computing

Research in Wireless and Mobile Computing 143

WAP Forum See Appendix–C

WDF Forum See Appendix–C

WSP Wireless Service Provider. An emerging Internet service thatcaters to subscribers with wireless interfaces (e.g. GSM, CDPD,iDEN). WSP will provide Internet access plus added value ser-vices such as Web content filtering and transcoding (e.g. WAPbased Wireless Markup Language), and information feed ser-vices.

Page 165: Mobile Computing

This page intentionally left blank

Page 166: Mobile Computing

APPENDIX–B: WIRELESSCELLULAR SYSTEMS

Analog Cellular Technology

NMT-Nordic Mobile TelephoneMobile Frequency Range (MHz):NMT-450: Rx: 463-468, Tx: 453-458NMT-900: Rx: 935-960, Tx: 890-915Multiple Access Method: FDMADuplex Method: FDDNumber of Channels:NMT-450: 200, NMT-900: 1999Channel Spacing:NMT-450: 25kHz, NMT-900: 12.5 kHzModulation: FMChannel Bit Rate: n/a

AMPS-Advanced Mobile Phone ServiceMobile Frequency Range (MHz):Rx: 869-894, Tx: 824-849Multiple Access Method: FDMADuplex Method: FDDNumber of Channels: 832Channel Spacing: 30 kHzModulation: FMChannel Bit Rate: n/a

TACS-Total Access Communication SystemMobile Frequency Range (MHz):ETACS: Rx: 916-949, Tx: 871-904NTACS: Rx: 860-870, Tx: 915-925Multiple Access Method: FDMADuplex Method: FDDNumber of Channels:ETACS: 1240, NTACS: 400

Page 167: Mobile Computing

146

Channel Spacing:ETACS: 25 kHz, NTACS: 12.5kHzModulation: FMChannel Bit Rate: n/a

Analog Cordless Technology

CTO-Cordless Telephone OMobile Frequency Range (MHz):2/48 (U.K.), 26/41 (France), 30/39 (Australia), 31/40 (Netherlands, Spain),46/49 (China, S. Korea, Taiwan, USA), 48/74, 45/48 (China)Multiple Access Method: FDMADuplex Method: FDDNumber of Channels: 10, 12, 15, 20 or 25Channel Spacing: 1.7,20,25 or 40 kHzModulation: FMChannel Bit Rate: n/a

JCT-Japanese Cordless TelephoneMobile Frequency Range (MHz):254/380Multiple Access Method: FDMADuplex Method: FDDNumber of Channels: 89Channel Spacing: 12.5 kHzModulation: FMChannel Bit Rate: n/a

CT1/CT1+ Cordless Telephone 1Mobile Frequency Range (MHz):CTI: 914/960, CTI+: 80Multiple Access Method: FDMADuplex Method: FDDNumber of Channels: CTI: 40, CTI+: 80Channel Spacing: 25 kHzModulation: FMChannel Bit Rate: n/a

Page 168: Mobile Computing

Research in Wireless and Mobile Computing 147

Digital Cellular Technology

DCS 1800Mobile Frequency Range (MHz):Rx: 1805-1880, Tx: 1710-1785Multiple Access Method: TDMA/FDMDuplex Method: FDDNumber of Channels: 374 (8 users/channel)Channel Spacing: 200 kHzModulation: GMSK (0.3 Gaussian Filter)Channel Bit Rate: 270.833 kb/s

PDC-Personal Digital CellularMobile Frequency Range (MHz):Rx: 810-826, Tx: 940-956Rx: 1429-1453, Tx: 1477-1501Multiple Access Method: TDMA/FDMDuplex Method: FDDNumber of Channels: 1600 (3 users/channel)Channel Spacing: 25 kHzModulation: 1/4 DQPSKChannel Bit Rate: 42 kb/s

IS-54/-136North American Digital CellularMobile Frequency Range (MHz):Rx: 869-894, Tx: 824-849Multiple Access Method: TDMA/FDMDuplex Method: FDDNumber of Channels: 832 (3 users/channel)Channel Spacing: 30 kHzModulation: 1/4 DQPSKChannel Bit Rate: 48.6 kb/s

IS-95: North American Digital CellularMobile Frequency Range (MHz):Rx: 869-894, Tx: 824-849Multiple Access Method: CDMA/FDMDuplex Method: FDDNumber of Channels: 20 (798 users/channel)Channel Spacing: 1250 kHzModulation: QPSK/OQPSKChannel Bit Rate: 1.2288 kb/s

Page 169: Mobile Computing

148

GSM: Global System for Mobile CommunicationsMobile Frequency Range (MHz):Rx: 925-960, Tx: 880-915Multiple Access Method: TDMA/FDMDuplex Method: FDDNumber of Channels: 124 (8 users/channel)Channel Spacing: 200 kHzModulation: GMSK (0.3 Gaussian Filter)Channel Bit Rate: 270.833 kb/s

Law-tier PCS: Personal Communications SystemsMobile Frequency Range (MHz):Rx: 1930-1990, Tx: 1850-1910Multiple Access Method:

– PACS (based on PHS cordless)

– DCT-U (based on DECT cordless)

– Composite CDMA/TDMA

High-tier PCS: Personal Communications SystemsMobile Frequency Range (MHz):Rx: 1930-1990, Tx: 1850-1910Multiple Access Method:

– PCS TDMA (based on IS-136 cellular)

– PCS CDMA (based on IS-95 cellular)– PCS 1900 (based on GSN cellular)

– Wideband CDMA

Digital Cordless Technology

CT2/CT2+ Cordless Telephone 2Mobile Frequency Range (MHz):CT2: 864/868, CT2+: 944/948Multiple Access Method: TDMA/FDMDuplex Method: TDDNumber of Channels: 40Channel Spacing: 100 kHzModulation: GFSK (0.5 Gaussian Filter)Channel Bit Rate: 72 kb/s

Page 170: Mobile Computing

Research in Wireless and Mobile Computing 149

PHS-Personal Handy Phone SystemMobile Frequency Range (MHz): 1895-1918Multiple Access Method: TDMA/FDMDuplex Method: TDDNumber of Channels: 300 (4 users/channel)Channel Spacing: 300 kHzModulation: 1/4 DQPSKChannel Bit Rate: 384 kb/s

DECT-Digital European Cordless TelephoneMobile Frequency Range (MHz): 1880-1900Multiple Access Method: TDMA/FDMDuplex Method: TDDNumber of Channels: 10 (12 users/channel)Channel Spacing: 1.728 MHzModulation: GFSK (0.5 Gaussian Filter)Channel Bit Rate: 1.152 Mb/s

Wireless Data Technology

CDPD: Cellular Digital Packet DataMobile Frequency Range (MHz):Rx: 869-894, Tx: 824-849Multiple Access Method: FDMADuplex Method: FDDNumber of Channels: 832 (4 users/channel)Channel Spacing: 30 kHzModulation: GMSK (0.5 Gaussian Filter)Channel Bit Rate: 19.2 kb/s

RAM-MobitexMobile Frequency Range (MHz):(North America): Rx: 935-941, Tx: 896-902(Europe/Asia): 403-470Multiple Access Method: TDMA/FDMDuplex Method: FDDNumber of Channels: 480Channel Spacing: 12.5 kHzModulation: GMSK (0.3 Gaussian Filter)Channel Bit Rate: 8 kb/s

Page 171: Mobile Computing

This page intentionally left blank

Team-Fly®

Page 172: Mobile Computing

APPENDIX–C: STANDARDSORGANIZATIONS

ANSI – http://www.ansi.org/

American National Standards Institute is the primary organization for fosteringthe development of technology standards in the United States. ANSI workswith industry groups and is the U.S. member of the International Organizationfor Standardization (ISO) and the International Electrotechnical Commission(IEC).

ATIS – http://www.atis.com

Alliance for Telecommunications Industry Solutions (formerly Exchange Carri-ers Standards Association). Houses Tl standards committees on wireless andswitching interfaces to telephone exchange systems. Many projects are donejointly with TIA.

BlueTooth – http://www.bluetooth.com

Bluetooth is a technology specification for small form factor, low-cost, shortrange radio links between mobile PCs, mobile phones and other portable de-vices. The Bluetooth Special Interest Group is an industry group consisting ofleaders in the telecommunications and computing industries that are drivingdevelopment of the technology and bringing it to the market.

CDPD Forum – http://www.cdpd.org

Cellular Digital Packet Data Forum is a trade association of carriers, equipmentsuppliers, and mobile application developers. As of May 1998, the CDPDForum will be expanded into the Wireless Data Forum (See WDF).

Page 173: Mobile Computing

152

CTIA – http://www.wow-com.com

Cellular Telecommunications Industry Association. Represents the Cellularand PCS system operators/license holders in the USA.

DECT Forum – http://www.dect.ch

Digital Enhanced Cordless Telephone system, also known as Digital EuropeanCT and DCT or PWT in the North American versions. The North Ameri-can versions use DQPSK modulation, while the European version uses digitalGMSK FM.

ETSI – http://www.etsi.fr

European Telecommunication Standards Institute. Administers GSM, DECT,and and other wireless/cellular standards. Publishes the standards which aresuccessors to the former CCITT standards.

FCC – http://www.fcc.gov

United States Federal Communications Commission. Writes and administersrules governing wire and radio/television communication in the USA.

Global Engineering Documents – http://global.ihs.com

Formerly Indian Head Systems. Franchised distributor of TIA standards doc-uments and other selected standards documents.

IEEE – http://www.ieee.org

Institute of Electrical and Electronics Engineers. Publishes the 802.x LANstandards.

Page 174: Mobile Computing

Research in Wireless and Mobile Computing 153

IEC – http://www.iec.ch

International Electrotechnical Commission. Publishes standards mainly de-voted to electrical and optical fiber connectors and power systems and graphicstandards for diagrams and labeling.

IETF – http://www.ietf.org

Internet Engineering Task Force. An open international community of networkdesigners, operators, vendors, and researchers concerned with the evolution andimprovement of the Internet.

IrDA – http://www.irda.org

Infrared Data Association. IrDA is an association of over 160 companies worldwide focused on providing IR standards to ensure the quality and interoper-ability of the Infrared Technology.

ISO – http://www.iso.ch

International Organization of Standards. ISO is a worldwide federation ofnational standards bodies from some 100 countries, one from each country.Among the standards it fosters is Open Systems Interconnection (OSI), a ref-erence model for communication protocol.

ITU – http://www.itu.ch

International Telecommunication Union. An international organization withinwhich governments and the private sector coordinate global telecommunicationnetworks and services.

The Salutation Consortium – http://www.salutation.org

Salutation is a non-profit organization defining an architecture for “find andbind” mobile environment. The architecture is intended to be used to discover

Page 175: Mobile Computing

154

services and capabilities among a diverse set of mobile information appliances,under different connectivity and mobility conditions. The architecture will leadto creating open servers for mobile client.

T1P1 – http://www.tl.org

Planning committee of ATIS. Issues many wireless standards jointly with TIA.

TelCordia (ex. Bellcore) – http://www.bellcore.com

Bell Communications Research. Split off from Bell Laboratories in 1984. Re-cently sold to Science Applications International Corp (SAIC). Issues TR (tech-nical reference) and TA (technical advisory) standards for the North Americanpublic telephone industry.

TIA – http://www.itsa.org

Telecommunications Industry Association. Issues standards on consumer endtelecom equipment (telephone sets, PBX equipment, etc.) and cellular/PCSsystems.

USTA – http://www.usta.org

U.S. Telephone Association (formerly US Independent Telephone Association).Affiliated with TIA. Represents the telephone industry.

WAP Forum – http://www.wapforum.org

The Wireless Application Protocol Forum. A non-for-profit organization withmembers from terminal (phones and PDA) and infrastructure manufacturers,operators, carriers, wireless software developers, and internet content providers.The goal of the WAP Forum is to develop a wireless protocol specificationthat works across differing wireless network technology types, for adoption byappropriate industry standards bodies.

Page 176: Mobile Computing

Research in Wireless and Mobile Computing 155

WDF Forum – http://www.cdpd.org

Wireless Data Forum (previously the CDPD Forum) is a nonprofit organiza-tion formed to promote the benefits of wireless data products and services toend-user communities, the telecommunications industry, the media, and theinformation technology industry.

Page 177: Mobile Computing

This page intentionally left blank

Page 178: Mobile Computing

REFERENCES

[1] Acts 1997 programme guide. http://www.de.infowin.org/ACTS/.

[2] Bbn’s bodylan. http://www.bbn.com/getsmart/sdp/tpo.htm.

[3] Bluetooth. http://www.bluetooth.com.

[4] The citrix thin client architecture. http://www.citrix.com.

[5] Cmu wearable computer project.http://www.cs.cmu.edu/afs/cs/project/vuman/www/home.html.

[6] Comparison of mobile ip and cdpd.http://www.ctr.columbia.edu/ chien/Courses/E6950/comparison.html.

[7] The IETF working group on IP routing for wireless/mobile hosts (Mo-bilelP).http://www.ietf.org/html.charters/mobileip-charter.html.

[8] IP in IP tunneling (RFC 1853).http://www.ietf.org/html.charters/mobileip-charter.html.

[9] Mit wearable computing project.http://lcs.www.media.mit.edu/projects/wearables/.

[10] Mobile network computer white paper.http://www.pc.ibm.com/networkstation/os/whitepaper.html.

[11] Mobility support in IPv6: Internet draft.http://www.ietf.org/html.charters/mobileip-charter.html.

[12] Network Computer Reference Profile (MNCRF) Standard.http://www.opengroup.org/public/pubs/catalog/c720.htm.

[13] Phs wristwatch phone. http://www.hil.ntt.co.jp/Speech/PHS/index-E.html.

[14] Route optimization in Mobile IP: Internet draft.http://www.ietf.org/html.charters/mobileip-charter.html.

Page 179: Mobile Computing

158 ANY TIME, ANYWHERE COMPUTING

[15] Seiko’s wristwatch pc. http://ruputer.com/englilsh.

[16] Toshiba’s desk area network (dan).http://www.toshiba.co.jp/tech/review/1998/high98/research/r2/index.htm.

[17] Intermediate system to intermediate system intra-domain routing ex-change protocol for use in conjunction with the protocol for providingthe connectionless-mode network service. February 90. (ISO 8473), ISODP 10589.

[18] Personal handy phone system. ARIB (Association of Radio Industries andBusinesses) Standard V2 RCR STD-28, December 95.

[19] Ip mobility support. October 96. IETF RFC 2002.

[20] Mission statement and IMT 2000 specifications. December 97.http://www.itu.ch/.

[21] Mission statement and UMTS project specifications. December 97.http://www.etsi.fr/.

[22] S. Acharya, R. Alonso, M. Franklin, and S. Zdonik. Broadcast disks: Datamanagement for asymmetric communications environments. In Proceed-ings of the ACM SIGMOD Conference on Management of Data, San Jose,CA, 1995.

[23] R. Alonso, D. Barbara, and L. L. Cova. Using stashing to increase nodeautonomy in distributed file systems. In Proceedings of the 9th Symposiumon Reliable Distributed Systems. IEEE, October 1990.

[24] M. Baker, X. Zhao, S. Cheshire, and J. Stone. Supporting mobility inmosquitonet. In Proceedings of the 1996 USENIX Conference, San Diego,CA, January 1996.

[25] A. Bakre and B. Badrinath. Indirect Transport Layer protocols For MobileWireless Environment. In T. Imielinski and H. Korth, editors, MobileComputing, chapter 8, pages 229–252. Kluwer Academic Publishers, 1996.

[26] H. Balakrishnan, S. Seshan, E. Amir, and R. Katz. Improving TCP/IPperformance over wireless networks. In First ACM International Confer-ence on Mobile Computing and Networking (MobiCom), November 1995.

[27] J. Bartlett. W4–the wireless world wide web. In Workshop on MobileComputing and Applications, December 1994.

Page 180: Mobile Computing

References 159

[28] S. Basagni, I. Chlamtac, V. R. Syrotiuk, and B. A. Woodward. A distancerouting effect algorithm for mobility (dream). In Fourth Annual Inter-national Conference on Mobile Computing and Networking (MobiCom),October 1998.

[29] Bellcore. Intelligent network (in) tutorial. 1999.http://www.webproforum.com/bell-atlantic/index.html.

[30] K. Buchanan, R. Fudge, and D. McFarlane. IMT-2000: Service provider’sperspective. IEEE Personal Communications, August 97.

[31] R. Caceres and L. Iftode. Improving the Performance of Reliable Trans-port Protocols in Mobile Computing Environments. In T. Imielinski andH. Korth, editors, Mobile Computing, chapter 7, pages 207–228. KluwerAcademic Publishers, 1996.

[32] P. Chrysanthis. ACTA, A Framework for Modeling and Reasoning aboutExtended Transactions. PhD thesis, University of Massachusetts, Amherst,1991.

[33] P. Chrysanthis. Transaction processing in mobile computing environment.In IEEE workshop on Advances in Parallel and distributed Systems, pages77–82, Oct 1993.

[34] Fumio Teraoka Kim Claffy and Mario Tokoro. Design, implementationand evaluation od virtual internet protocol. volume Proceedings of 12thInternational Conference on Distributed Computing SYstems, pages 170–177. IEEE Computer Soceity, June 1992.

[35] Nielsen J. D. Transactions in Mobile Computing. PhD thesis, DIKU, 1995.

[36] N. Davies, G. Blair, K. Cheverst, and A. Friday. Supporting adaptiveservices in a heterogeneous mobile environment. In Workshop on MobileComputing and Applications, Santa Cruz, California, December 1994.

[37] A. Demers, K. Petersen, M. Spreitzer, D. Terry, M. Theimer, and B. Welch.The BAYOU architecture: Support for data sharing among mobile users.In Workshop on Mobile Computing and Applications, December 1994.

[38] R. Drooms. Dynamic Host Configuration Protocol, rfc 1541 edition.

[39] M. Dunham, A. Helal, and S. Balakrishnan. A mobile transaction modelthat captures both the data and movement behaviour. ACM-BaltzerJounal on Mobile Networks and Applications (MONET), 2(2):149–162,October 1997.

Page 181: Mobile Computing

160 ANY TIME, ANYWHERE COMPUTING

[40] J. Duran and A. Laubach. Virtual personal computers and the portablenetwork. In Proceedings of the IEEE International Performance, Commu-nication, and Computing Conferenc (IPCCC’99), Phoenix, AZ, February1999.

[41] A. Elmagarmid, editor. Database Transaction Models For Advanced Ap-plications. Morgan Kaufman, 1991.

[42] D. Gilbert et. al. Intelligent agent strategy. Technical report, IBM Corpo-ration, Research Triangle Park, NC, 1995.

[43] P. Evaggelia and B. Bharat. Revising transaction concepts for mobilecomputing. In First IEEE Workshop on Mobile Computing Systems andApplications, pages 164–168, Dec 1994.

[44] P. Evaggelia and B. Bharat. Maintaining consistency of data in mobiledistributed environments. In 15th International Conference on DistributedComputing Systems., 1995.

[45] G. Forman and J. Zahorjan. The challenges of mobile computing. IEEEComputer, 27(4):38–47, April 1994.

[46] Thomas Frey. Performance of adaptive antenna arrays for cellular cdmasystems. In The Second European Personal Mobile Communications Con-ference, Bonn, Germany, September 97.

[47] A. Friday, G. Blair, K. Cheverst, and N. Davies. Extensions to ANSAwarefor advanced mobile applications. Technical report, Department of Com-puting, Lancaster University, Lancaster, U.K., 1996.

[48] V. Garg and J. Wilkes, editors. Wireless and Personal CommunicationSystems. Prentice Hall PTR, 1996.

[49] W. Gary and C. Panos. Supporting semantics-based transaction process-ing in mobile database applications. In Proceedings of the 14th IEEESymposium on Reliable Distributed Systems, Sept 1995.

[50] J. Gibson. The Mobile Communications Handbook. CRC Press, IEEEPress, 96.

[51] J. Gilmore. Boot Strap Protocol, rfc 1541 edition.

[52] J. Gray, P. Helland, P. O’Neil, and D. Shasha. The dangers of replica-tion and a solution. In Proceedings of the ACM SIGMOD internationalconference on Management of data, pages 173–182, June 1996.

Team-Fly®

Page 182: Mobile Computing

References 161

[53] L. Harte and S. Prokup. Cellular and PCS/PCN Telephones and Systems.APDG Publishing, 1996.

[54] Lawrence Harte and Steve Prokup. Cellular and PCS/PCN Telephonesand Systems. APDG Publishing, 1996. ISBN 0-9650658-1-2.

[55] C. Hedrick. The routing information protocol. June 88. Internet RFC1058.

[56] W. Honcharenko, J. Kruys, D. Lee, and N. Shah. Broadband wirelessaccess. IEEE Communications Magazine, January 97.

[57] B. Housel and D. Lindquist. Webexpress: A system for optimizing webbrowsing in a wireless environment. In Second Annual International Con-ference on Mobile Computing and Networking (MobiCom’96), November1996.

[58] T. Imielinski and H. Korth, editors. Mobile Computing. Kluwer AcademicPublishers, 1996.

[59] J. Ioannidis, D. Duchamp, and Jr G. Maguire. Ip-based protocols formobile internetworking. SIGCOMM, pages 235–245, 1991.

[60] D. Johnson. Scalable support for transparent mobile host internetworking.Wireless Networks, 1:311–321, October 1995.

[61] D. Johnson and D. Maltz. Dynamic source routing in ad-hoc wireless net-works. In T. Imielinski and H. Korth, editors, Mobile Computing, chap-ter 5, pages 153–179. Kluwer Academic Publishers, 1996.

[62] D. Johnson and D. Maltz. Protocols for adaptive wireless and mobilenetworking. IEEE Personal Communications, 3(1), February 1996.

[63] David B. Johnson. Mobile host internetworking using ip loose source rout-ing. Technical report, School Of Computer Science, Carnegie Mellon Uni-versity, Feb 1993.

[64] A. Joseph, J. Tauber, and M. Kaashoek. Mobile computing with therover toolkit. IEEE Transactions on Computers: Special issue on MobileComputing, 43(3), 1997.

[65] A. Joshi, R. Weerasinghe, S. Mc.Dermott, B. Tan, G. Bernhardt, andS. Weerawarana. Mowser: Mobile platforms and web browsers. Bulletinof the Technical Committee on Operating Systems and Applications, 8(1),1996.

Page 183: Mobile Computing

162 ANY TIME, ANYWHERE COMPUTING

[66] M. Kaashoek, T. Pinckney, and J. Tauber. Dynamic documents: Mo-bile wireless access to the www. In Workshop on Mobile Computing andApplications, December 1994.

[67] K. Kammerlander. Benefits of combined TDMA/CDMA operation forthird generation mobile radio systems. In IEEE 4th International Sympo-sium on Spread Spectrum Techniques and Applications, Mainz, Germany,August 96.

[68] R. Katz and E. Brewer. The bay area research wireless access network(BARWAN). In Workshop on Mobile Computing and Applications, De-cember 1994.

[69] R. Katz and E. Brewer. The case for wireless overlay networks. InT. Imielinski and H. Korth, editors, Mobile Computing, chapter 23, pages621–648. Kluwer Academic Publishers, 1996.

[70] J. Kistler and M. Satyanarayanan. Disconnected operation in the coda filesystem. ACM Transactions on Computer Systems, 1992.

[71] Y. B. Ko and N. H. Vaidya. Location-aided routing (lar) in mobile ad hocnetworks. In Fourth Annual International Conference on Mobile Comput-ing and Networking (MobiCom), October 1998.

[72] M. Kojo, K. Raatikainen, and T. Alanko. Connecting mobile workstationsto the internet over a digital cellular telephone network. In MobidataWorkshop, November 1994.

[73] G. Kuenning and G. Popek. Automated hoarding for mobile computers. InProceedings of the 16th ACM Symposium on Operating Systems Principles(SOSP), 1997.

[74] G. Livingston. Third generation wireless standards to shape internet’sfuture. Wireless NOW, November 97.http://www.commnow.com/3rd_Generation.html.

[75] Q. Lu and M. Satyanaranyanan. Isolation-only transactions for mobilecomputing. ACM Operating Systems Review, 28(2), 1994.

[76] A. Myles, D. Johnson, and C. Perkins. Mobile host protocol supportingroute optimization and authentication. IEEE Journal of Selected Areas inCommunications, 13(5):839–849, June 1995.

[77] B. Noble, M. Price, and M. Satyanarayanan. A programming interface forapplication-aware adaptation in mobile computing. In Proceedings of theSecond USENIX Symposium on Mobile and Location-Independent Com-puting, 1995.

Page 184: Mobile Computing

References 163

[78] S. Jajodia P. Liu, P. Ammann. Incorporating transaction semantics to re-duce reprocessing overhead in replicated mobile data applications. In Pro-ceedings of the 19th IEEE International Conference on Distributed Com-puting Systems, Austin, Texas, May 1999.

[79] T. Parker. Put on your 3G glasses. PCS Edge, August 97.http://www.internettelephony.com.

[80] C. Perkins. Mobile-IP Design principles and Practices. Addison-WesleyWireless Communications Series, 1998.

[81] C. Perkins and D. Johnson. Mobility support in ipv6. In Second Annual In-ternational Conference on Mobile Computing and Networking (MobiCom),November 1996.

[82] C. Perkins and E. Royer. Ad-hoc on-demand distance vector routing. InSecond IEEE Workshop on Mobile Computing Systems and Applications,February 1999.

[83] R. Prasad, J. Nijhof, and H. Cakil. Hybrid tdma/cdma multiple accessprotocol for multi-media communications. In IEEE International Confer-ence on Personal Wireless Communication, February 96.

[84] K. Ramamritham and C. Pu. A formal characterization of epsilon serial-izability. IEEE Transactions on Knowledge and Data Engineering, 1994.

[85] M. Satyanarayanan. Fundamental challenges of mobile computing. InACM Symposium on Principles of Distributed Computing, 1995.

[86] M. Satyanarayanan. Mobile information access. IEEE Personal Commu-nications, 3(1), 1996.

[87] K. Seal and S. Singh. Loss profiles: A quality of service measure in mobilecomputing. Wireless Networks, 2:45–61, February 1996.

[88] James Solomon. Mobile IP: The Internet Unplugged. Prentice Hall Pro-fessional Technical Reference, 1998. ISBN 0-13-856246-6.

[89] C. Sunshine and J. Postel. Addressing Mobile Hosts in ARPA Environ-ment. Internet Engg. Note IEN 135, Mar 1980.

[90] G. Voelker and B. Bershad. Mobisaic: An information system for a mobilewireless computing environment. In Workshop on Mobile Computing andApplications, December 1994.

[91] Mark Weiser. The computer for the twenty-first century. Scientific Amer-ican, 265(3):94–104, September 1991.

Page 185: Mobile Computing

164 ANY TIME, ANYWHERE COMPUTING

[92] R. Yavatkar and N. Bhagawat. Improving end-to-end performance of tcpover mobile internetworks. In Workshop on Mobile Computing and Appli-cationsr, December 1994.

Page 186: Mobile Computing

INDEX

ACTS, 72ARDIS, 21Access Point, 20Ad-Hoc Mode, 20Ad-Hoc Routing, 105BARWAN, 114BOOTP Protocol, 101BIueTooth, 19, 55BodyLAN, 53Broadband services, 64Broadcast Disks, 123CDMA, 16CDPD, 21, 106Cellular Networks, 20Cellular Systems, 13Client Proxy, 90Clio, 38Communicator, 38Continuous Queries, 4D-AMPS, 15DAN, 55DHCP Protocol, 101Degree of Mobility, 22Diffie-Hellamn Protocol, 109DirecPC, 21Direct Sequence, 16Dynamic URL, 120EPOCH, 11Ear Phone, 59Enhanced TDMA (ETDMA), 15Ericsson, 26Fast Retransmission TCP, 112Foreign Agent, 102Fourth generation wireless networks, 86Frequency Hopping, 16Frequency Reuse, 16GPRS, 107

GSM, 15GeOS, 11HLR, 9Hand-held Computers, 37Home Agent, 102I-TCP, 111IDEN, 15, 21IEEE 802.11, 20, 95IMT 2000, 67, 69IMT-2000 spectrum assignment, 71IPv4, 102, 105IPv6, 105IS-136, 15IS-54, 15Infrared, 19Infrastructure Mode, 20IrDA, 19Isolation-Only Transaction, 129Kangaroo Transactions, 127LEO, 21Laptops, 45Loose Source Routing, 101Loss Profile, 112Lucent, 20MEMS, 60MNCRF, 119MOWSER, 120MPEG 1, 26MPEG 2, 26Macro Cell, 22Marco, 40Micro Cell, 22Microsoft Exchange, 93Microsoft, 11MobiTex, 21Mobile Agents, 133Mobile IP, 102

Page 187: Mobile Computing

166 INDEX

Mobile Transactions, 126Mobile-IP, 102MobileWare Office Server, 93Mobility Middleware, 21NCR, 20NTT, 56Netscape, 93Noise to Signal Ratio (NSR), 17Nokia 9000, 40Nomadic Computing, 2Notebooks, 43Novell, 93Odyssey, 124OpenAir Interface, 20Oracle Lite, 91Oracle Mobile Agents, 91Oracle Replication Manager, 92Oracle Software Manager, 92PCS, 15PDA, 38PHS Wristwatch Phone, 56Palm Pilot, 36Palm-OS, 11Proxim, 20QoS, 11Quality of Service, 109RAM Network, 21RBOC, 21RangeLAN, 20Remote Access, 93Remote-Node, 89Replication, 90Ring Keyboard, 58Rover, 125SDMA, 84

Shiva PPP, 93–94Spread Spectrum, 16Sub-notebooks, 43Sybase SQL Remote, 92System on Chip (SOC), 61TCP/IP, 107TD-CDMA, 73TIA, 15TIA/EIA-136, 15The Open Group, 119Thin Client, 134Third generation network requirements,

69Time-Division Multiple Access (TDMA),

15UMTS schedule, 72UMTS, 71Ubiquitous Computing, 2, 50VLR, 9W-ATM, 86W-CDMA, 26, 77WAP Protocol, 116WaveLAN, 20Wearable Computer, 52WebExpress, 121Windows-CE, 11, 20Wireless Application Protocol (WAP),

96Wireless LAN, 20Wireless Service Provider (WSP), 4Wireless World Wide Web (W4), 119Wireless network evolution, 73X.25, 107Zaurus, 38