AN ARCHITECTURE FOR CONTEXT-SENSITIVE

TELECOMMUNICATION APPLICATIONS

Agnieszka Lewandowska, Maik Debes and Jochen Seitz

Division of Communication Networks

Technische Universit

at Ilmenau

Keywords:

Context deﬁnition, context information, adaptation, personalisation, ubiquitous computing.

Abstract:

Nowadays, everybody utilizes information from different sources and services from various providers. How-

ever, one must know how to ﬁnd the interesting information and how to access services. Wouldnt it be better

if information and service access could adapt to the user and his interests? This approach is the central point

of the ubiquitous computing research. And the adaptation is done based on the current context describing

the user and his equipment. Although there are many projects that claim to be context sensitive, a general

architecture for context-sensitive applications has not yet been introduced. This paper closes this gap and

describes an architecture that has been developed for assisting handicapped tourists during their holidays in

the Thuringian Forest. However, this architecture can easily be generalized for any context-sensitive telecom-

munication application, because it gives a well-deﬁned and ﬂexible model for context information and deﬁnes

several processes to gather, transfer, process and apply context information.

1 INTRODUCTION

Information has become inestimable goods in today’s

society. With the rise of the Internet, information can

easily be retrieved from any computer connected to

the network, and when using mobile communication

technologies, access to information is possible at any

time and from any place. However, information is al-

ways linked to the person who supplies them, and not

to the person who would like to have access to them.

An obvious example is trying to ﬁnd information in

the Internet: one normally uses an Internet search en-

gine, gets thousands of entries matching the speciﬁed

keywords, and has to eliminate all the entries that do

not contain the desired information. Thus, informa-

tion that is well suited for the person trying to retrieve

them is desirable.

Ubiquitous computing is an approach to alleviate

the access to information by supplying user interfaces

that do not rely on traditional computing technique. In

a ubiquitous computing environment, the user inter-

acts with the contained components in a rather com-

mon way without having to become accustomed to

new interfaces. The technologies weave themselves

into the fabric of everyday life until they are indistin-

guishable from it (Dey, 2001). The interface to the

environment, hence, adapts to the persons wanting to

utilize it. And to carry this idea one step further, not

only the interface but also the information exchanged

through this interface must be adapted to the desires

of the user.

This personalization of services and information is

usually directly coupled with the collection and us-

age of context information. In order to collect, trans-

mit and process context information, they must be

modeled and described in a formally deﬁned language

(Debes et al., 2005). Then, they are used to adapt ser-

vices and information to the user.

Although context-sensitivity is application inde-

pendent, it is easier to start with a certain applica-

tion to deﬁne speciﬁc context information. In our

project, we are working on a system aiding handi-

capped tourists on their holidays in the Thuringian

Forest (TAS, 2005). We want to provide them with

barrier-free access toany service and information they

desire. Thus, context-sensitive services play a major

role in this project. The tour planned for the tourists

must be adapted to their abilities and handicaps. Fur-

thermore, the tourist is provided with services and

information which are optimized to the current con-

text. Service providers are able to store their offers

and services, or at least information about their ser-

Lewandowska A., Debes M. and Seitz J. (2005).

AN ARCHITECTURE FOR CONTEXT-SENSITIVE TELECOMMUNICATION APPLICATIONS.

In Proceedings of the First International Conference on Web Information Systems and Technologies, pages 40-47

DOI: 10.5220/0001228700400047

 SciTePress

vices and where these services are accessible, in a

central database. Furthermore, this database contains

a detailed map of the according tourist area. The

subscriber is now able to access information about

navigation, routing and additional useful information

(opening hours, sights, shopping possibilities etc.).

The architecture of TAS enables the seamless integra-

tion of new services and the exploitation of innovative

context information. Communication is conducted

through existing networks, which are connected with

the Internet. The basic ideas are described in more

detail in (Debes et al., 2003).

The paper is structured as follows. In the next sec-

tion, several recent and current projects and systems

are introduced that have the same scope as the archite-

cure described in section 3. There, we show how con-

text information are gathered, transported to the cen-

tral server, analysed and used for adapting services

and information offers. Finally, section 5 will sum-

marize the paper and give an outlook on future work.

2 RELATED WORK

Context-sensitive systems have emerged in different

areas of today’s life. However, the term “context” is

treated differently in the literature. Furthermore, on

one hand a number of authors use a deﬁnition con-

sidering quite concrete properties which is directly

related to a considered object. On the other hand,

one ﬁnds quite general deﬁnitions in the literature but

which are only valid for special application cases.

Nevertheless, in our opinion the deﬁnition of A.

Dey and G. Abowd is the most appropriate. They

state that context deﬁnes a subset of a physical or con-

ceptual status, which is of interest to a certain entity.

Their deﬁnition includes the complete range of con-

text in such a universal way that it can be applied to

very different applications. Dey and Abowd propose

the following deﬁnition (Dey and Abowd, 2000):

Context is any information that can be used

to characterize the situation of entities (i.e.

whether a person, place or object) that are con-

sidered relevant to the interaction between a user

and an application, including the user and the ap-

plication themselves. Context is typically the lo-

cation, identity and state of people, groups and

computational and physical objects.

Projects, which are related to context-sensitivity,

are numerous so the following list of projects cannot

be complete. Therefore, we focus our list on projects

in the area of navigation and localization, since lo-

calization information belongs to the most important

type of context information (Hazas et al., 2004). Nev-

ertheless, this list shall provide a general impression

of the importance of this research.

• The aim of the Lancaster Guide project is to in-

vestigate the provision of context-sensitive mobile

multimedia computing support for city visitors. In

essence, the project will develop a number of hand-

portable multimedia end-systems which will pro-

vide information to visitors as they navigate an ap-

propriately networked city (Davies et al., 2005).

• The project REAL (Resource-Adaptive Localiza-

tion) was done at the Universit

at des Saarlands in

Saarbr

ucken, Germany. It utilizes GPS navigation

for outdoor movements and localization based on

infrared technologies indoors. The system reacts

on the situation the user is in and, thus, adapts to

the user context (Baus et al., 2002).

• The project DRISHTI from the University of

Florida in Gainesville focused on a navigation sys-

tem for visually impaired and disabled persons in

indoor and outdoor areas (Helal et al., 2001).

• At the Technische Universit

at Vienna, the project

LoL@ (Local Location Assistant) implements a

prototype that offers several services over UMTS.

Being a mobile GIS application, these services con-

sist of map- and position-dependent multimedia in-

formation. Routes through the town center of Vi-

enna are calculated, which allow multimedia inter-

action, an electronic travel diary and information

about the main points of interest (Umlauft et al.,

2002).

• Keidl and Kemper from the University of Passau

propose a context framework that facilitates the de-

velopment and deployment of context-aware adapt-

able web services. This framework led to the Ser-

viceGlobe system, which is an open and distributed

web service platform (Keidl and Kemper, 2004).

However, we still have not found a general archi-

tecture for implementing context-sensitive telecom-

munication applications. Combining all the experi-

ences of the projects we have investigated, we came

to the general architecture described in the next sec-

tion.

3 THE PROPOSED

ARCHITECTURE

This section introduces the architecture developed for

the TAS system. Figure 1 gives an overview of this

architecture. It can be roughly separated into four

views:

1. The INPUT view is concerned about collecting

context information.

2. The PROCESSING view deals with analyzing and

applying context information.

AN ARCHITECTURE FOR CONTEXT-SENSITIVE TELECOMMUNICATION APPLICATIONS

Figure 1: An overview of the introduced architecture.

3. In order to select a service appropriate for the given

context, a third view called SERVICE SELECTION

was created.

4. Finally, the service and/or information required by

the user must be presented according to the current

context, which is done in the OUTPUT view.

Thus, context information must be gathered, trans-

mitted, analyzed and applied to the selection and pre-

sentation of information and services. The following

subsections detail these speciﬁc tasks.

3.1 Gathering Context Information

Gathering context information (describing the INPUT

view) can be accomplished differently. For the TAS

architecture, three options are taken into account. The

ﬁrst option, called explicit feedback is based on di-

rectly interrogating the user for context information.

This can be done either by a direct interview or by

a questionnaire the user has to ﬁll out. This method

is very accurate and results in dependable context in-

formation. However, the user has to take some time

to complete the questionnaire or the interview and is

thus distracted from the activities he wanted to do,

which might lead to discontent. Furthermore, there

is the risk of asking too much of the user so that he

might not be willing to answer any further questions.

Thus, explicit feedback should only be applied if few

and clear questions can be formulated that lead to well

deﬁned but rather invariant context information.

The second option, according to the ﬁrst option

named implicit feedback (Poslad et al., 2001), tries to

extract context information from observing user inter-

actions with the equipment he utilizes. Thus, when-

ever the user asks for certain services of information,

the systems logs these requests and tries to classify

them. When a certain class is requested more often

than others, it looks like the user is interested in this

special kind of service and the system tries to orga-

nize information and service offers accordingly. Es-

pecially for advertisements, this kind of feedback is

very useful. The user is not directly involved in es-

tablishing the context and thus not molested like in

the explicit feedback method. However, the correct-

ness of the information cannot be proven, though it is

rather likely that the derived context information align

with the user preferences and interests, especially if

this logging process can be done over a quite long pe-

riod of time.

Furthermore, in our system, a third option of

gathering context information has been implemented.

This option is based on sensing the environment of the

user and collection these sensor information, which

was implemented by several other projects as well,

e.g. in the Context Toolkit (Salber et al., 1999). There-

WEBIST 2005 - INTERNET COMPUTING

fore, we call this option sensory feedback. Sensors

may be attached to the users themselves (e.g. to es-

timate the physical conditions), to the mobile equip-

ment (e.g. to measure the ambient temperature) or

in the close vicinity to the user (e.g. for information

about weather conditions). Additionally, these sen-

sors can be complemented by special actors. A rather

prominent example is using RFID technology (Ra-

dio Frequency IDentiﬁcation). The environment is

equipped with several RFID tags and the user equip-

ment is augmented with an RFID reader. This reader

can receive the answers sent out by RFID tags which

contain the tag’s identiﬁcation. With this identiﬁ-

cation, the actual position of the user can be accu-

rately determined and thus GPS localization can be

improved.

It is quite obvious that all three options for gath-

ering context information must be combined and im-

plemented to build up a comprehensive model of the

actual user context. Thus, our system starts with in-

terviews and questionnaires in the explicit feedback

phase. Then during the user’s stay, the context is be-

ing amended by implicit and sensory feedback.

3.2 Transmitting Context

Information

All context information must then be transmitted to a

central server for evaluation. Therefore, the TAS ar-

chitecture must be detailed, which is depicted in ﬁg-

ure 2.

The heart of this architecture is the TAS headquar-

ter, where a central server is located that is responsi-

ble for processing the context information. It is con-

nected to the Internet via an Internet service provider

and it incorporates ISDN router so that a direct dial-up

connection via the telephone system is also possible.

This option was implemented, because the TAS ar-

chitecture also provides barrier-free access to services

or information by information terminals specially de-

signed for handicapped persons by “systems engineer-

ing ilmenau sei”, one of the cooperation partners in

the TAS project. These terminals provide information

via a web interface accessible through a touch screen,

which is adjustable in three dimensions.

Users can also get information (and pass their con-

text information) using their PC at home or in an In-

ternet cafe. But the main focus of the TAS project is

on users on their way through the Thuringian Forest.

Since the users are mobile, the transport of context

information from the mobile user equipment should

rely on wireless technology. The TAS system does

not restrict information transport to certain network-

ing technologies. It rather is open for any current or

even evolving mobile communication technology.

Currently, the mobile equipment can use Bluetooth,

Wireless LAN (according to IEEE 802.11), GSM,

GPRS or even UMTS to transfer the collected sen-

sor information to a central server. This ﬂexibility

could be achieved by using the Internet Protocol as

a basis for information transport. Utilizing a spe-

cial handover mechanism (Evers, 2004) the mobile

equipment always chooses the best available com-

munication network to send the context information.

Furthermore, this mechanism allows a seamless han-

dover between different network technologies. Thus,

if the user started to communicate in a Wireless LAN

and left the radio coverage of this WLAN, the mo-

bile equipment would change to another network, e.g.

GPRS. Although the mobile equipment would then

receive a new Internet address, ongoing communica-

tion would not be interrupted. This is accomplished

by using a proxy architecture based on the SOCKSv5

standard (Leech et al., 1996).

3.3 Analyzing Context Information

Once context information have arrived at the central

server, they must be analyzed and evaluated, the sub-

ject of the PROCESSING view. The essence of all the

transmitted information is then stored in a data base,

which is constantly being updated. Thus, a concise

data base model for context information is required.

In the TAS project, the following three types of con-

text information are distinguished: user context, tech-

nical context and environmental context.

3.3.1 User context

According to (Cheverst et al., 2000), user context

comprises all information that are directly connected

to a speciﬁc person. In our project, the following in-

formation are important:

• User data

This item contains all information about a user,

which can be further subdivided into his personal

data (name, given name, date of birth, home ad-

dress, visiting address, etc.), his interests (hobbies,

relevant news, etc.) or his preferences (e.g. about

food or telecommunication service provider). Fur-

thermore, user data may be enhanced by logged

user activities as described in section 3.1.

• Group membership

For helping handicapped tourists on their holidays,

it is useful to know, whether they are accompanied

by other persons or whether they are on their own.

• Physical status

Since the system should be useful for physically

handicapped persons, this person’s context par-

tially consists of information about sensor data

describing the health status, like heart frequency,

blood pressure, body temperature and alike, the

AN ARCHITECTURE FOR CONTEXT-SENSITIVE TELECOMMUNICATION APPLICATIONS

Internet

Router

ISDN

Router

TAS

Headquarter

Internet

Service Provider

RFID Tag

Tourist

on the way

Tourist

at Home

Head

Office

Server

Data

Web

Con-

tents

Server

Data

Backup

Internet

Router

Mobile

Terminal

Equipment

Web

Con-

tents

Barrier-free

Info Terminal

RFID Tag

Barrier-free

Info Terminal

Figure 2: The communication infrastructure of TAS.

emotional or mental state, the speed of motion or

the current physical activity.

• Handicap

Finally, kind and degree of the person’s handicap

must be stored in the context information data base.

3.3.2 Technical context

Technical context describes the user equipment and

the technical communication infrastructure that is or

can be used. Thus, the context information database

contains the following information:

• User equipment

User equipment is described by its type (PDA, mo-

bile phone, laptop, etc.), its grade of mobility, its

display characteristics (size, resolution, color, etc.),

its input features (e.g. number of hardware but-

tons), its audio capabilities, its speech recognition

and synthesis features or its battery status.

• Technical environment

For the technical environment, especially the avail-

able communication networks with their communi-

cation capabilities are recorded.

3.3.3 Environmental context

Finally, all other context information come from the

environment. According to (Schmidt-Belz et al.,

2002), this group of context information consists of

• Time

The entry “time” has many different facets. Of

course, the current time is necessary to character-

ize the context entries. Furthermore, time also re-

lates to speciﬁc objects in the environment (just like

opening hours of a museum or the remaining time

in a public swimming pool). Thus, time must be

speciﬁed using a type “entry”.

• Localization

Another very important information is the current

WEBIST 2005 - INTERNET COMPUTING

position of the user. Firstly, it is distinguished

whether the person is inside a building or outside.

According to this information, coordinates give the

current position, e.g. GPS data, recently received

RFID tag or alike.

• Orientation

A very valuable enhancement to the localization

part of the environmental context is the orienta-

tion of the user, which may be derived from dif-

ferent sensors like head-mounted displays or GPS

receivers. Thus, the different information sources

must be taken into account in this context entry.

• Environment

Rather general is the last entry that describes infor-

mation about the environment like weather condi-

tions, road conditions, or other travel information.

3.3.4 Data base entry for context information

As described earlier, the processed context informa-

tion is stored in a data base. Therefore, we deﬁned

the data base structure as shown in table 1.

This database entry is then used to adapt informa-

tion and services for the concerning user.

3.4 Applying Context Information

Once context information is stored in the data base,

the next view called SERVICE SELECTION is re-

sponsible for utilizing them to choose and adapt ser-

vices and information.

In this view, three kinds of interaction are distin-

guished. In the user driven interaction, the user ex-

plicitly requests a certain service or some actual infor-

mation. According to the stored context information,

the service provider is chosen, which ﬁts best for the

requirements of the user. A simple example might be

that the user on his way through the Thuringian Forest

would like to know the way to the nearest restaurant.

Being a vegetarian (which is of course stored in the

context information data base), he certainly would be

very annoyed if the system would choose a restaurant

serving only a very restricted selection of vegetarian

food. Moreover, according to the “time” information

the system checks whether the restaurant is open and

ready to serve some food. In a second step, the user

interface for this service is adapted to the user pref-

erences and the current technical environment. Thus,

if the user is impaired of seeing, an acoustical pre-

sentation of the route to the restaurant should be pre-

ferred to a simple textual one, if the user equipment

provides an acoustical output device. And in align-

ment with the current networking technology the user

can communicate over, the information is coded to

achieve minimum delay and cost.

The second kind of interaction is the server driven

interaction. This is started whenever the context in-

formation is changing beyond a given threshold. As

described in section 3.1, context information is being

collected and updated constantly. Hence, the server

might notice a change of context the user should be

informed about. For example, a user has requested

a route to a certain point of interest and received a

route. Then, sensors notify about changing weather

conditions, which means a storm is on its way. Thus,

the system sends out a warning that the walk to the

chosen point of interest would become quite unpleas-

ant and suggests another point of interest according to

the user’s interests.

The third kind of interaction is ﬁnally the user

equipment initiated interaction. As user equipment

also contributes to the context information, it also

may observe the context state. Whenever critical con-

ditions occur, e.g. the battery is quite low or the heart

frequency drops below a given threshold, it may initi-

ate interaction with the user, the TAS headquarter or

with other persons the user is in a group with. This

helps to initiate emergency actions when a user gets

lost or when his health state requires immediate coun-

termeasures.

4 IMPLEMENTATION

The TAS system is currently being implemented. Pro-

totypes have been set up and demonstrated. In March

2005, the system will run in Georgenthal, a village in

the Thuringian Forest, where special installations for

handicapped persons already have been put up.

During all the project time, the team members co-

operated with several associations for handicapped

persons to be able to understand their special needs

and requirements. Nevertheless, we found out early

that handicapped persons will never rely on technical

equipment only. Although we tried to increase reli-

ability as much as possible, simple technical events

like low battery, unavailable GSM coverage or inac-

curate GPS positioning prevent the system from be-

ing totally reliable. Therefore, we think of TAS as

being an assisting system that simpliﬁes the holidays

for handicapped tourists in the Thuringian Forest.

5 CONCLUSION

Within this paper, we presented the TAS system and

focused on the context-sensitivity of the system. We

showed how context information is gathered, trans-

mitted, processed and applied. We have developed a

comprehensive and ﬂexible data base scheme for stor-

AN ARCHITECTURE FOR CONTEXT-SENSITIVE TELECOMMUNICATION APPLICATIONS

Table 1: Example of a data base entry

0176

PERSONAL DATA

name

Anna

hobby

music

USER DATA

INTEREST

current interests

shopping

individual person

GROUP

MEMBERSHIP

group

---

heart frequency (strokes/minute)

blood pressure (mmHg)

120/80

body temperature (°C)

36,2

HEALTH STATE

sugar content (mg/dl)

---

slow

---

normal

STATE

SPEED OF MOTION

quick

---

type

seeing hindrance

PERSONAL CONTEXT

HANDICAP

degree (%)

TYPE

Pocket PC

fixed

---

MOBILITY

mobile

resolution (dpi)

240x320

DISPLAY

colour

monochrome

number

HARDWARE BUTTONS

emergency call

AUDIO OUTPUT

SPEECH SYNTHESIS

state of battery (%)

EQUIPMENT

BATTERY

operating time remaining (minutes)

ISDN

---

GSM/ GPRS

TECHNICAL CONTEXT

TECHNICAL

ENVIRONMENT

NETWORK

WLAN

---

KIND available time (hours)

2,5

date

15.12.2004

TIME

TYPE

hour

08:30:46

outside of buildings

PLACE

within buildings

---

GPS

°°

56´22´´eL

°°

40´57´´nW

GSM

---

LOCALIZATION

KIND OF DATA

RFID-data

---

direction

north

ORIENTATION

BEARING KIND

coordinates

---

temperature (°C)

ENVIRONMENTAL CONTEXT

ENVIRONMENT

WEATHER

weather conditions

rain

ing and updating context information and considered

different sources of context information.

With context information, services and information

provided to users can be adapted and ﬁltered accord-

ing to the user’s interests and preferences. However,

one often experiences that smart applications rapidly

start to annoy the user because they do not allow him

to directly inﬂuence the way the application works.

We therefore will closely investigate the effect of our

adaptation on the user and try to ﬁnd out how content

he or she is with the system. As we do not assume

that the user of our system is a computer expert, we

will have to do many interviews to come to reliable

statements about the acceptance of the system.

The work described in this paper is part of a project

for the InnoRegio programme (BMBF, 2004) called

”Assisting System for Tourists” (Touristisches Assis-

tenzsystem TAS), which is funded by the German Fed-

eral Ministry of Education and Research (BMBF).

REFERENCES

Baus, J., Kr

uger, A., and Wahlster, W. (2002). A resource-

adaptive mobile navigation system. In International

Conference on Intelligent User Interfaces (IUI2002),

San Francisco, California, USA. ACM Press.

BMBF (2004). InnoRegio - more than a funding pro-

gramme.

http://www.innoregio.de/index

e.php. Bundesminis-

terium f

ur Bildung und Forschung (German Federal

Ministry of Education and Research) BMBF.

Cheverst, K., Davies, N., Mitchell, K., Friday, A., and Ef-

stratiou, C. (2000). Developing a context-aware elec-

tronic tourist guide: Some issues and experiences. In

Proceedings of the 2000 Conference on Human Fac-

tors in Computing Systems (CHI-00), The Hague, The

Netherlands. ACM Press.

Davies, N., Cheverst, K., and Mitchell,

K. (2005). The guide project.

http://www.guide.lancs.ac.uk/GuideOverview.html.

Debes, M., Heubach, M., Seitz, J., and Tosse, R. (2003). In-

formation without barriers – a new form of ubiquitous

computing. In 48. Internationales Wissenschaftliches

WEBIST 2005 - INTERNET COMPUTING

Kolloquium IWK 2003, Ilmenau, Germany. Technis-

che Universit

at Ilmenau, Fakult

at f

ur Elektrotechnik

und Informationstechnik, Technische Universit

at Il-

menau (Th

uringen).

Debes, M., Lewandowska, A., and Seitz, J. (2005). De-

ﬁnition and implementation of context information.

In Kyamakya, K., Jobmann, K., and Kuchenbecker,

H.-P., editors, Joint second Workshop on Positioning,

Navigation and Communication 2005 (WPNC ’05)

& ﬁrst Ultra-Wideband Expert Talk 2005 (UET ’05),

pages 63–68, Aachen. IANT, University of Hannover,

Shaker Verlag. ISBN 3-8322-3746-1.

Dey, A. (2001). Understanding and using context. Personal

and Ubiquitous Computing Journal, 5(1):4–7.

Dey, A. and Abowd, G. (2000). Towards a better under-

standing of context and context-awareness. In Work-

shop on the What, Who, Where, When and How of

Context-Awareness, afﬁliated with the CHI 2000 Con-

ference on Human Factors in Computer Systems, The

Hague, The Netherlands. ACM Press.

Evers, F. (2004). Unterst

utzung eines vertikalen Handovers

zwischen GPRS und WLAN durch einen Proxy.

Diplomarbeit, Division of Communication Networks,

Technische Universit

at Ilmenau, Ilmenau, Germany.

(in German).

Hazas, M., Scott, J., and Krumm, J. (2004). Location-aware

computing comes of age. IEEE Computer, 37(2):95–

97.

Helal, A., Moore, S., and Ramachandran, B. (2001). Dr-

ishti: An integrated navigation system for visually im-

paired and disabled. In 5th International Symposium

on Wearable Computers ISWC2001, Zurich, Switzer-

land.

Keidl, M. and Kemper, A. (2004). Towards context-aware

adaptable web services. In Proceedings of the 13th

International World Wide Web Conference, pages 55–

65, New York, NY, USA. ACM. ISBN 1-58113-912-

Leech, M., Ganis, M., Lee, Y., Kuris, R., Koblas, D., and

Jones, L. (1996). Socks protocol version 5. Request

for Comments 1928, Internet Egineering Task Force –

Network Working Group. Category: Standards Track.

Poslad, S., Laamanen, H., Malaka, R., Nick, A., Buckle,

P., and Zipf, A. (2001). CRUMPET: Creation of user-

friendly mobile services personalized for tourism. In

Proceedings of the IEE Second International Confer-

ence on 3G Mobile Communication Technologies (3G

2001), number 477 in IEE Conference Publication,

London, UK.

Salber, D., Dey, A., and Abowd, G. (1999). The context

toolkit: Aiding the development of context-enabled

applications. In ACM SIGCHI Conference on Human

Factors in Computing Systems (CHI 99), pages 434–

441, Pittsburgh, Pennsylvania, USA.

Schmidt-Belz, B., Poslad, S., and Zipf, A. (2002). Creation

of user-friendly mobile tourism services. In Work-

shop on Mobile Tourism Support Systems in conjunc-

tion with the fourth International Symposium on Hu-

man Computer Interaction with Mobile Devices (Mo-

bile HCI’02), Pisa, Italy.

TAS (2005). Touristisches Assistenzsystem f

ur bar-

rierefreien Zugang zu Urlaubs-, Freizeit- und

Bildungsaktivitten (TAS).

http://rudi4.systemtechnik.tu-

ilmenau.de/webseiten

tas/index.html. (in German).

Umlauft, M., Pospischil, G., Niklfeld, G., and Michlmayr,

E. (2002). Designing lol@, a mobile tourist guide

for UMTS. In 4th International Symposium on Hu-

man Computer Interaction with Mobile Devices (Mo-

bile HCI ’02), Pisa, Italy.

AN ARCHITECTURE FOR CONTEXT-SENSITIVE TELECOMMUNICATION APPLICATIONS