Design Principle for Ubiquitous Computing - unifr.ch · Design Principle for Ubiquitous Computing Chen Kai DIUF – Department of Informatics University of Fribourg, Switzerland [email protected]

Design Principle for Ubiquitous Computing

Chen Kai

DIUF – Department of InformaticsUniversity of Fribourg, Switzerland

[email protected]

Abstract

This paper intends to make a state of the art of designprinciples applicable in the ubiquitous computing (ubi-comp) domain. Since design principles are well defined,accepted and used in Human-Computer interaction, it isnot the case in ubicomp. We also analyze a concrete ubiq-uitous system project, establish a list of applied designprinciples and check if design principles for GUIs fit tothis particular system.

Keywords: Design principle, Ubiquitous computing,Pervasive computing.

1. Introduction

Since 30 years, the development of computing technol-ogy has changed drastically our daily life, and the inter-action between human and computer has evolved accord-ingly. If at the beginning, users were operating the ma-chine only (computer centered design), nowadays theyare the center of the problem and computer programsor systems are developed in regards to user’s needs andhabits (user center design). The user being in the cen-ter, program and system designers have to think abouthow users interact with computers and what kind of de-sign principles should be applied in order to design usablesystems.

In this paper, we intend to specifically present somedesign principles for ubiquitous computing (ubicomp)presented in divers papers. In section 2, we present prin-ciples of design every day objects proposed by D. A. Nor-man [2]. In section 3, we show how these principles areapplied in design the graphic user interface. In section4, we list some design principles for ubiquitous comput-ing system found in the literatures. In section 5, we showa example of the ubiquitous computing system and thedesign principles which has been applied, and find outwhich of the design principles considered in GUI can beapplied to this system. In section 6, we draw a conclusionabout design principles of ubicomp and future work.

2. Design principlesDesign principles are derived from a mix of theory-basedknowledge, experience, the common sense. They tendto be written in a prescriptive manner, suggesting to de-signers what to provide and what to avoid in the interface(does and don’ts of interaction design). More specifically,they intend to help designers explain and improve theirdesigns [5]. D. A. Norman [2] has identified the designprinciple which should be consider before creating anyobjects with which we will interact in our life.

In his paper [3], D. Gelbcite has made an overview ofthe main principles defined by Norman:

• Visibility. To be able to control the object, we haveto see their important components. For example,when we drive a car we should be able to see thestate of device and possible action. The car controlshould be place in a obvious position, so we caneasily found and use them. As a bad example, wetake the auto faucet. It is difficult to use, becausepeople like me will always try to find the switch todeactivate it.

• Feedback. It’s important to know what the systemis doing now. The feedback should be immedi-ate and synchronized with the user action. For in-stance, when playing piano, the pianist be able tohear the sound when he presses the key.

• Affordance. Perceiving the properties of a objectshould give the means of its operation. For exam-ple, the different shape of the door handle shouldtell us to push or pull it to open the door.

• Mapping. Mapping means the relationship be-tween controls and their action. For example, thecontrol of MP3 player has the buttons with direc-tion triangles. Just by seeing the direction of trian-gle, we know how to jump to the next song or goback to the previous one.

• Constraint. Restricting some interactions can re-duce the chance of errors. For example, in the

menu list some options are unavailable in somemoment context.

• Consistency. The interfaces which have similar op-erations should be consistent. For example, thepower on buttons are similar in different devicesor the access to menu are always the same.

In next section, we introduce how these principles areapplied in graphic user interfaces.

3. Design principle for graphical userinterface

In the previous section, we have seen the main designprinciples for everyday objects. Now we present howthese principles are applied specifically in the graphicaluser interfaces (GUI).

To make the program more user friendly, user inter-face must be intuitive and easy to use. In the paper ”Prin-cipals of Good GUI Design”, J. Hobart [4] has shown thecauses that contribute to a bad design:

• Forgetting user

• Failing to Give user control

• Providing to many feature at the top level

In [5] (ch. 1.6), they identify Norman’s design prin-ciples applicable specifically to GUI:

• Visibility. Intuitive position of control componentslike buttons and switches in the GUI can help userto use the program more easily instead of guessinghow to make it work.

• Feedback. To be able to know what has been doneand how to continue, the user need a feedback.For instance, during a backup operation, the userwould appreciate to know the remaining time in or-der to organize his time. J. Hobart [4] states that ingeneral, most users like to have a message dialogbox with a progress indicator displayed when op-erations are going to take longer than seven to tenseconds.

• Constraints. Constraints means to restrict certainkinds of user interaction. The advantage is that itreduce the chance of making mistakes. A commondesign practice in graphical user interfaces is to de-activate certain menu options.

• Consistency. J. Hobart shows a good example ofconsistency in software development. Suppose weare writing the business application, we should pro-vide the user with consistent behaviors. The busi-ness users of your software always have the similarneeds. In GUI, the benefits of consistent interface

is that they are easier to learn and use. Users haveto learn only a single mode of operation that is ap-plicable to all objects.

• Affordance. It is a term used to refer to an attributeof an object that allows people to know how touse it [5]. For example the graphical elements likeicons should be designed to afford clicking, scroll-bars to afford moving up and down, buttons to af-ford pushing.

• Mapping. J. Hobart calls it Provide TraceablePaths which means users should be able to knowhow to go to a window, how to achieve some func-tions.

• Simplicity. Keep the important and interesting el-ements in the screenful space. Getting the bal-ance between aesthetic appeal and simplicity[4].Google is a good example, the interface containsonly the necessary elements (e.g. the text field andbutton) to achieve the task.

In the next section, we present some design principlesidentified for ubicomp systems.

4. Design Principle for Ubiquitouscomputing

The concept of ubicomp has been developed in the 80’by Marc Weiser and stated that computers will be every-where, disappear into our environment and will be em-bedded into everyday objects [1].

In his vision, Weiser predicted a radical change ofparadigm in computer sciences passing from ”many users- one computer” to ”one user - many computers”. Alsoubicomp will allow to access information every time andeverywhere.

This new paradigm force us to redefine the interactionmode with such systems and reconsider the design prin-ciples that should be applied in order to create usable,intuitive, secure computing systems.

We have found several papers describing design prin-ciples. It appears that there are general principles whichcover everything in every domains as it is the case forGUI for instance. In the field of ubicomp, there are moredifferent approaches which not necessary overlap. Wehave selected only the paper describing what we considerto be relevant principles.

4.1. Computer system issues

In the paper ”Pervasive Computing: Vision and Chal-lenges” [8] M. Satyanarayanan et al. show the de-sign principles in computer issues. At the beginningthe authors present the evaluation of computer system.The ubiquitous computing environment is saturated withcomputing and communication capability. The difference

between ubiquitous systems and other computing sys-tems is that they become integrally part of people’s life,and is ”invisible”. Ubiquitous systems contain the princi-ple of distributed system and mobile computing system.They include the problematic of:

• Remote communication

• Distributed security

• Mobile networking

• Location sensitivity

Besides those concerns, ubicomp contains the addi-tional principles:

4.1.1. Effective use of Smart Spaces

The smart space could be an enclosed area, for exampleclassroom, meeting room. Smartness means the use ofthe ubiquitous computing technology. The system in thespace can adapt user’s requirements or provide specificservices to users depending on user’s information, for ex-ample activities or movements.

4.1.2. Invisibility

As mentioned by Weiser, the most profound technologyare the one which disappear. The invisibility is the mainfeature of ubiquitous system. The disappearance of thetechnology decreases the distraction. If the system al-ways meets the user’s requirement then it will be inte-grated into people’s life.

4.1.3. Localized Scalability

Since the interaction between the surroundings anduser’s computer space increase, the density changes andstrongly depends on the number of users. The nowadayscomputing systems (e.g. web server, email server) ignorethe distance from user to the server, whether they are lo-cated in the same room or across countries, they get thesame services. But a ubicomp system is location sensi-tive, in other words the location of the user is importantand changes the behaviour of the system.

4.2. Design principles for educational ubiquitouscomputing system

The goal of ubicomp is to help people in the daily tasks.In an ubicomp environment, users are in the center. Theidea is to put computers into the background. M. Eisen-berg et al. [7] argue that, as a design principle, ”invisibil-ity” has advantages in some domains; but that it has pow-erful, and ultimately counterproductive, connotations foreducational design.

They figure out that at the office work, professionalbusiness benefit of the ”invisible” technology. In the of-fice work context, the goal of the ubiquitous computing

system is to fit user’s requirement, therefore the workerdoes not have to know what happen in the background.So they can focus on their work and increase the effi-ciency. But in educational context, the goal is to interestthe students. If the technology is completely invisible,then people don’t know how it works and that will de-crease their curiosity. That is why they present three cen-tral themes for ubiquitous education computing.

4.2.1. Curiosity Enhancement

”Ubiquitous Computing as a Means of ConveyingIdeas”. To motivate the students, we need to increase thecuriosity of the objects with which they interact. For that,people should be able to design, program, customize theobjects. The authors show their lab to show how to argu-ment the curiosity. Their project is called ”SmartTiles”.SmartTiles are tiles-like pieces which contains a micro-controller, LED, and piezoelectric touch sensor. The tilesmay be placed in a background fabric that supplies com-munication connections (between neighboring tiles) anda source of power. Each tiles are programmable and theycan be combined into arrays to communicate with eachothers. As shown in Figure 1, the independently pro-grammed SmartTitles can be combined together to sim-ulate the famous ”Game of life”. This shows the ap-proach of ”curiosity enhancement” for educational ubiq-uitous computing.

Figure 1: SmartTiles array running the ”Game of Life”cellular automaton simulation.

4.2.2. Control and Programmability

”Turning Ubiquitous Computing Artifacts into Means ofExpression”. Instead of doing the ”screen based” pro-gramming, the educational ubiquitous computing enablesstudents to program the behaviors of the object in thereal world. ”a bicycle might be equipped with a pro-grammable display whose operation is influenced by theactions of the rider; a set of wind chimes could be aug-mented with programs that cause the chimes to produce

distinct sounds in response to distinct patterns of wind.”In [7], they also describe projects such as: LED bracelet(fig. 2) and tanktop shirt (fig. 3), they are ”programmablewearable” artifact objects. The user can write very sim-ple code and sent to the object to produce variety of visualeffect.

Figure 2: Programmable bracelet with an array of LEDsequence as a display.

Figure 3: Programmable tanktop shirt. Here, the weareris ”reprogramming” the shirt wirelessly with a new start-ing cellular automaton pattern.

4.2.3. Aesthetics

”Making Beautiful Ubiquitous Computational Artifacts”.The authors suggest that the ubiquitous computing envi-ronment should be beautiful and comfortable, increasingthe motivation and creativity when working in this space.

4.3. Transaction supporting

A transaction is an agreement between the buyer andseller for an exchange of items of value, such as goods,services and money. Transactions are the most commonway of interaction among people. (e.g. baggage han-dling, check out system).

In computer science, the most prevalent daily humanand computer interaction is the retail transaction. Wefound they can be benefit from the ubiquitous computing.A simple example of retail transaction settings is the ”au-tomated checkout”. The idea is that customers enter thestore, choose the products they want and leave the storewithout waiting in the queue to achieve the payment, theamount is translated from their credit card to the accountof the store automatically.

T. Salvador et al.[10] suggest in their paper that whenwe apply ubiquitous computing in transaction system wehave to consider the following principles: accountability,real-time inspectability, exercise of resource.

In this paper, the authors present the scenario of aRFID check-out system. The advantage of using RFIDis that the customers do not need to wait in the line topay their items. By checking the RFID tags on the itemsand the user’s credit card the ubiquitous check out systemachieves the transactions. On the other hand, they findsome problems. When users want to use another creditcard, they don’t want the system to access their accountto get some information. Also the cashiers need to usu-ally check the customer’s basket in order to ensure allarticles are paid and the payment is done with a valid andauthentic credit card (right owner).

With this scenario, they summarize following theprinciples of design in the context of transaction:

4.3.1. Accountability

Ubicomp systems will hide their actions of the transac-tions (i.e. ubicomp must be accountable). In the ubiqui-tous check out system, the transaction should be trustedby the client. In other words, the payment must beachieved correctly.

4.3.2. Real-Time Inspectability

It means in the context of check-out that at all time cus-tomers as well as the cashiers should be able to see”what’s going on”. The customers want to ensure theitems are payed by the right credit card, and cashier haveto ensure the customer did not forget items in the basketor cart and the credit card belongs to the customer.

4.3.3. Exercise of Recourse

The last element to support transactions is that both par-ties must have access to recourse. Before leaving thestore, the customers can review their receipt and make

sure that everything is in order. If mistakes happen dur-ing the transaction, the customers must be able to ask forcorrection.

4.4. Privacy awareness

The protection of personal information is another impor-tant issue in the principle of designing the ubicomp sys-tem. Our future world will be full of smart and cooper-ative ubiquitous computing devices. People will interactwith them every time and every where. These systemswill get some information from your: location, move-ment, intent or even identification.

M. Langheinrich[11] believes that there are two so-lutions to ensure that not everyone sees everything orto find a technology that guarantee, our personal in-formation could stay private by applying encryptionand anonymization. He also figures out that even ifanonymization technology can make tracing personaldigit identity impossible and encryption scheme seems tobe hard to break, we must realize that our presence in thereal world can not be completely hidden nor anonymized.In his paper, he proposes the privacy awareness system(pawS) which aims to find a solution balancing those twoapproaches.

The pawS focuses on ubiquitous computing environ-ments that allow data collectors to both announce andimplement data usage policies, as well as providing datasubjects with technical means to keep track of their per-sonal information as it is stored, used, and possibly re-moved from the system.

The Figure 4 shows a simple scenario which imple-ments pawS concepts:

• When the user enters the building, a privacy bea-con lists the data collection of each services

• A Privacy assistant on the user mobile connectswith a privacy proxy

• After comparing these privacy policies with theuser’s privacy policy configurations, the user’s mo-bile device disables some services that building of-fers.

To be able to create this system, the author lists themain principles:

• Notice. The data collections in the ubiquitous com-puting system should declare their privacy policies,and there should be efficiently communicated tothe user.

• Choice and consent. There should be a selectionmechanism to help users to indicate their preferredservice by comparing their privacy policy to theprivacy policies of the services.

Figure 4: Scenario implementing pawS

• Proximity and locality. By using the informationof location and proximity, it can restrict the user toaccess some data collections.

• Access and recourse. There should be an inter-face for the user to access their personal informa-tion. The usage and the location of their informa-tion should be noticed to the users.

5. The design principles applied inubiComp in a concrete use case

In this section, we present an ubicomp system and its de-sign principles and try to define if the design principlesfor GUI can also be applied in this project.

Project Aura [13] is an ubiquitous computing envi-ronment at Carnegie Mellon university. Instead of ex-plaining the Aura system itself, we choose the two hypo-thetical scenarios which are described in the [13].

• Scenario 1 :

Jane is at Gate 23 in the Pittsburgh airport, wait-ing for her connecting flight. She would like to sendthe documents by email by using wireless. Unfor-tunately, bandwidth is miserable. Aura observesthat at the current bandwidth Jane won’t be ableto finish sending her documents before her flightdeparts. Consulting the airports network weatherservice and flight schedule service, Aura discov-ers that wireless bandwidth is excellent at Gate 15.A dialog box pops up on Jane’s screen suggestingthat she go to Gate 15. Jane accepts Aura adviceand walks to Gate 15. She watches CNN on the TVthere until Aura informs her that it is close to be-ing done with her messages, and that she can startwalking back. The last message is transmitted dur-ing her walk, and she is back at Gate 23 in time forher boarding call.

• Scenario 2 :

Fred is in his office preparing for a meeting atwhich he will give a presentation and a softwaredemonstration. It is time to leave, but Fred is notquite ready. Aura transfers the state of his workfrom his desktop to his handheld, and allows himto make his final edits using voice commands dur-ing his walk. Aura infers where Fred is going fromhis calendar and the campus location tracking ser-vice. It downloads the presentation and the demon-stration software to the projection computer, andwarms up the projector. Fred finishes his edits justbefore he enters the meeting room. As he walks in,Aura transfers his final changes to the projectioncomputer. As the presentation proceeds, Fred isabout to display a slide with highly sensitive bud-get information. Aura senses that this might be amistake: the room face detection and recognitioncapability indicates that there are some unfamiliarfaces present. It therefore warns Fred. Realizingthat Aura is right, Fred skips the slide. He moveson to other topics and ends on a high note, leavingthe audience impressed by his polished presenta-tion.

From these two scenarios, we can identify several de-sign principles for ubicomp.

• Proactivity. Which is the most important designprinciple of ubicomp. The transitional computingsystem we are using are poor user’s intention detec-tion. Therefore they offer little adaption and proac-tivity [8]. Contrarily, Aura serves people by captur-ing their intent and giving suggestion to users. Inthe first scenario, Aura discovers that Jane want tosend documents by wireless, but since there are toomany peoples using the wireless at the same gate,Jane will not succeed in sending these documentsbefore depart. Aura looks for a more free gate, cal-culates the time for getting to there and comparewith the time of sending the document. Then Auragives Jane an advise to achieve her task.

• Privacy and Trust. Ubicomp must obtain the infor-mation about the users. Some information will besensitive for someone. In the second scenario, theprojector is turn on before Fred comes to the con-ference room. The user should first be authenticateto have the right to achieve that, and also duringthe presentation, when Fred go through the slidewhich contains sensitive data. Aura detects suchthat there are untrusted person in this room, so itsuggest Fred to skip some slide.

• Context awareness. This is another feature of ubi-comp. The system changes its comportment ofuser’s state and surrounding. In the first scenario,

Aura finds that the bandwidth of the current loca-tion is poor such that Jane will not be able to finishsending her email. Therefore, Aura suggests Janeto go to another gate (of course there should beenough time to achieve the task before the flight).

• High-level Energy Management. Since the numberof mobile devices around us increases and also theybecome more proactive and context-aware, there isa continuous increase of energy demand. The au-thors propose solutions such as the power-awarememory management [14] and the Energy-awareadaptation [15].

As we can see, the design principles applied in thetwo scenarios are specific to ubicomp. We try toapply more general design principles as defined byNorman and try to determine if human-ubicompsystem interaction can be design as GUI’s are.

– Visibility. To decrease the distraction, Auramust be invisible. Its operations are in thebackground. It changes its comportmentwhen user’s context changed, and it gives theuser advice when it is necessary.

– Affordance. Aura is an autonomous system.Users do not have to operate it, therefore ”af-fordance” is not important for Aura.

– Consistency. Although the users do not haveto operate Aura, there are at least some com-ponents in its user interface. For example,when Aura pops up the message box, the con-firm and cancel buttons must be the same asin other applications.

– Feedback. Aura has to give users intuitivefeedback. When every tasks are done, Auramust inform the user by popping up a mes-sage box, and the user can also check theprogress of the task.

We see that in this case the interaction with thesystem can follow the four main principles used inGUI design. Only the affordance is not importanthere.

6. ConclusionIn this paper, we have looked at design principles thatare applicable to ubicomp systems. To do this we havefirst defined the concept of design principles and their ap-plications in different domains such as every day objectdesign and GUIs. We have found that the variety of ubi-comp systems in different domains make difficult to de-fine general design principles as it is the case in HCI andGUI. However, we have identified some that appear to beimportant in the design phase of an ubicomp system. We

also found that even if specific design principles must beapplied when creating an ubicomp system, some of theadopted ones in HCI can be taken into consideration es-pecially when the user has to explicitly interact with thesystem. As conclusion, we can say that is an open fieldof research for now although different attempts have beendone already. We miss a proper classification and gener-alization of ubicomp design principles like it is the casein HCI domain.

7. AcknowledgementI would like to thank my supervisor, Pascal Bruegger. Hegives me a lot of ideas about how to writing a scientificpaper, and helps me to correct my English writing mis-takes very patiently. Without his support, I would nothave been able to terminate this seminar.

8. References[1] M. Weiser. The Computer for the 21st Century. Sci-

entific American, Inc, 1991.

[2] D. A. Norman. The Design of Everyday Things. TheMIT Press, forth edition, 2001.

[3] D. Gelb. An overview of Don Norman’s design prin-ciples. 2007.

[4] J. Hobart. Principal of Good GUI Design. 2007.

[5] H. Sharp, Y. Rogers, J. Preece. Interaction designbeyond human-computer interaction. Wiley, 2ndEdition, 2006

[6] S. Erikson. Designing for Accountability. Depart-ment of Human Work Science and Media Technol-ogy Blekinge Institute of Technology, 2002.

[7] M. Eisenberg, A. Eisenberg, L. Buechley, and N.Elumeze. Invisibility Considered Harmful: Revisit-ing Traditional Principles of Ubiquitous Computingin the Context of Education. Department of Com-puter Science, University of Colorado, 2006.

[8] M. Satyanarayanan. Pervasive Computing: Vi-sion and Challenges. School of Computer ScienceCarnegie Mellon University, 2001.

[9] D. Chalmers, M. Chalmers, J. Crowcroft, M.Kwiatkowska, R. Milner, E. O’Neill, T. Rodden, V.Sassone, M. Sloman. Ubiquitous Computing: Ex-perience, Design and Science. 2006.

[10] T. Salvador, S. Barile, J. Sherry. UbiquitousComputing Design Principles: Supporting Human-Human and Human-Computer Transactions. 2004.

[11] M. Langheinrich. A Privacy Awareness System forUbiquitous Computing Environments. Institute ofInformation Systems, ETH Zurich.

[12] A. Boedcher, K. Mukasa, D. Zuehlke. CapturingCommon and Variable Design Aspects for Ubiqui-tous Computing with MB-UID. 2005.

[13] D. Garlan, Daniel P. Siewiorek, A. Smailagic, P.Steenkiste Project Aura: Toward Distraction-FreePervasive Computing. Carnegie Mellon University,2002.

[14] A. R. Lebeck, X. Fan, H. Zeng, C. Ellis PowerAware Page Allocation. Department of ComputerScience Duke University, 2002.

[15] J. Flinn and M. Satyanarayanan Energy-awareadaptation for mobile applications. 17th ACMSymposium on Operating Systems Principles (SOSP ’99), Published as Operating Systems Review,34(5):48C63, Dec. 1999.