Content uploaded by George Giaglis
Author content
All content in this area was uploaded by George Giaglis on Jan 07, 2014
Content may be subject to copyright.
1
A Design Theory for Pervasive Information Systems
Panos E. Kourouthanassis
Athens University of Economics and Business
Department of Management Science and Technology
pkour@aueb.gr
Abstract
Pervasive Information Systems (PIS) constitute an
emerging class of Information Systems where
Information Technology is gradually embedded in
the physical environment, capable of
accommodating user needs and wants when
desired. PIS differ from Desktop Information
Systems (DIS) in that they encompass a complex,
dynamic environment composed of multiple
artefacts instead of Personal Computers only,
capable of perceiving contextual information
instead of simple user input, and supporting
mobility instead of stationary services. This paper
aims at proposing a design theory for PIS. In
particular, we have employed Walls et al. (1992)
framework of Information Systems Design
Theories (ISDT) to develop a set of prescriptions
that guide the design of PIS instances. The design
theory addresses both the design product and the
design process by specifying four meta-
requirements, nine meta-design considerations, and
five design method considerations. The design
theory has been validated through the
implementation of an Information System in the
retail context following the theory’s prescriptions.
A field experiment revealed that the design theory
is capable of leading to valuable and acceptable PIS
instances.
1 Introduction
Information technology (IT) artefacts are already
embedded in more places than just our desktop
computers, providing innovative services in ways
unimaginable in the near past. This shift in the
viewpoint of information systems (IS) is commonly
referred to as ‘post-desktop’ (Jonsson, 2002) or
‘ubiquitous’ computing (Weiser, 2002). This trend
has fired a shift away from computers towards
computerised artefacts. A new generation of
information appliances has emerged (Roussos,
2003), differing from traditional general-purpose
computers in what they do and in the much smaller
learning overhead they impose on the user. Instead
of having IT in the foreground, triggered,
manipulated, and used by humans, nowadays we
witness that IT (irrespectively whether it comprises
of computers, small sensors, or other
communication means) gradually resides in the
background, monitoring the activities of humans,
processing and communicating this information to
other sources and intervening should it be required.
This new class of IS is often called ‘Pervasive
Information Systems’ (PIS) (Birnbaum, 1997) and
enables new interaction means beyond the
traditional desktop paradigm.
This position paper aims at proposing a
methodological approach that may facilitate
designers to develop PIS instances. Specifically, we
will present a design theory for PIS. The following
section presents the design challenge of PIS.
Section 3 briefly introduces the methodological
framework that will be used to specify the design
theory, while section 4 presents the theory itself.
Section 5 is concerned with the activities we
undertook to validate the proposed design theory.
The final section concludes with a critical appraisal
of the proposed design theory and its practical
usefulness.
2 The Design Challenge of PIS
2.1 PIS Novel Characteristics
PIS introduce new elements in multiple dimensions
spanning different IS domains, such as Human-
Computer Interaction (HCI) and Software
Engineering, which admonish us to examine them
as a new Information Systems class. In essence, PIS
revisit the way we interact with computers by
introducing new input modalities and system
capabilities. So far, the interaction paradigm for
Information Systems has been the desktop. Thus,
the design and implementation of Information
Systems was based on this paradigm. PIS extend
this paradigm by introducing a set of novel
characteristics that are summarised in the following
paragraphs.
First, PIS deal with non-traditional computing
devices that merge seamlessly into the physical
environment. As such, the desktop (in the form of
the Personal Computer) is just ‘another access
device’. Consequently, conventional HCI design
methods and interaction schemes may not be
appropriate for the new IS class since the physical
2
interaction between users and the system will, most
certainly, not resemble the prevailing DIS
keyboard/mouse/display paradigm. On the contrary,
PIS simulate the way that humans interact with the
physical world. Abowd and Mynatt (2000) argue
that since humans speak, gesture, and use writing
utensils to communicate with other humans and
alter physical artefacts, such actions can and should
be used as explicit or implicit input to PIS. Burkey
(2000) argues that the next step in this progression
refers to environmental interfaces where the
environment is the interface and the user exists in it.
This is fully aligned with PIS where, ultimately,
every artefact can interact with the system user.
Thus, apart from solely physical interactions with
the system, PIS may also incorporate elements of
ambient interactions with devices or objects from
the physical space (Schur, Decker, & May, 1999).
According to the authors, “these interactions
should be lauded for their increased learnability
and general ease of use”. Additionally, they may
be used by people with disabilities or IT
unfamiliarity for whom the traditional mouse and
keyboard are less accessible.
Moreover, PIS support a multitude of
heterogeneous device types that differ in terms of
size, shape (more diverse, ergonomic, and stylistic),
and functionality (simple mobile phones, portable
laptops, pagers, PDAs, sensors, and so on),
providing continuous interaction which moves
computing from a localised tool to a constant
presence. Opposed to desktop environments where
the access devices are stationary, PIS support
nomadic devices which may be carried around by
users and present location-based information. Since
these devices are not required to be a fixed part of
the pervasive system, PIS need to support
spontaneous networking, implying ad-hoc detection
and linking of the participating devices into a
temporary pervasive network creating dynamic
dependencies among the linked devices.
Furthermore, the participating elements of a
pervasive system are highly embedded in the
physical environment. This implies that they will
inevitably interact with the existing architecture of
the environment. Understanding architecture,
therefore, plays a key role in designing PIS.
Designers need to consider how architecture’s
manipulation of space, aiming to minimise
obstacles (Bentley, 1985), interacts with the
pervasive artefacts that will eventually be deployed.
Ultimately, architecture can be seen as a large-scale
system, and as such, we can learn from its presence
and design.
In addition, PIS emerge a revised viewpoint in the
way we perceive system design. ‘Conventional’
system design incorporated more and more of the
physical world inside the computer. In this sense,
the actual system intelligence has been purely
cybernetic, comprising of software designed to
execute predefined tasks and activities efficiently.
Moreover, systems were designed in such ways that
enhanced overall utility and productivity, especially
when applied to organisational contexts. In the case
of PIS many computerised artefacts (instead of a
single computer) monitor and support the user. The
system’s intelligence no longer resides solely in the
computer, but it is embedded in the physical world.
Thus, each artefact may be specialised to support a
single task performed in a more efficient way. This
task may depend on a geographical location or may
be triggered by an event such as a user request, a
sensor reading change, and so on.
Building on the above, in desktop environments
designers typically assume that user profiles are
known in advance (Grudin, 1991a, , 1991b; Lynch
& Gregor, 2004; Poltrock & Grudin, 1994), thus
allowing for systematic requirements analysis. In
PIS, the opposite may be true: it is highly unlikely
for the system designer to know in advance the
kinds of users who will be interacting with the
system. Users may range from being vaguely
familiar with IT to expert users. In addition, PIS
users may be opportunistic in the sense that they
may use the system only sporadically, implying that
they may not be subject to training prior to system
use.
Finally, PIS introduce the property of context-
awareness as a result of the pervasive artefacts
capability to collect, process, and manage
environmental or user-related information on a real-
time basis. Opposed to desktop computing, where
user action precedes system response, PIS promote
system pro-action based on environmental stimuli.
This can be accomplished through the deployment
of sensors and actuators in the physical world. The
following table summarises the differences among
the desktop paradigm (Desktop Information
Systems – DIS) and PIS under six dimensions
answering the following questions:
• What is the generic profile of a PIS user
compared to DIS? (User)
• What interactions is the user expected to
perform? (Task)
• How does the interaction take place? (Medium)
• Where does the interaction take place? (Space)
• What will be eventually designed? (Product)
• When will the system be used? (Time)
3
Table 1: DIS and PIS differentiating elements
Desktop Information Systems Pervasive Information Systems
User • Committed
• Known
• Trained
Role Model: Office Clerk
• Opportunistic
• Unknown
• Untrained
Role Model: Citizen
Task • Specific
• Focused on utility and productivity • Generic
• Focused on service delivery and experience
Medium • Localised
• Homogeneous
• ‘Point and Click’ paradigm
• Constant Presence
• Heterogeneous
• Natural interaction and multimodal paradigm
Space • Cybernetic • Physical
Product • Virtual • Tangible and Virtual
Time • Reactive • Proactive
2.2 The Need of a Design Theory for PIS
DIS, through their form of personal computers or
other stationary access devices, were designed to fit
into an office environment and the activities taking
place there. They were designed to be efficient tools
in the hands of professionals. Thus, their practice of
interaction design is directed towards this setting.
Obviously, everyday life is quite different from
office work, and therefore other ‘places’, interfaces,
and appearances have to be explored in order to
find a broader repertoire of strategies for creating
human-centered technology. Moreover, these
systems are designed for use; this means that their
design and evaluation are accomplished on the
basis of some definition of their functionality and
perceived usage. Thus, designers seek a solution
that satisfies the basic criteria for usability such as
efficiency in use, low error rate, and support for
recovery from error, based on a general knowledge
about what to do and what not to do to meet such
criteria (Hackos & Redish, 1998; Nielsen, 1994).
The objective is to achieve maximum usability with
respect to a general, precise notion of use, and the
design is motivated by this ambition.
Consequently, the existing design approaches for
DIS follow the same rationale. Design methods
such as SSADM (Weaver, Lambrou, & Walkley,
1998), ETHICS (Mumford & Weir, 1979), SSM
(Checkland, 1981), or Object-Oriented Analysis
(Mathiassen, Munk-Madsen, Nielsen, & Sage,
2000) to name but a few popular methods, consider
systems that support predefined tasks and in many
cases assume a job or office environment.
Moreover, such methods rely on the knowledge of
the designer in order to recognise potential
problems and mostly offer a generic approach to
design. Also, such methods tend to focus on details
of systems, something which is possible with
traditional, static systems, but which cannot always
happen for the dynamic and rich environments
supported by pervasive systems. That is why the IS
literature has classified such methods into broad
categories each capable of tackling different
problem classes. Finally, all the traditional
approaches are oriented towards the fundamental
tenets of Human-Computer Interaction: design for a
specific user, performing a specific task, in a
specific domain (Preece, Rogers, & Sharp, 2002).
However, user interactivity with a pervasive system
may not address these particular tenets.
It should be noted that traditional design
approaches are still applicable for PIS environment.
User participation in all its forms (user-centered
design, participatory design, and so on) may still be
employed to augment user acceptance of the final
system. Structured design methods may be
followed to develop pervasive systems. Still, they
should be perceived under the prism of the novel
PIS features. PIS are complex systems. This
complexity may generate increased uncertainty to
designers (e.g. whether they have collected and
accommodated all system requirements). That is
why current approaches to PIS design follow an
iterative and incremental system implementation.
Systems are designed as proof-of-concept
prototypes that gradually evolve into final-cut
systems. Emphasis is given to the successful
deployment of technology and not on its actual use.
Thus, priority is given to the resolution of
engineering challenges, thus forcing designers to
selectively omit from their design important system
qualities such as usability, privacy management,
and aesthetics.
This argument illustrates the value and utility of our
research. A design theory may provide designers
with a tool that holistically examines PIS and
identifies, in a nut-shell, the key features that
4
should be addressed during the design process.
Such a tool may influence the traditional design
approaches and methods and suggest how these
may be adapted to efficiently guide the design
process in pervasive environments.
At the same time, the PIS domain itself requires the
formulation of a comprehensive framework that
supports designers throughout the design process.
So far, PIS design has been stemmed from practice.
It is a trial-and-error process. Thus, the knowledge
generated so far presents only fragments of the PIS
picture as a whole. Since PIS is a technology-driven
phenomenon, some efforts that attempt to guide
designers in a systematic way also perceive PIS
design from a technical perspective. Others, view
PIS design from a purely social perspective and
emphasise on privacy management or
environmental management issues. To summarise,
there is no consolidated framework for PIS design.
The absence of a common frame of reference forces
most of current PIS implementations to be
designed, deployed and evaluated in relative
isolation. Indeed, current PIS implementations
follow a vertical and ad-hoc approach
implementing from scratch all the required
elements based on the unique characteristics of the
application domain.
3 Information Systems Design Theories
To specify our design theory we have followed the
framework of Information Systems Design
Theories (ISDT) that was first articulated by
(Walls, Widmeyer, & El Sawy, 1992). An ISDT
aims at the design of classes of Information
Systems, rather than the development of specific IS
instances. To this end, design theories are
prescriptive, in the sense that they provide
constructs and guidelines for the achievement of
stated goals, rather than explaining phenomena
(explanatory theories) or predicting outcomes
(predictive theories). Finally, design theories are
theories of procedural rationality (Simon, 1996), as
their objective is to prescribe both the properties
that an artefact should have if it is to achieve certain
goals, and the methods of artefact construction.
According to Walls et al. (1992), design theories
have two aspects; one dealing with the product of
design (the artefact that will form the outcome of
applying the design theory) and another dealing
with the process of design (i.e. the method by
which the design product can be realised). We will
use this distinction to describe the components that
form an ISDT. These components are summarised
in Table 2.
The framework has already been used to develop
design theories for, amongst others, Executive
Information Systems (EIS) (Walls, Widmeyer, & El
Sawy, 1992), Decision Support Systems (Kasper,
1996), Group Decision Support Systems (Limayem,
1996), Organisational Memory Information
Systems (Stein & Zwass, 1995), Simulation
Systems for IS Evaluation (Giaglis, 1999), and
Emergent Knowledge Processes (Markus,
Majchrzak, & Gasser, 2002). The following section
presents the proposed design theory for PIS.
Table 2. ISDT Components
DESIGN PRODUCT
Kernel Theories Theories from reference disciplines that govern design requirements
Meta-Requirements The class of goals to which the theory applies
Meta-Design A class of artefacts hypothesised to meet the meta-requirements
Design Product Hypotheses Used to test whether the meta-design satisfies the meta-requirements
DESIGN PROCESS
Kernel Theories Theories from reference disciplines that govern the design process
Design Method Description of procedures for artefact construction
Design Process Hypotheses Used to test whether the design method results in an artefact consistent with
the meta-design
5
4 Design Theory Components
4.1 Meta-Requirements Elicitation
The first meta-requirement refers to the profile of
prospective PIS users. In the PIS context it is highly
unlike for the system designer to know in advance
the types of people who will be using the system.
Take into consideration the users of the different
implementations of tour guides existing in the PIS
literature (Abowd et al., 1997; Bederson, 1995;
Bellotti, Berta, De Gloria, & Margarone, 2001;
Davies, Cheverst, Mitchell, & Efrat, 2001); users
may range from people that are vaguely familiar
with IT (mainly due to their interaction with
commonplace IT artefacts such as mobile phones)
or (at extreme cases) techno-phobic. Moreover,
these types of users are opportunistic in the sense
that they will use the tour guides for a particular
time frame and for a particular reason (in this
example, to augment their visiting experience). To
this end, it is highly unlikely that these users will be
subject to thorough training in the system’s use as
in the case of DIS users. Conclusively, a PIS
instance should support all the different user types
by employing sufficient mechanisms that enhance
or facilitate user interactions with the system, while
at the same time haste users’ learning curve for
using the PIS.
The second meta-requirement refers to the PIS
capability to support the multitude of different
device types that may participate in the pervasive
environment. Hansmann et al. (2003) distinguish
among four types of devices: information access
devices, intelligent appliances, smart controls, and
entertainment systems. Nevertheless, ideally a PIS
should support any device that has built-in active
and passive intelligence. Therefore, devices’
heterogeneity is the most important element that
should be addressed during the design of a PIS.
Moreover, the plurality and diversity of pervasive
devices generate additional system requirements in
terms of connectivity and integration between them
(Saha & Mukherjee, 2003). Furthermore, the design
of the application and user interface should take
into account the unpredictability of end devices. In
the case that an application follows the user and
moves seamlessly between devices, it is implied
that this application will have to adapt to changing
hardware capabilities (different types of pointing
devices, keyboards, network types, and so on) and
variability in the available software services
(Banavar & Bernstein, 2002).
Designing for manipulation of contextual
information, implies that a PIS should be able to
perceive relevant information of its environment
(with location sensitivity and user identity capturing
being the minimum requirement as stated by
(Abowd & Mynatt, 2000; Dey, 2001), process it,
and adapt to changes in the environment taking into
account both historical and current data. Although
at present contextual information refers mainly to
the users’ current location, we expect that in the
near future PIS will be able to perceive
simultaneously multiple stimulants that may be
contradictory one to the other. Thus, this meta-
requirement suggests that PIS should accommodate
an appropriate mechanism that will filter the
different contextual information particles, process
them, and adjust their behaviour according to the
information that best suits the current occasion.
The final meta-requirement suggests that pervasive
artefacts should be ‘gracefully’ embedded in the
physical space. This smooth integration does not
suggest that these IT artefacts should be completely
invisible to the system users, as implied by most
visionary research papers in the field (Norman,
1999; Satyanarayanan, 2001; Weiser, 1993, , 2002).
On the contrary, we follow (Redstrom, 2001)’s
considerations that pervasive technology should be
governed by meaningful presence, promoting
unobtrusiveness. Thus, the challenge is to design
PIS in such a way that users perceive them as part
of the environment. Universal design principles
(Story, Mueller, & Mace, 1998) may be applied to
create remembrances allowing for system usage
with the minimal distraction. Likewise, the PIS
designer should also focus on the aesthetical
qualities of the pervasive artefacts (Djajadiningrat,
Wensveen, Frens, & Overbeeke, 2004).
4.2 Meta-Design Elicitation
The first meta-requirement clearly suggests that PIS
should support opportunistic or inexperienced users
that do not have the luxury, or time, for training to
the system’s functionality. To this end, PIS
designers should devise means that facilitate the
interaction of such users with the system, and
minimise its learning curve. The solution to this
problem is to employ natural, easy to use and easy
to learn interfaces that facilitate a richer variety of
communications capabilities between humans and
computation artefacts. At the same time, PIS
designers should expect that the cases of PIS
misuse will be increased compared to DIS. This is
the result of the potential PIS users’ profile (lack of
experience and/or sufficient training).
Consequently, the system design should incorporate
appropriate mechanisms that minimise the degree
of errors, or guide the user in such a way that
prevents errors from even occurring. Finally, PIS
should be able to perceive users’ current skill level
through both the contextual information and current
6
system usage and adapt their functionality
accordingly.
The diversity and plurality of pervasive devices
poses new challenges for information delivery
applications in this environment. To meet the
demands in this heterogeneous environment, it is
necessary for the information to be customized or
tailored according to the user's preferences, client
capabilities and network characteristics (Held,
Buchholz, & Schill, 2002). As such, PIS should
incorporate a sufficient adaptation system that can
accommodate all different types of adaptations
between different formats. Already several such
mechanisms have been proposed in the literature
(Berhe, Brunie, & Pierson, 2004; Gajos & Weld,
2004; Han et al., 1998; Lei & Georganas, 2001;
Mohan, Smith, & Li, 1999; Ponnekanti, Lee, Fox,
Hanrahan, & Winograd, 2001) providing the system
designer with the option to select the most
appropriate for its system requirements. Moreover,
a PIS should be scalable. Scalability refers to the
ability to incrementally increase the abilities of a
system, whilst maintaining, or improving,
performance (Saha & Mukherjee, 2003). As such,
the issue of scalability refers mainly to ensuring
smooth and unobtrusive communication among PIS
clients and backend hardware infrastructure (such
as sensors and actuators, backend systems, and so
on).
Contextual management implies that sensing
artefacts should be able to effectively communicate
the information they collect and process as well as
trigger events that deem necessary to support PIS
users. Opposed to DIS where the user initiates the
interaction with the system, the PIS vision for
invisibility and unobtrusiveness suggests that the
system is always active, continuously collecting
contextual information, and pro-acting (rather than
re-acting) to the needs and demands of the end
users before they even start expressing them.
Conclusively, the PIS designer should devise an
appropriate mechanism that supports proactive
system operation. Similarly, since it is extremely
difficult to program each participating device and
application in the pervasive environment to receive
and communicate uniformly the information it
collects, PIS designers should devise a
representation format that is efficient enough to
model, process and communicate context.
The final two meta-design considerations stem from
the requirement of smoothly embedding the
pervasive artefacts to the physical space. On the one
hand, PIS should be easily accessible. If designers
follow the extreme suggestion to completely hide
the IT infrastructure, we might end up with a
system that is completely inaccessible due to the
fact that users are unaware of how to use it or, at
extreme cases, of its existence. On the other hand,
pervasive artefacts should be smoothly embedded
in the physical environment. As such, PIS designers
should make sure that the systems which they
create do not conflict with or challenge the
architecture of the place they will be integrated. To
reach that goal, co-operation between two
previously completely different disciplines (namely
IS software engineering and civil architecture)
seems to be the logical solution. Consequently, the
design of PIS should exploit the existing material of
the physical environment and gracefully embed
ITin the physical world.
4.3 Design-Method Elicitation
The first design method consideration suggests that
PIS designers orchestrate the design around the
informal and unstructured activities that users
perform. This is also illustrated in the functionality
of various PIS implementations. For example,
domestic PIS, in their multiple instantiations,
support such activities as home automation
(inventory management, light and heat adjustment,
and so on), or home entertainment. The focus on
activities, as opposed to tasks, is a crucial departure
from traditional HCI design. Of course, activities
and tasks are not unrelated to each other. Often an
activity will comprise several tasks, but the activity
itself is more than these component parts. The
challenge in designing for activities is
encompassing these tasks in an environment that
supports continuous interaction.
The second design method consideration suggests
that prototyping through its various forms
(sketching, low-fidelity designing, and mock-ups,
just to name as a few) should be a core activity of
the design process. Demonstrating prototypes to the
system users will ensure the implementation of
user-friendly interfaces, as well as the incorporation
of user feedback regarding several dimensions of
the system such as usability, functionality, privacy
protection, and so on. Moreover, since PIS have
significant impact to the physical environment
(through the embedment of several pervasive
artefacts), prototyping may assist designers to fix
functional user requirements without implementing
the PIS on a full-scale basis, thus, minimizing
implementation costs in terms of time and financial
resources.
The third design method consideration suggests
integrating conceptual design in the early phases of
the design process. Contextual design (CD) may be
used in the form of sketching or conceptual
scenarios to demonstrate the system’s functionality
to the system’s stakeholders. Given that PIS may
re-engineer the way users perform their tasks and
7
activities, CD may be employed as a technique that
presents alternative usage scenarios, technological
solutions, or interaction techniques, so that system
users may evaluate them, and select the most user-
friendly or appropriate, meeting their goals and
aspirations.
The fourth design method consideration aims at
addressing issues relating to user privacy. Indeed,
context-awareness implies that the system will be
able to monitor and process personal information
such as the users’ current location, activities, even
information related to the human body (e.g. users’
temperature, heartbeats, or respiration levels).
Combining these features with the capability to
store this information for future utilisation, it is not
surprising that privacy protection is considered as
one of the most major properties that a PIS should
take into consideration from the very beginning of
the design process with many researchers proposing
specific guidelines or models that may be applied
(Beckwith, 2003; Beresford & Stajano, 2003;
Jacobs & Abowd, 2003; Langheinrich, 2001; Palen
& Dourish, 2003).
Finally, the design of PIS should not be the concern
of software engineers only. Software engineers
have the necessary skills to analyse and design a
system from an IS perspective: create entity-
relationship diagrams, data-flows, large databases,
select the most appropriate technical solution to the
given problem, and so on. Therefore, they, most
probably, lack the necessary skills to apply their
design solution to the physical space on an effective
manner, not to mention to propose the most
effective solution based on the design problem. As
such, the design team should be enriched with
additional members that may involve types of
people such as architects, or internal decorators,
each providing a different perspective to the design
of PIS. These people should be involved from the
early stages of the design process in order to
counsel software engineers on how to exploit
environments’ smart spaces. Likewise, they may
advise them on how and where to place pervasive
artefacts in an unobtrusive, aesthetical, and possibly
invisible, to the system users’, manner. Finally,
they may recommend alternative layout
propositions regarding the effective exploitation of
the physical space in order to minimise hardware
placement and environment restoration costs.
5 Theory Validation
Because ISDTs propose theoretical contributions
(Walls, Widmeyer, & El Sawy, 1992, , 2004) we
need to empirically test their propositions. Only the
accumulated weight of empirical evidence will, in
essence, establish the validity of any design theory
without any doubt. In our case, the proposed design
theory has been employed for the design of a
Pervasive Retail Information System (PRIS).
Specifically, we followed the theory’s prescriptions
to design and implement a PIS capable of
enhancing the shopping experience in
supermarkets. In the supermarket environment, the
shopper can pick up a wirelessly connected
shopping cart equipped with a display device and a
Radio Frequency Identification (RFID) sensor
capable of scanning the contents of the cart.
The system functionality is summarized to the
following scenario. The shopper could use her
loyalty card to log in the system, which welcomed
her and presented her the shopping list she had
uploaded prior to her visit to the store. She could
then start navigating within the store as usual,
picking up products and placing them inside the
shopping cart. Each time a product was placed in
the cart, the display device showed its description
and price, and updated the total cost of the cart
contents. At the same time, the product was
removed from the shopping list. Moreover, at any
time the shopper could request additional
information about a product (e.g. nutritional value
or ingredients), get informed about ongoing
promotional activities (fully personalised based on
the shopper’s profile and past consumption
patterns), and request navigation assistance within
the store. Finally, during checkout, the system
transmitted the list of purchased products along
with the total amount to the cashier, who billed the
shopper and issued the receipt, thus alleviating the
need for queues and product loading/unloading at
the checkout. The detailed design of the pervasive
system has been published in (Roussos et al., 2003)
and (Kourouthanassis & Roussos, 2003).
To validate the propositions of the design theory we
have generated a set of validation hypotheses
measuring the instance’s value and acceptance. To
assess them, we organised a field experiment in a
Greek supermarket (ATLANTIK) where the PRIS
was used by supermarket shoppers to conduct part
of their shopping. The field spanned over a two-
week period. Sixty loyalty club members
participated in the trial.
Initially, the trial environment was prepared in
terms of (i) selecting the appropriate products, (ii)
preparing a specially modified corridor inside the
supermarket, (iii) preparing the technical
infrastructure to support the trial, and (iv) inviting
the trial participants. Participants were invited
through phone interviews. The majority of the
participants (85%) belonged to the age range of 30-
54 years old, while 77% of the participants were
female. The level of education was high, with over
71% of the participants having university or higher
8
education. Regarding familiarity with information
technology, 66% were relatively familiar with PCs
while 15% had never used personal computers
before.
The execution of the trial was organized in a
sequence of three distinct steps. Initially, the trial
participants were shown the system functionality by
a facilitator (10 minutes). After the end of the
system demonstration, shoppers were prompted to
use the system on their own (30 minutes). Shoppers
were able to purchase the products that were
displayed in the modified corridor. Following their
interaction with the system, the participants
completed a questionnaire evaluating the effect of
the system on their shopping experience.
All research hypotheses have been validating which
indicates that the proposed design theory results to
valuable and acceptable PIS instances. In particular,
subjects evaluated their shopping experience using
the system as amusing and pleasant (with over 70%
characterizing it as exciting), while 78% of them
stated that the system enabled them to monitor
effectively the products in their shopping cart while
at the same time, organizing their supermarket
purchases better. Furthermore, 85% of the
participants stated that the system saved them time
to search for additional information or promotional
offers regarding the products they want to purchase.
Although the prototype was not fully integrated
within the supermarket, shoppers identified that the
use of such a system will eventually improve the
check-out process, while almost all of them stated
that they expect to wait less at the cashiers using
such a system. An interesting observation derived
from the fact that 89% of the trial participants stated
that waiting less time in the cashiers would
influence their decision to shop at a certain
supermarket. Furthermore, shoppers perceived that
the system improved the effect of time pressure
within the supermarket since over 80% responded
that the system offered them more time to conduct
their shopping, while at the same time reduced the
sense of time pressure and contributed to less hurry
in the supermarket.
Regarding the continuous monitoring of the cart
total value, 93% of the participants stated that the
system can help them monitor their budget more
effectively and that such a system allowed them not
to spend more money than they have budgeted for.
Finally, they responded that such systems may
improve promotions effectiveness by presenting
and organizing product promotions efficiently.
System acceptance was measured through the three
major constructs of the Technology Acceptance
Model (Davis, 1989) (perceived usefulness,
perceived ease of use, and perceived intention of
use). The model was enhanced with two additional
variables stemming from the design theory
prescriptions (perceived privacy and perceived
aesthetics). In all cases, the model fitted within
acceptable parameters.
The detailed field experiment results concerning the
system’s value have been published in
(Kourouthanassis, 2004). The field experiment
results concerning the system’s acceptance are
available at (Kourouthanassis, 2006).
6 Conclusions
This position paper outlined the components of a
design theory for the development of PIS. The
theory consists of a set of meta-requirements, a set
of meta-design considerations, and a set of design
method considerations. This research stemmed
from the lack of a consolidated framework to guide
the design process of PIS.
We argue that the proposed design theory is an
important theoretical contribution for several
reasons that are summarised in the points below.
First, the design theory prescribes the design
process and products of an emerging IS class,
which is expected to be the subject of significant
investigation in coming years. This is evident by
several publications in established IS journals,
which advocate the significance of PIS in both
practical and managerial terms, and specify agendas
for further research in the field (Abowd & Mynatt,
2000; Lyytinen & Yoo, 2002; Lyytinen et al.,
2004). Section 2 argued that this emerging field is
in need of a systematic and holistic framework that
consolidates the existing fragmented efforts
regarding the design and development of PIS. Our
work addresses this need by formulating a design
theory for PIS, prescribing both the design product
and the design process.
Second, the design theory provides guidance to
practitioners by demonstrating how traditional IS
design methods need to be modified to support the
development of PIS. In most cases, traditional IS
design approaches provide designers with little
guidance about what to do and hot to do it. In new
and emerging IS areas (such as PIS), the existing
knowledge base is often insufficient for design
purposes and designers must rely on intuition,
experience, and trial-and-error methods. The
proposed PIS design theory collects the current
knowledge and development practices in the PIS
domain and presents them in a systematic and
holistic way enhancing the efficacy of the design
process. Similarly, in accordance to Hevner et al.
(2004) call for effective communication of design
9
science outcomes, the PIS design theory facilitates
managers to conceptualise and understand the PIS
phenomenon. Our research identified the
constituting components and critical factors
affecting the deployment of pervasive technologies
in a particular application domain, as well as their
acceptance by users. Our work provides managers
with a descriptive mechanism that helps them
understand the PIS phenomenon, identify the
capabilities and limitations of pervasive
technologies, and recognise prospective business
opportunities.
The most important limitation of our work refers to
the validation of our theoretical propositions
through a combination of single field experiment.
We acknowledge that our design principles should
be tested empirically in other situations where the
same theoretical conditions hold; namely, PIS
environments. Only the accumulated weight of
empirical evidence will reinforce the validity of our
theoretical propositions. We expect that our design
theory will provide the initial grounds for further
research in this area, collecting PIS common
patterns, and specifying a set of generic design
principles that may be followed by designers when
developing PIS instances.
At the same time, our validation efforts focused on
the utilitarian perceptions of the designed artefact
for its users. We acknowledge that additional
validation is required to address the perceptions of
designers since they are the primary beneficiaries of
this work. Although the design team that was
involved in the prospective validation of the design
theory provided us with several interesting insights,
we believe that our theoretical propositions need to
be exposed to a wider audience comprising of PIS
designers that embrace a variety of design
experiences and IS skills. It should be noted that an
initial step towards this objective has been
performed during the 3rd International Workshop on
Ubiquitous Computing which was organised in
Paphos, Cyprus during May 2006, where we
presented our design theory principles. The
workshop participants agreed, in principle, with our
propositions and expressed the feeling that such an
approach is required for the PIS field. Nevertheless,
we posit that the PIS design theory needs to be the
subject of additional debate and refinement in a
more formal manner.
Finally, the prospective evaluation of the design
theory did not address all of the theory’s
statements. Specifically, the design team did not
address the principle concerning the adaptation of
PIS functionality by identifying its users’ current
level of IT experience. Initially, the design team
aspired to develop a mechanism that passively
identifies the current level of IT experience based
on the pace and number of iterations that shoppers
needed to scan the RF-tag, and the number of clicks
in the system’s available options. However, they
decided not to develop this mechanism because the
expected duration of each shopping session in the
field experiment would be inadequate to properly
test such a feature. The PIS would be used by each
shopper only once since the supermarket’s
management did not provide us with the
opportunity to conduct a longitudinal study.
Moreover, each user would use the system for a
time frame of approximately 30-45 minutes. As
such, we realised that the system would not have
the time required to adjust its services accordingly.
Therefore, testing this element is left for future
research.
The value of this research will be determined by the
application of the proposed design theory by other
scholars, in their effort to design PIS instances. We
believe that this research provides an aggregated
approach to describe PIS characteristics and
prescribe their development. Moreover, it
represents the only consolidated approach, to our
knowledge, that investigates the problem of PIS
design. In any case, the proposed design theory
aims at specifying generic design principles that
should be inherited by PIS instantiations. We
decided to follow that paradigm in order to ensure
that the proposed design prescriptions are
applicable to all PIS instances, irrespective of their
application domain. It is up to the PIS designers to
interpret and apply these prescriptions based on
their design problem.
References
Abowd, G. D., Atkeson, C. G., Hong, J., Long, S.,
Kooper, R., & Pinkerton, M. (1997). Cyberguide: A
Mobile Context-aware Tour Guide. Wireless
Networks, 3(5), 421-433.
Abowd, G. D., & Mynatt, E. D. (2000). Charting
Past, Present, and Future Research in Ubiquitous
Computing. ACM Transactions on Computer-
Human Interaction, 7(1), 29-58.
Banavar, G., & Bernstein, A. (2002). Software
Infrastructure and Design Challenges for
Ubiquitous Computing Applications.
Communications of the ACM, 45(12), 92-96.
Beckwith, R. (2003). Designing for Ubiquity: The
Perception of Privacy. IEEE Pervasive Computing,
2(2), 40-46.
Bederson, B. B. (1995). Audio Augmented Reality:
A Prototype Automated Tour Guide. Proceedings
10
of the ACM Conference on Human Factors in
Computing (CHI '95), New York, 210-211.
Bellotti, F., Berta, R., De Gloria, A., & Margarone,
M. (2001). User Testing a Hypermedia Tour Guide.
IEEE Pervasive Computing, 1(2), 33-41.
Bentley, I. (1985). Responsive Environments: A
Manual for Designers. London, UK: Architectural
Press.
Beresford, A. R., & Stajano, F. (2003). Location
Privacy in Pervasive Computing. IEEE Pervasive
Computing, 2(1), 46-55.
Berhe, G., Brunie, L., & Pierson, J. M. (2004).
Modeling Service-Based Multimedia Content
Adaptation in Pervasive Computing. Proceedings of
the CF'04, Ischia, Italy, April, 14-16.
Burkey, C. (2000). Environmental Interfaces:
HomeLab. Proceedings of the Conference on
Human Factors in Computing Systems, The Hague,
The Netherlands, 47-48.
Checkland, P. (1981). Systems Thinking, Systems
Practice. Chichester, UK: John Wiley and Sons.
Davies, N., Cheverst, K., Mitchell, K., & Efrat, A.
(2001). Using and Determining Location in a
Context-Sensitive Tour Guide: The GUIDE
Experience. IEEE Computer, 34(8), 35-41.
Davis, F. D. (1989). Perceived Usefulness,
Perceived Ease of Use, and User Acceptance of
Information Technology. MIS Quarterly, 13(3),
318-340.
Dey, A. K. (2001). Understanding and Using
Context. Personal and Ubiquitous Computing, 5, 4-
7.
Djajadiningrat, T., Wensveen, S., Frens, J., &
Overbeeke, K. (2004). Tangible Products:
Redressing the Balance Between Appearance and
Action. Personal and Ubiquitous Computing, 8(5),
294-309.
Gajos, K., & Weld, D. (2004). SUPPLE:
Automatically Generating User Interfaces.
Proceedings of the Intelligent User Interfaces (IUI)
2004, Funcha, Portugal.
Giaglis, G. M. (1999). On the Integrated Design
and Evaluation of Business Processes and
Information Systems. Communications of the AIS,
2(5).
Grudin, J. (1991a). Interactive Systems: Bridging
the Gaps Between Developers and Users. IEEE
Computer, 24(4), 59-69.
Grudin, J. (1991b). Systematic Sources of
Suboptimal Interface Design in Large Product
Development Organizations. Human-Computer
Interaction, 6(2), 47-196.
Hackos, J., & Redish, J. (1998). User and Task
Analysis for Interface Design. New York, NY: John
Wiley and Sons.
Han, R., Bhagwat, P., LaMaire, R., Mummert, T.,
Perret, V., & Rubas, J. (1998). Dynamic Adaptation
in an Image Transcoding Proxy for Mobile WWW
Browsing. IEEE Personal Communications, 5(6),
8-17.
Hansmann, U., Merk, L., Nicklous, M. S., &
Stober, T. (2003). Pervasive Computing: The
Mobile World (Second ed.). Berlin, Heidelberg:
Springer-Verlag.
Held, A., Buchholz, S., & Schill, A. (2002).
Modeling of Context Information for Pervasive
Computing Applications. Proceedings of the 6th
World Multiconference on Systemics, Cybernetics
and Informatics (SC12002), Orlando, FL, USA,
July, 14-18.
Hevner, A. R., March, S. T., Park, J., & Ram, S.
(2004). Design Science in Information Systems
Research. MIS Quarterly, 28(1), 75-105.
Jacobs, A. R., & Abowd, G. D. (2003). A
Framework for Comparing Perspectives on Privacy
and Pervasive Technologies. IEEE Pervasive
Computing, 2(3), 78-84.
Kasper, G. M. (1996). A theory of Decision
Support System design for user calibration.
Information Systems Research, 7(2), 215-232.
Kourouthanassis, P. (2004). Can technology make
shopping fun? ECR Journal, 3(2), 37-44.
Kourouthanassis, P. (2006). A Design Theory for
Pervasive Information Systems. Unpublished
Dissertation, Athens University of Economics and
Business, Athens, Greece.
Kourouthanassis, P., & Roussos, G. (2003).
Developing Consumer-Friendly Pervasive Retail
Systems. IEEE Pervasive Computing, 2(2), 32-39.
Langheinrich, M. (2001). Privacy by Design –
Principles of Privacy-Aware Ubiquitous Systems.
Proceedings of the Ubicomp 2001, Berlin,
Heidelberg, 273-291.
11
Lei, Z., & Georganas, N. D. (2001). Context-based
Media Adaptation in Pervasive Computing.
Proceedings of the IEEE Canadian Conference on
Electrical and Computer (CCECE'01), Toronto,
Canada, May.
Limayem, M. (1996). A design methodology for
embedding decision guidance into GDSS. Group
Decision and Negotiation, 5(2), 143-164.
Lynch, T., & Gregor, S. (2004). User participation
in decision support systems development:
Influencing system outcomes. European Journal of
Information Systems, 13(4), 286-301.
Lyytinen, K., & Yoo, Y. (2002). Issues and
Challenges in Ubiquitous Computing.
Communications of the ACM, 45(12), 63-65.
Lyytinen, K., Yoo, Y., Varshney, U., Ackerman, M.
S., Davis, G., Avital, M., et al. (2004). Surfing the
Next Wave: Design and Implementation Challenges
of Ubiquitous Computing Environments.
Communications of the Association for Information
Systems, 13, 697-716.
Markus, M. L., Majchrzak, A., & Gasser, L. (2002).
A Design Theory for Systems that Support
Emergent Knowledge Processes. MIS Quarterly,
26(3), 179-212.
Mathiassen, L., Munk-Madsen, A., Nielsen, P. A.,
& Sage, J. (2000). Object Oriented Analysis and
Design. Aalborg, Denmark: Marko Publishing.
Mohan, R., Smith, J., & Li, C. S. (1999). Adapting
Multimedia Internet Content For Universal Access.
IEEE Transactions on Multimedia, 104-114.
Mumford, E., & Weir, M. (1979). Computer
Systems in Work Design - The ETHICS Method.
London: Associated Business Press.
Nielsen, J. (1994). Usability Engineering:
Academic Press.
Norman, D. A. (1999). The Invisible Computer:
Why Good Products Can Fail, the Personal
Computer Is So Complex, and Information
Appliances Are the Solution: MIT Press.
Palen, L., & Dourish, P. (2003). Unpacking
“Privacy” for a Networked World. CHI Letters, 5,
129-136.
Poltrock, S. E., & Grudin, J. (1994). Organizational
Obstacles to Interface Design and Development:
Two Participant Observer Studies. IEEE
Transactions on Human-Computer Interaction,
1(1), 54-80.
Ponnekanti, S. R., Lee, B., Fox, A., Hanrahan, P., &
Winograd, T. (2001). I-Crafter: A Service
Framework for Ubiquitous Computing
Evironments. Proceedings of the UBICOMP2001,
Georgia, Atlanta, September-October, 56-75.
Preece, J., Rogers, Y., & Sharp, H. (2002).
Interaction Design: Beyond Human-Computer
Interaction. New York: Wiley and Sons.
Redstrom, J. (2001). Designing Everyday
Computational Things. Unpublished Doctoral
Dissertation, Goteborg University, Goteborg,
Sweden.
Roussos, G., Spinellis, D., Kourouthanassis, P.,
Gryazin, E., Pryzbliski, M., Kalpogiannis, G., et al.
(2003). Systems Architecture for Pervasive Retail.
Proceedings of the ACM SAC Track on E-
Commerce, 350-356.
Saha, D., & Mukherjee, A. (2003). Pervasive
Computing: A Paradigm for the 21st Century. IEEE
Computer, March, 25-31.
Satyanarayanan, M. (2001). Pervasive Computing:
Visions and Challenges. IEEE Personal
Communications, 10-17.
Schur, A., Decker, S. D., & May, R. (1999). Design
issues for next generation interfaces (NGI).
Proceedings of the Conference on Human Factors
in Computing Systems (CHI'99), Pittsburgh,
Pennsylvania, 130-131.
Simon, H. A. (1996). The Sciences of the Artificial
(3rd Edition ed.): MIT Press.
Stein, E. W., & Zwass, V. (1995). Actualising
organisational memory with Information Systems.
Information Systems Research, 6(2), 85-117.
Story, M. F., Mueller, J. L., & Mace, R. L. (1998).
The Universal Design File: Designing for People of
All Ages and Abilities: The Center for Universal
Design, NC State University.
Walls, J. G., Widmeyer, G. R., & El Sawy, O. A.
(1992). Building an Information System design
theory for vigilant Executive Information Systems.
Information Systems Research, 3(1), 36-59.
Walls, J. G., Widmeyer, G. R., & El Sawy, O. A.
(2004). Assessing Information System Design
Theory In Perspective: How Useful Was Our 1992
Initial Rendition? Journal of Information
Technology Theory and Application, 6(2), 43-58.
12
Weaver, P. L., Lambrou, N., & Walkley, N. (1998).
Practical SSADM+. London: Pitman.
Weiser, M. (1993). Some Computer Science Issues
in Ubiquitous Computing. Communications of the
ACM, 36(7), 75-84.
Weiser, M. (2002). The Computer of the 21st
Century. IEEE Pervasive Computing, 1(1), 19-25.
13
