Eindhoven University of Technology

MASTER

Beyond reality the effect of a virtual reality experience on the user acceptance of smart homes

Honselaar, P.; Scharnigg, S.M.A.K.

Award date: 2015

Link to publication

Disclaimer This document contains a student thesis (bachelor's or master's), as authored by a student at Eindhoven University of Technology. Student theses are made available in the TU/e repository upon obtaining the required degree. The grade received is not published on the document as presented in the repository. The required complexity or quality of research of student theses may vary by program, and the required minimum study period may vary in duration.

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain Beyond Reality The effect of a Virtual Reality experience on the User Acceptance of Smart Homes

Master Thesis

Colophon

Master thesis Real Estate Management & Development Master Architecture, Building and Planning Department of the Built Environment Eindhoven University of Technology

Authors P. (Patrick) Honselaar 0752989 S.M.A.K. (Sanne) Scharnigg 0726216

September, 2015

Supervisory committee dr.ir. B. de Vries dr.ir. P.E.W. van den Berg ir. A.J. Jessurun

BEYOND REALITY THE EFFECT OF A VIRTUAL REALITY EXPERIENCE ON THE USER ACCEPTANCE OF SMART HOMES

Preface In front of you lies our research report about the effect of a Virtual Reality experience on the user acceptance of Smart Homes. Before starting our graduation project, we were searching for a challenging subject in the field of Smart Homes and ICT in the built environment. Guess we found it all in this thesis. In a rather short period of time we got the chance to set up a Virtual Reality Smart Home, in which we learned more about different modelling software and how to program. For real estate students to develop a model in Virtual Reality without any prior experience and knowledge was regarded surreal and did not seem possible at the start of our graduating process, but yet we succeeded. We are very grateful for this opportunity and the fact that we could perform this thesis together, which ensured that we could challenge ourselves and achieve so much more in this final university project.

The future‐oriented concept of Smart Homes has proven to be highly relevant for current society and it is hard to avoid the topic as newspapers entail related articles almost on a daily basis. We are privileged to contribute in this field of expertise through scientific research. The enthusiasm of the people around us definitely encouraged us to maintain high quality and to push ourselves to the limit. The fact that we could conduct our thesis at the department of Real Estate Management & Development in combination with the department of Design and Decision Support Systems did not only broaden our horizon scientifically, but also socially as we felt at home from day one in the research lab on the top floor of Vertigo.

We want to express our gratitude towards our supervisors Bauke de Vries, Pauline van den Berg and Joran Jessurun. Having the support of three professors with three different areas of expertise really helped us in finalising this thesis. Their kindness and dedication made us feel comfortable, appreciated and motivated us to cast a critical eye over our own work. A special thanks to the 134 respondents, as this thesis could not have been realized without all those who took part in our experiment. In advance, we could not have imagined that the experiment would become so successful and that so many people would participate in and talk about our thesis. However, the biggest thanks goes out to our family and friends, who have been a continuous support throughout our careers.

Martijn, Nancy, Paula & Bert, Arnout, I would like to thank my parents Carla Yara, Lenie & Bert, thanks for fitting a and Richard for the opportunity they supporting road trip in your busy provided me to study at a university schedules. Dad, mom, thank you both level. Their faith gave me the strength to for your advice, support and for keeping overcome all my obstacles. My brother me on my toes. A special thought goes Michel and sister Monique, who inspired out to my grandfather who passed away and supported me on every step of the during this project, but who would be so road. Of course, my beloved grandfather proud to know that I have become an whose enthusiasm will always remain ingenieur. Finally, I want to thank Patrick dear to me. Biggest heart goes out to for taking on this adventure with me and Sanne for being my bright and dedicated for always standing by my side. partner in crime on this smart journey.

Sanne Scharnigg Patrick Honselaar

September, 2015

Summary Smart Homes are presenting themselves as a new type of dwelling within the real estate market. A Smart Home is a home environment in which technology and space are integrated consequently resulting in an intelligent and comfortable way of life for its residents. Ever since its emergence, Smart Homes struggled to penetrate the real estate market and their potential has still been largely untapped. There are several reasons for this slow entrance, like costs, lack of awareness, privacy concerns, the limited interoperability of Smart Home technology, and the mismatch between the market and end‐user. All these barriers for the adoption of Smart Homes have a common ground: a low user acceptance of technologies used within Smart Homes. Therefore, the focus of the Smart Home discussion needs to be shifted from the technological to the social perspective. In order for Smart Home deployment to be widespread, the user acceptance of Smart Home technology needs to be addressed properly. In the upcoming decades, modern society faces many diverging challenges and Smart Homes might be capable of addressing some of these challenges, like aging or sustainability. Not only can Smart Homes support residents in their daily activities, they can also provide time and energy savings, convenience, and safety. Nowadays, most conducted research on Smart Home technology lays within the area of Smart Home devices, whereas research on the user acceptance is very scarce, especially within the field of the fully automated applications and (sensor) networks of devices. Creating user awareness through an acquaintance with these systems could highly benefit the implementation of Smart Homes.

The formulated research question of this thesis is twofold. At first it is necessary to identify the factors affecting the intention to use of Smart Homes in order to develop a model, before examining to what extent a user experience can enhance the user acceptance and contribute to the implementation of Smart Homes. Within this research, a user experience is provided through a Virtual Reality Smart Home experiment. In order to validate the impact of the experience, a pre‐ and post‐survey are included, which are derived from a newly developed user acceptance model for Smart Homes. The scientifically substantiated claim that results obtained in Virtual Reality (VR) strongly correspond to results in real‐life situations supports the fact that VR can be deployed to expose the potential relations of the scientific user acceptance model.

Since awareness and understanding are very important for user acceptance, it is vital to understand Smart Home and sensor technology. A sensor is a device that converts a physical phenomenon into an electrical or optical signal that can be quantified. Sensors can trigger actuators to start an application based on such a phenomenon, like motion, humidity or heat. There are several types of sensors and normally they do not operate as individual nodes in a Smart Home environment, but in a (wireless) network. In order to develop an accurate experiment, one holistic wireless sensor network protocol is selected to create a realistic representation of specific characteristics and features. Through research, it is identified that the wireless sensor network protocol MyriaNed is the most appropriate platform to conduct this study. MyriaNed is developed by Devlab, a company operating from within the Eindhoven University of Technology campus. Especially in its scalability, power consumption, reliability and security, being crucial characteristics for Smart Home appliances, MyriaNed distinguishes itself from other protocols.

Instead of creating a user acceptance model from scratch, it is useful to understand the most common models applied for technology acceptance, like TAM, UTAUT, UTAUT‐SE, and TAUM. Based on their knowledge and the Smart Home barriers, an own model was created in which perceived usefulness, perceived ease of use and facilitating conditions are the main factors affecting intention to use. The relationships between these variables can be influenced by certain personal characteristics, like gender, age, technological knowledge and social influence. As Smart Home technologies were not physically available and this study was not longitudinal, actual use could not be measured. Instead, the most important predictor of actual use, being intention to use, forms the dependent variable of the model. The survey questions necessary for the experiment were derived from the individual variables, resulting in two almost identical pre‐ and post‐surveys. Whereas personal characteristics only had to be asked once in the pre‐survey, questions involving the VR experience were included in the post‐survey.

The experiment was conducted at the Eindhoven University of Technology and was held for three weeks starting from the 10th until the 30th of June 2015. A Powerwall, 3D glasses and a wand were applied to create the Smart Home Virtual Reality tour. Through this, participants were immersed into the VR world and could move around freely. Several interactions and activities were included in the tour, in order for the participants to really experience the Smart Home and its technologies. Within the environment, a kitchen, living room, outdoor, bedroom, and bathroom activity were performed. In total, 134 participants individually performed the experiment. The results of the sample and descriptive statistics indicated an adequate reliability and normal distribution for all variables of the user acceptance model, except for the moderating effect technological knowledge. The survey questions measured diverging areas of the variable and therefore, technological knowledge had been split up into current IT use and the familiarity of people with Smart Homes. Analyses showed that more than 90% of the participants perceived that the VR experience helped in understanding what Smart Home technology could mean for them. More than 60% believed that the VR experience changed their attitude towards Smart Homes and influenced their opinion about implementing Smart Home technology. 79% of the respondents stated to have a better understanding of the term Smart Home and 75% believed that applying Smart Home technology could provide them energy conservations or other financial benefits. Considering VR as a tool, almost 16% of the participants felt some kind of uneasiness, either being unstable or nausea. This drawback needs to be addressed as this can prevent the implementation of VR on a larger scale.

The data analysis of the pre‐ and post‐survey indicated that the user acceptance model as developed needs modifications in order for the model to be a valid tool in Smart Home research. The proposed relationships between the variables were verified to be moderate or strong, in which ease of use was indicated as most important predictor of intention to use. In the linear regressions, the explanatory power of the pre‐ and post‐survey were confirmed substantial and show an R Square of respectively .464 and .493. This indicates that the data retrieved from the experiment adequately fits the statistical model. However, the moderating variables of the personal characteristics affected different relationships than predicted in advance. Being a male or female turned out to have no significant effect on any of the relations with intention to use, whereas age only showed a minor effect. The analysis indicated that older participants require more and better

facilitating conditions to increase their intention to use, as they are more dependent of the additional support. Social influence affects the same relationship, as well as the relation between perceived usefulness and intention to use. Those participants who are less influenced by their social environments, pay more attention to the content of the support network behind the Smart Home technology, and are therefore more affected by the facilitating conditions. Considering the technological knowledge, participants that lack knowhow on Smart Homes or possess multiple innovative technologies are more intended to use Smart Home technology if they perceive these technologies as easy to use. Furthermore, it was indicated that for those who use less innovative technologies, it is crucial to be aware of the usefulness of Smart Home technology. Besides age and social influence, familiarity with Smart Homes affects the relation between facilitating conditions and intention to use. By including these effects, the degree to which the data fitted the statistic model increased considerably, from an R Square of .464 to .601 in the pre‐survey and an explanatory power of .493 to .564 in the post‐survey model. All moderating effects of both the pre‐ and post‐survey were combined to form a new, validated user acceptance model. This model is shown in Figure 0.1. As the variable of intention to use showed an overall increase after the experiment, it is shown that the user acceptance of Smart Homes can be enhanced by a Virtual Reality user experience.

Requisites

Perceived Usefulness

Perceived Ease Intention to of Use use

Facilitating Conditions

Personal Characteristics

Technological Technological Age Knowledge – Knowledge – Social Influence Current IT Use Familiar with SH

Figure 0.1 The user acceptance model regarding Smart Homes and Smart Home technology

Virtual Reality might be the missing link for many companies or organizations in bringing their product on the market and to communicate with different parties. Vice versa, as the technology is perceivable and tangible, customers are more aware of what they are potentially buying. It is expected that VR and Smart Homes will play a far more significant role within society in the near future. It is assumed that the nauseous effect some respondents experienced in VR will diminish over time, as the human body is remarkably able to adapt to new circumstances and the game community is still improving the technology. The success of introducing Smart Home technology through a Virtual Reality home setting can become a catalyst for new areas of VR implementations. Specifically in combination with Smart Homes or other types of real estate, and in a wider perspective for the entire built environment.

Samenvatting Smart Homes presenteren zichzelf als een nieuw type woning binnen de vastgoedmarkt. Een Smart Home is een woonomgeving waarin technologie en ruimte consequent geïntegreerd zijn om zodoende een intelligente en comfortabele manier van leven te realiseren voor de bewoners. Sinds haar ontstaan heeft de Smart Home moeite gehad om de vastgoedmarkt te betreden en is haar potentieel grotendeels onbenut gebleven. Er zijn verscheidene redenen voor deze langzame toetreding, waaronder de kosten, het gebrek aan bewustzijn bij de klant, privacy zorgen, de beperkte interoperabiliteit van de Smart Home technologie en de mismatch tussen de markt en eindgebruiker. Al deze barrières voor de ingebruikname van Smart Homes hebben een gemeenschappelijk raakvlak: een lage gebruikersacceptatie van de technologieën binnen een Smart Home. De focus van de Smart Home discussie dient te verschuiven van het technologische naar het sociale perspectief. De gebruikersacceptatie van Smart Home technologie dient adequaat in kaart gebracht te worden om Smart Homes op grote schaal te kunnen inzetten. In de komende decennia wordt de moderne samenleving geconfronteerd met diverse uitdagingen. Smart Homes zijn mogelijk in staat een aantal van deze uitdagingen, zoals vergrijzing en duurzaamheid, aan te pakken. Naast het ondersteunen van bewoners in hun dagelijkse behoeften, kunnen Smart Homes ook gemak, veiligheid, energie‐ en tijdbesparingen opleveren. Tegenwoordig richten de meeste Smart Home technologie onderzoeken zich op de apparaten zelf, terwijl onderzoek naar de gebruikersacceptatie zeer schaars is, vooral op het gebied van volledig geautomatiseerde toepassingen en (sensor)netwerken van apparaten. Het creëren van gebruikersbewustwording door middel van een kennismaking met deze technologieën kan de implementatie van Smart Homes sterk bevorderen.

De onderzoeksvraag van deze scriptie is tweeledig. In eerste instantie is het noodzakelijk de factoren vast te stellen die de gebruiksintentie van Smart Homes beïnvloeden, om zodoende een model te ontwikkelen. Hierna kan er pas onderzocht worden in hoeverre een ervaring de gebruikersacceptatie kan verhogen en kan bijdragen aan een verbeterde implementatie van Smart Homes. Dit onderzoek verstrekt een gebruikerservaring met Smart Homes door middel van een experiment in Virtual Reality (VR). Om het effect van de ervaring te valideren, wordt het experiment begeleid door een voor‐ en na‐enquête. Deze enquêtes worden afgeleid uit het nieuw ontwikkelde gebruikersacceptatie model voor Smart Homes. De wetenschappelijk onderbouwde stelling dat VR resultaten sterk overeenkomen met de in werkelijkheid behaalde resultaten, ondersteunt de aanname dat potentiële relaties van het gebruikersacceptatie model bloot te leggen zijn in VR.

Aangezien bewustwording en begrip nauw verbonden zijn aan gebruikersacceptatie is het essentieel om Smart Home en sensor technologie te begrijpen. Een sensor is een apparaat dat een fysisch verschijnsel omzet in een kwantificeerbaar elektrisch of optisch signaal. Sensoren kunnen actoren activeren om een applicatie te starten op basis van een dergelijk verschijnsel, zoals beweging, vochtigheid of warmte. Er bestaan verschillende type sensoren die normaliter niet afzonderlijk opereren, maar in een (draadloos) netwerk binnen een Smart Home omgeving. Om een nauwkeurig experiment te realiseren is er één alomvattend, draadloos sensor netwerk protocol geselecteerd om een realistische weergave van de specifieke kernmerken en functies te geven. Onderzoek heeft uitgewezen dat MyriaNed hiervoor het meeste geschikte platform is. MyriaNed is

ontwikkeld door Devlab, een onderneming die gesitueerd is op de campus van de Technische Universiteit Eindhoven. Vooral in schaalbaarheid, energieverbruik, betrouwbaarheid en veiligheid, cruciaal voor Smart Home applicaties, onderscheidt MyriaNed zich van andere protocollen.

Input voor het nieuwe gebruikersacceptatie model is verkregen vanuit de Smart Home barrières en bekende technologie acceptatie modellen, zoals TAM, UTAUT, UTAUT‐SE en TAUM. Waargenomen bruikbaarheid, waargenomen gebruiksgemak en ondersteunende condities komen in het nieuwe model naar voren als de belangrijkste factoren die de intentie tot gebruik beïnvloeden. De relaties tussen deze variabelen worden beïnvloed door de persoonlijke kenmerken geslacht, leeftijd, technische kennis en sociale invloed. Aangezien dit onderzoek geen longitudinale studie betrof en de Smart Home technologieën niet fysiek beschikbaar waren, kon het daadwerkelijke gebruik niet gemeten worden. In plaats daarvan werd de belangrijkste voorspeller van daadwerkelijk gebruik, zijnde de intentie tot gebruik, opgenomen als afhankelijke variabele in het model. De vrijwel identieke voor‐ en na‐enquêtevragen zijn verkregen uit de voorgaande variabelen. De persoonlijke kenmerkvragen worden enkel gesteld in de voor‐enquête en de vragen aangaande de VR ervaring worden alleen opgenomen in de na‐enquête.

Het experiment werd van 10 tot 30 juni 2015 uitgevoerd aan de Technische Universiteit Eindhoven. Voor de Smart Home Virtual Reality tour werden een Powerwall, 3D bril en een handbediening aangewend. De deelnemers werden als het ware de VR wereld ingezogen en konden zich daar vrij bewegen. Verschillende interacties en activiteiten zijn geprogrammeerd om deelnemers de Smart Home en haar technologieën te laten ervaren. Er worden activiteiten uitgevoerd in de keuken, woonkamer, slaapkamer, badkamer en buiten. In totaal hebben 134 mensen deelgenomen aan het experiment. Uit de resultaten van de steekproef beschrijving blijkt dat alle variabelen in het gebruikersacceptatie model voldoende betrouwbaar en normaal verdeeld zijn, met uitzondering van het interactie‐ effect van technologische kennis. De enquêtevragen meten uiteenlopende aspecten en daarom dient de variabele te worden opgesplitst in huidig IT gebruik en bekendheid met Smart Homes. Analyses wijzen uit dat ruim 90% van de deelnemers ervoer dat de VR ervaring heeft geholpen om te begrijpen wat Smart Home technologie voor hen kan betekenen. Meer dan 60% is van mening dat de VR ervaring hun houding ten opzichte van Smart Homes veranderd heeft en tevens hun mening heeft beïnvloed aangaande het gebruik van Smart Home technologie. 79% van de respondenten geeft aan een beter beeld te hebben van de term Smart Home en 75% gelooft dat het toepassen van Smart Home technologie energiebesparingen en andere financiële voordelen kan opleveren. Aangaande VR als instrument voelde bijna 16% een soort van instabiliteit of misselijkheid. Deze keerzijde dient aangepakt te worden om VR op grotere schaal te kunnen inzetten.

Uit de data‐analyse bleek dat modelaanpassingen noodzakelijk waren om een geldig Smart Home gebruikersacceptatie model te verwerven. De vooropgestelde verbanden tussen de variabelen waren matig tot sterk, waarin gebruiksgemak werd aangeduid als belangrijkste voorspeller van intentie tot gebruik. De lineaire regressie toonde aan dat het verklarend vermogen van de voor‐ en na‐enquête substantieel waren met een R Square van respectievelijk .464 en .493. Dit betekent dat de experimenteel verkregen data overeenkomt met het statistisch model. Echter, de interactievariabelen van de persoonlijke kenmerken beïnvloedden andere relaties dan vooraf voorspeld. Geslacht

bleek geen invloed te hebben op de relaties met intentie tot gebruik, terwijl leeftijd slechts geringe invloed had. Meer en betere ondersteunende condities zijn vereist voor ouderen om hun intentie tot gebruik te verhogen, aangezien ze afhankelijker zijn van aanvullende ondersteuning. Sociale invloed beïnvloedt dezelfde relatie, evenals het verband tussen de waargenomen bruikbaarheid en de intentie tot gebruik. Deelnemers die minder beïnvloedbaar zijn besteden meer aandacht aan de inhoud van de ondersteunende condities achter de Smart Home technologie. Zij die meerdere innovatieve technologieën bezitten of waar de kennis over Smart Homes ontbreekt, zijn meer geneigd om Smart Home technologie te gebruiken als ze het gebruiksgemak inzien. Verder wees de analyse uit dat voor diegene die minder innovatieve technologieën gebruiken, het van belang is om bewust te zijn van de bruikbaarheid van Smart Home technologie. Naast dat leeftijd en sociale invloed de relatie tussen de ondersteunende condities en de intentie tot gebruik beïnvloeden, heeft ook de bekendheid met Smart Homes invloed op deze relatie. Door het gebruikersacceptatie model te verrijken met deze interactie‐effecten stijgt het verklarend vermogen van het model aanzienlijk, in het voor‐enquête model met een R Square van .464 tot .601 en in het na‐enquête model van .493 tot .564. Alle interactie‐effecten van beide enquêtes werden gecombineerd tot een nieuw, gevalideerd gebruikersacceptatie model afgebeeld in Figuur 0.2. De totale toename van de intentie tot gebruik na het experiment toont aan dat de gebruikersacceptatie van Smart Homes versterkt kan worden door een ervaring in VR. Voorwaarden

Waargenomen bruikbaarheid

Waargenomen Intentie tot gebruiksgemak gebruik

Ondersteunende condities

Persoonlijke kenmerken

Technologische Technologische Leeftijd kennis – kennis – Sociale invloed Huidig IT gebruik Bekendheid met SH

Figuur 0.2 Het gebruikersacceptatie model aangaande Smart Homes en Smart Home technologie Voor veel bedrijven en organisaties kan Virtual Reality de ontbrekende schakel zijn om hun producten op de markt te brengen en om te communiceren met verschillende partijen. Vice versa, als de technologie waarneembaar en tastbaar is, kunnen klanten meer bewust worden van hetgeen ze potentieel gaan kopen. De verwachting is dat VR en Smart Homes een veel grotere rol zullen gaan spelen in de maatschappij in de nabije toekomst. Er kan worden aangenomen dat de misselijkheid, die sommige respondenten ervoeren in VR, zal afnemen met de tijd, aangezien het menselijk lichaam in staat is zich geleidelijk aan te passen aan nieuwe omstandigheden en de game gemeenschap nog steeds de technologie aan het verbeteren is. Het succesvol introduceren van Smart Home technologie via een thuis setting in Virtual Reality kan een katalysator zijn voor nieuwe VR implementatie gebieden. Met name in combinatie met Smart Homes of andere typen vastgoed, maar ook in een breder perspectief voor de gehele bouwkunde.

Table of Contents Preface ...... Summary ...... Samenvatting ...... 1 Introduction ...... 1 1.1 Introduction into Smart Homes ...... 1 1.2 Research relevance ...... 3 1.2.1 Societal relevance ...... 3 1.2.2 Scientific relevance ...... 3 1.3 Demarcation ...... 4 1.4 From Virtual Reality to real life ...... 6 1.5 Goal and research question ...... 7 1.5.1 Goal ...... 7 1.5.2 Research question and sub‐questions ...... 7 1.6 Research design and readability ...... 8 1.7 Thesis outline ...... 10 2 Sensor technology: The essentials ...... 11 2.1 Introduction ...... 11 2.2 Sensor terminology: sensors, transducers and actuators ...... 12 2.3 Sensor characteristics ...... 14 2.4 Sensor types for Smart Homes ...... 16 2.5 Sensor networks and criteria ...... 17 2.6 Communication protocols within a WSN ...... 19 2.7 Smart Home communication protocols – a comparison ...... 21 2.8 MyriaNed ...... 22 2.9 Conclusion ...... 24 3 User acceptance model for Smart Homes ...... 26 3.1 Introduction ...... 26 3.2 Technology Acceptance Model ...... 28 3.2.1 TAM: Model and construct definition ...... 28 3.2.2 TAM: Statistical methods ...... 29 3.3 Unified Theory of Acceptance and Use of Technology ...... 30 3.3.1 UTAUT: Model and construct definition ...... 30 3.3.2 UTAUT: Statistical methods...... 31 3.4 Unified Theory of Acceptance and Use of Technology for Smart Environments32

3.4.1 UTAUT‐SE: Model and construct definition ...... 32 3.4.2 UTAUT‐SE: Statistical methods ...... 33 3.5 Telecare Acceptance and Use Model ...... 34 3.5.1 TAUM: Model and construct definitions ...... 34 3.5.2 TAUM: Statistical methods ...... 35 3.6 Conclusion ...... 35 4 From acceptance model to survey ...... 37 4.1 Introduction ...... 37 4.2 Defining and operationalizing the model ...... 38 4.3 Survey questions, measurement scales and sample size ...... 41 4.4 Conclusion ...... 44 5 The Virtual Reality Smart Home experience ...... 45 5.1 Introduction ...... 45 5.2 Technological implementation and activities ...... 47 5.3 Smart Home design ...... 50 5.4 Virtual Reality tour ...... 53 5.5 Interaction through programming ...... 57 5.6 Conclusion ...... 59 6 Data collection and sample description ...... 60 6.1 Introduction ...... 60 6.2 Data collection ...... 60 6.3 Sample and descriptive statistics ...... 62 6.4 Conclusion ...... 70 7 Data analysis ...... 71 7.1 Introduction ...... 71 7.2 Analysing the relationships ...... 71 7.3 Analysing the moderator effects ...... 72 7.4 Evaluation of the pre‐ and post‐model ...... 78 7.5 Conclusion ...... 79 8 Conclusions and recommendations ...... 81 8.1 Conclusions ...... 81 8.2 Discussion ...... 84 8.3 Recommendations and directions for future research ...... 85 9 References………………………………………………………………………………………………………...... 89

1 Introduction 1.1 Introduction into Smart Homes Smart Homes are presenting themselves as a new type of dwelling within the real estate market. When people refer to the holistic concept of Smart Homes, they might use other denominations, like home automation, home networking, sensor‐embedded houses or intelligent homes (DiCarlo, 2010). The term Smart Home is generally used to describe a residence that is equipped with technological features in order to improve the quality of life. However, the concept comprises more than that. Despite being a ubiquitous computing home environment, its main purpose is to respond to the residents’ needs and comfort. By integrating information technology in a home environment, a more intelligent way of life is created. Due to the automated appliance control, comfort, convenience, healthcare, security, entertainment, and energy conservation can be regulated according to the users’ desires and needs (Alam, et al., 2012; Aldrich, 2003; Allameh, et al., 2012).

The way in which people regard Smart Homes can differ significantly. Some authors stress that the focus of Smart Homes is within the topic of independent living and the aging population (Chan, et al., 2008; Chan, et al., 2009; Ehrenhard, et al., 2014; Okeyo, et al., 2014), while others believe its energy‐related capabilities are key (Capitanelli, et al., 2014; Zhang, et al., 2015; Balta‐Ozkan, et al., 2013). Most of the authors solely define Smart Homes by its technology (Belley, et al., 2014; Peine, 2009; Missaoui, et al., 2014; Sanaee, 2014; Alam, et al., 2012), whereas only few mention the user (Ding, et al., 2011; Balta‐ Ozkan, et al., 2014; De Silva, et al., 2012) or approach the concept from an end‐user’s perspective (Fang, et al., 2014; Allameh, et al., 2012; Aldrich, 2003). Considering the technology, Bierhoff et al. (2007) state that Smart Home technology is best defined as an integration of technology and services through home networking for an enhanced quality of life.

Regarding the previous, the following definition of a Smart Home is used within this thesis:

‘’A Smart Home is a home environment in which technology and space are integrated consequently resulting in an intelligent and comfortable way of life for its residents.’’

The greatest difference between this definition and the probably more generally available ones, is the user‐centred point of view. In general, people tend to define Smart Homes by its technology. It is vital for Smart Homes to raise general awareness for their beneficial features concerning near‐future problems. Furthermore, the definition points out that a Smart Home is not only about technology, the technology is just the instrument utilized to provide a new way of life for its users. Therefore, technology and space are mere means to achieve the higher goal. An intelligent way of life for its residents can result in an enhanced comfort for the end‐user (D'Ulizia, et al., 2010; Augusto & Nugent, 2006). The technology involved within Smart Homes has to become familiar to the public in

1 order for them to embrace the concept of Smart Homes (Davis, 1993; Balta‐Ozkan, et al., 2013).

In the Netherlands, the first impression of a Smart Home was presented back in the mid‐ eighties by television scientist Chriet Titulaer. Considering the timeframe, he was miles ahead of his generation when introducing his House of the Future, in which he came remarkably close to reality (Titulaer, 1985; NOS, 2014). Whereas the history of home automation seems rather short, home appliances had set in long before their presence. The 20th century entails a revolutionary transition of technology within the domestic environment. Domestic technology transformed beyond any recognition since electricity entered the homes. In that era, a domestic servant decline coincided with the introduction of electricity‐powered machines. The early home appliances of that period consisted of devices like the vacuum cleaner or sewing machine. Within decades, washing machines, cookers, hair dryers, electric razors, televisions, refrigerators, and thermostats were added to that list (Aldrich, 2003). The technology could provide conveniences to its users in their daily activities and thereby cause time savings, while other inventions were adopted to use within this newly created spare time.

Actual signs of home automation started to appear in the 1980s. Although efforts had already been made in private projects on a small scale, it was the National Association of Home Builders in the United States of America that first referred to a new design of homes as Smart Houses in 1984. The main focus of their working theory comprised the essence of including useful technology and home automation into a new type of dwelling. Although revolutionary at that time, it did not really cause a serious enhanced public interest in the Smart Home concept (Aldrich, 2003). Perhaps the advanced technology did not fit the zeitgeist of that period. It was not until the nineties that the concept started to gain popularity and by the beginning of the 21st century, the first Smart Homes arose and their technological devices started to enter the consumer market.

Ever since its emergence, Smart Homes struggled to penetrate the real estate market. Although Smart Homes have been presenting themselves to the public for quite some decades, their potential has still been largely untapped. There are several reasons for this slow entrance (Gann, et al., 1999; Balta‐Ozkan, et al., 2013; Ehrenhard, et al., 2014; Bernheim Brush, et al., 2011). Firstly, a Smart Home encompasses many different products and technologies, which all have different protocols that support different sets of Smart Home equipment. Due to the lack of standardisation and common protocol, interoperability has been limited, resulting in a low market adoption. Secondly, consumer investment is high in order to obtain a fully automated system. Potential end‐users need to be aware of what this information technology can provide them, before even considering buying it. In addition, the perceived usability and attitude towards the technology are critical for the actual usage behaviour (Davis, 1993). Thirdly, it is especially vital within the current economy that developers understand and tailor their products to the user needs. By doing so, they should wield a market‐pull strategy instead of a technology‐push strategy (Saidi, 2011). By listening to their customers, companies try to fix the mismatch between their available products and the customer needs.

Despite the technical issues, all these barriers for the adoption of Smart Homes are linked to a common ground: a low user acceptance of technologies used within Smart Homes. In order for the Smart Home deployment to be widespread, the user acceptance of Smart

2 Home technology needs to be addressed properly (Dengler, et al., 2007; Gaul & Ziefle, 2009). The Buildings Services Research and Information Association (BSRIA), a British market intelligence provider within the built environment, states that now is the time for Smart Homes to increase their market share by focussing on comfort and convenience for the end‐user. Smart Homes can benefit from the growth in the smartphone and tablet market, as well as from concepts like the Internet of Things (Honeywell, 2014). The Internet of Things refers to the current development in which various embedded devices become IP‐enabled. IP is the abbreviation for Internet Protocol, a communication protocol that enables networks to communicate with one another by transmitting data. Through this, a variety of objects and things like sensors, radio‐frequency identification (RFID) or mobile phones can be connected to each other and the Internet, creating an enormous integral network of devices (Atzori, et al., 2010; Shelby & Bormann, 2009). Therefore, especially within this timeframe, it is crucial for the Smart Home technology to be accepted by its potential users on a large scale. 1.2 Research relevance 1.2.1 Societal relevance In the upcoming decades, modern society faces many diverging challenges within the demographical, environmental, technological and social context. The concept of Smart Homes might be capable of addressing some of these challenges, like aging or sustainability. However, in order to address these aspects it is key for the concept to be widely accepted and implemented within society. Therefore, the focus of the Smart Home discussion needs to be shifted from the technical perspective to the social perspective (Davidoff, et al., 2006; Allameh, et al., 2013). Due to the wide variety of benefits, Smart Homes can be deployed within multiple settings for different purposes. Not only can Smart Homes support its residents in their daily activities, it can also provide time savings, efficiency, control, and safety (Robles & Kim, 2010). 1.2.2 Scientific relevance The possibilities with and implementation of Smart Homes are a common subject within the scientific literature. However, the user acceptance of Smart Home technologies and the actual use are underexposed in these conducted studies. Nevertheless, several articles do emphasise the importance of these issues (Bal, et al., 2011; Chan, et al., 2009; De Silva, et al., 2012; Sponselee, 2013; Allameh, et al., 2013; Sponselee, et al., 2008). Privacy and ethical issues are mentioned in various articles as current barriers and provide key challenges for the user acceptance (De Silva, et al., 2012; Chan, et al., 2009; Chan, et al., 2008; Ding, et al., 2011; Fabian & Feldhaus, 2014; Abi‐Char, et al., 2010; Bal, et al., 2011; Busnel & Giroux, 2010). Furthermore, Allameh et al. (2013) state the user acceptance of Smart Homes is hampered, due to the complexity of the building, since Smart Home technologies become more and more invisibly embedded in the home environment. User acceptance is taken into consideration in most articles, parallel to the research or as a subject for future research (Arcelus, et al., 2007; Gaddam, et al., 2009; Kim, et al., 2013; Chan, et al., 2008; Bal, et al., 2011). The articles that do study the user acceptance of Smart Homes only perform surveys or develop systems (Sponselee, et al., 2008; Sponselee, 2013). These developed systems focus on the aspects that increase the user acceptance level, like ease of maintenance, the way of surveillance and monitoring, and the information sharing with third parties (Abi‐Char, et al., 2010; Fabian & Feldhaus,

3 2014; Busnel & Giroux, 2010; Kim, et al., 2013; Gaddam, et al., 2009). Considering the literature, it is shown that there is a discrepancy between the theoretical and practical use. The research of the user acceptance and implementation of Smart Homes is very scarce. Exactly these two elements are necessary to make the step to actual use. Sponselee et al. (2008) study the effectiveness of Smart Home technology in home care situations and they state the difference between intentional use and actual use. Actual use is seen as the optimum acceptance. If technology supports the goals of well‐being, the technology is useful and consequently will be implemented more often. Intention to use is considered as the main determinant for actual use (Sponselee, et al., 2008). Sponselee et al. (2008) developed a model that uses aspects of the widely implemented Technology Acceptance Model (TAM), but also take into account the ‘behavioural‐ acceptance’ and the factors that determine the actual use. They state that the original TAM does not cover all areas necessary to predict the actual Smart Home technology user acceptance and use, as user reactions in other studies show that the acceptance of a new technology also depends on costs, information supply and experience (Sponselee, et al., 2008). Furthermore, experiencing Smart Home technology is researched by actually implementing technology in a home (Almende, 2011; Almende, 2010; DevLab, 2013; DevLab, 2011; Engineersonline, 2013; Gomez & Paradells, 2010) or by Virtual Reality (VR) programs (Allameh, et al., 2013; Heidari, et al., 2014; Lertlakkhanakul, et al., 2008). Those studies merely intended to test the technology or to elicit user preferences and how the user would react to the Smart Home technology, not to research the user acceptance of Smart Home technology. Furthermore, Virtual Reality is also utilized to measure user acceptance (Lai, et al., 2009), yet not within the subject of Smart Home technology. 1.3 Demarcation Smart Home technology comprises a wide range of networking devices, services and equipment. It is impossible to study the user acceptance of Smart Home technology in total. Therefore, it is necessary to demarcate the technology. DiCarlo (2010) distinguishes three main categories of Smart Home technology, being 1) simple, fixed applications, 2) applications programmable by users, and 3) applications that are flexible, programmable, and automated. However, defining Smart Home technology by these categories seems slightly outdated, since there is an important difference between for instance a Smart Wall and an automatic home security system for the windows. It would be more difficult to position those applications into DiCarlo’s categorisation. Consequently, the classification needs to be updated. Therefore, based on his study and recent developments, the following classification is formulated and used within this thesis:

1. Simple, straightforward applications with pre‐defined functions that have a fixed position in the house, like a video phone at the door or an intercom. 2. Applications and devices that can be set and programmed by its users within a certain range of options, like a television or microwave. 3. Fixed or flexible, semi‐automatic applications and networks of devices that can communicate with each other and can be manually set or programmed by its users, by using a smartphone, tablet or the device itself, like a Smart Kitchen, Smart Furniture or smart thermostat. 4. Fully automated applications and networks of devices, like automatic doors, daylight systems, monitoring devices or self‐adjusting blinds.

4 The suggested hierarchy within the categories also entails a certain timeline in which the first two technologies date back some decades ago, while the latter two are considerably new. Devices like the videophone or the microwave are already widespread within households and accepted by its users on a large scale (Caprani, et al., 2012). Therefore, it is unnecessary to conduct further research on the user acceptance within these two categories. Devices that can be controlled and monitored by using a smartphone, on the other hand, still struggle to enter the home environment. For instance, a current discussion in Dutch society is based on the services gained and information retrieved from the so‐called Slimme Meter, a smart digital gas and energy meter that can be controlled from within or outside a home by using a mobile phone. Main focus are the privacy concerns that come along with this technology, since the electricity consumption is recorded and communicated back to the energy supplier and grid operator. The meters might provide users insight into their own energy consumption during the day, but at the same time they are sharing that information with the public utilities (Heck, 2009). Despite the pros and cons, a widespread user acceptance is needed in order for the technology to succeed (Sponselee, et al., 2008). Due to the huge consternation, the Dutch government decided that the use of the Slimme Meter should be voluntary instead of mandatory (Ministerie van Economische Zaken, 2014). Whereas a smart meter is employed to get insight in the energy consumption at a home, a smart thermostat can be utilized to adjust the home temperature, even when the user is outside his or her home. The thermostat can be connected to a smartphone, through which various applications can be accessed (Lee, et al., 2013). Although barriers are already quite high, most of the time users of Smart Home technology still have the capability to somehow control an application or device, since they can adjust it by using their mobile device. In this case, technology is semi‐automatic when there is user intervention needed in order to set an activity in motion.

In this thesis, the focus will be on the fourth category of the Smart Home technology classification, the fully automatic applications. In the third category, the users still have some control and they need to keep their smartphone with them at all times. The smartphone industry still has a growing market share (Gartner, 2015). In 2013, the smartphone exceeded the annual sales of the feature phone, being the ‘normal’ cell phone, for the first time (Gartner, 2014). This indicates that smartphones and the systems are getting more and more accepted. A wireless sensor network, on the other hand, is a fully automatic application whose possibilities are relatively unknown and need more research in both the network integration and the user acceptance (Chan, et al., 2008). These systems provide advantages when it comes to privacy and maintenance (Alam, et al., 2012; DevLab, 2013). When using a smartphone, a connection is made outside the home environment, whereas when using a self‐regulated wireless sensor network, no information has to be sent or accessed outside the home. By creating user awareness through an acquaintance with these systems, the acceptance of wireless sensor network technology in Smart Homes could benefit (Lai, et al., 2009).

In the case of fully automatic applications, there is no device for users to control the system. Therefore, the feeling of control is out of their hands. For instance, there are smart thermostats that use motion sensors to detect whether residents are present (Lu, et al., 2010; Whitehouse, et al., 2012). These occupancy sensors are located within the ambient home environment. Jeng (2009) states that, since ubiquitous computing

5 technologies become smaller, hardware components can be embedded in a largely invisible way in the home environment and can be mutually connected through a wireless network. Through this, a more complex home environment with various systems is created. In addition, Allameh et al. (2013) emphasize that this home complexity contributes to a low user acceptance level of Smart Homes.

Considering the scientific literature, nowadays most research on Smart Home technology lays within the area of Smart Home devices, whereas these sensor networks are also main part of the Smart Home technology. Therefore, in this thesis the user acceptance of sensor technology in fully automated applications and networks of devices of Smart Homes is addressed. Like Smart Home technology, there are multiple sensor technologies available. Even outside of the Smart Home concept, sensor technology is used within multiple disciplines. Therefore, in order to examine the user acceptance of sensor technology in Smart Homes, the sensor technology itself needs to be specified. Furthermore, the demarcation of possibilities with sensor technology results in specific protocols and guidelines, which can be applied under provided circumstances. Within this thesis, the wireless sensor network MyriaNed will be used. 1.4 From Virtual Reality to real life As stated, the original Technology Acceptance Model is not complete to test the user acceptance of new technology, since costs, information sharing and experiencing the technology are also a major part in the users’ consideration (Sponselee, et al., 2008). The costs are defined by the market and the company who develops the technology, while the company is also responsible for providing information about their products. User acquaintance with the new technology is something a company can provide, but the way in which customers experience the technology is hard to predict. When adding technology to Smart Homes, this makes it even harder since a Smart Home is not a simple device that can easily be presented to potential users. When conducting a user‐centred approach to assess user experience, it is essential to have explicit, observable, and tacit knowledge of the consumers, since it is crucial to know what the consumers think, say, do, know or feel. A visual design tool can serve as a platform to link users’ thoughts and ideas (Sanders, 2002). In order to introduce people to new equipment or technology, a Living Lab can be developed through which researchers can learn about the users’ attitude and view towards the newly presented product. Within Living Labs, a variety of methodologies and tools are applied in real world environments to validate open innovations by involving end‐users in the development of a new product. By applying a user‐centred instead of a technology‐centred approach, innovations are tested through co‐creation, exploration, experimentation and evaluation. It can be an important tool to test user acceptance and to explore how users are experiencing the technology at hand (Kusiak, 2007; Molinari, 2011; Schumacher & Feurstein, 2007). In the exploration phase, it is needed to involve all potential end‐users as soon as possible, since they will assist in the co‐creation process. Through this, new settings, appliances and behaviours are detected by exposing the users to live scenarios in real or virtual environments (Pallot, 2009; Schumacher, 2011). This emphasises that Living Lab environments can be created in real life environments, although these environments might also be virtual, whether this is Virtual Reality or augmented or mixed reality. Likewise, Bergvall‐Kåreborn et al. (2009) state that Living Labs try to create realistic situations through contexts, users, use

6 circumstances, and technologies. Moreover, they claim there is no distinction between the real or virtual world and results obtained in the virtual world could as well be valid for real market situations (Bergvall‐Kåreborn, et al., 2009). Taking this into account, conducting research on user acceptance and user experience of sensor technology in Smart Homes can also take place in VR. Through this, the user experience can be taken to the customer instead of users coming to a Living Lab. Considering cost and time aspects, this seems the most appropriate approach within this thesis. Furthermore, the different target groups, for instance elderly, that are necessary to participate within this study, need to be able to attend to the Virtual Reality experiment. 1.5 Goal and research question 1.5.1 Goal There exists limited research in the area of user acceptance, nor has there been research about user experience with sensor technology and the relation to the acceptance and implementation of Smart Homes in VR. This represents a problem between the practical and theoretical use, which leads to the goal of this research:

To investigate if experiencing Smart Home sensor technology, named MyriaNed, through Virtual Reality can lead to a higher user acceptance and implementation of Smart Homes.

1.5.2 Research question and sub‐questions This goal has been formulated into the following research question:

Which factors affect intention to use and to what extent can user experience with sensor technology through Virtual Reality enhance the user acceptance and contribute to the implementation of Smart Homes?

The formulated research question of this thesis is twofold. To answer the research question , several sub‐questions are formulated:

Intention to use  How can the intention to use of Smart Homes be measured?  Which constructs are necessary in order to examine the user acceptance of Smart Homes? Users / Sample  How are the data obtained and what is the composition of the sample?  How do the participants perceive the Virtual Reality experience? Sensor technology  What comprises sensor technology and what can be measured?  How is sensor technology applied within Smart Homes?  How can sensor technology support users of Smart Homes?  How can MyriaNed be virtually implemented within the VR experiment?

7 Virtual Reality experience  What is necessary to set up a realistic Virtual Reality program?  How will MyriaNed be implemented within the virtual Smart Home?  How is the virtual Smart Home designed to create a realistic setting? User acceptance model  How can the constructs and moderators of the user acceptance model be operationalized for the survey?  How can the Virtual Reality experience and observations be included in the survey? Outcome and implementation  Which statistical analyses are necessary to analyse the data?  What are the results of the pre‐ and post‐survey comparison?  How can the results of the analyses be interpreted? 1.6 Research design and readability In order to understand the process followed within this thesis, a research design is composed in Figure 1.1. The fact that the research question is twofold will be explained by the following. At first, it is necessary to understand the sensor technology MyriaNed and to identify the factors affecting the intention to use of Smart Homes in order to develop a user acceptance model. Next, it can be examined to what extent a user experience can enhance the user acceptance and contribute to the implementation of Smart Homes. Within this research, a user experience is provided through a Virtual Reality Smart Home experiment. In order to validate the impact of the experience, a pre‐ and post‐survey are included, which are derived from a newly developed user acceptance model for Smart Homes. The surveys in combination with the Virtual Reality experience form the experiment from which data will be collected, analysed and interpreted.

The research method conducted in this thesis will be a quantitative research, due to the fact that besides the Virtual Reality Smart Home tour, a pre‐ and post‐survey are utilized. The experiment will be held at the Eindhoven University of Technology and paper surveys will be provided to the participants. The target group required is hardly demarcated, since it is necessary for the research to reach participants out of every age category, social, educational and technological level. The questions of the surveys are based on the user acceptance model that will be developed and operationalized for this research. As results from the experiment need to be generalizable to other populations, a sufficient sample size consists of at least one hundred respondents (Baarda & De Goede, 2006). Only then can the potential relationships between constructs and moderators be analysed. Within these analyses, it is essential to assess the reliability and validity of the model. Eventually, it has to be examined whether a change in results between the pre‐ and post‐survey can attribute in answering the research question as formulated in this chapter.

8 Step Literature study 1

Choose gap: user acceptance of Smart Homes

Research introduction

Formulate Formulate Step Formulate motive problem research 2 & relevance definition & goal questions

Choose Development VR method: VR & program survey

Define and validate variables and set up experiment Convert to Step Define target User Acceptance practical and Define MyriaNed Prepare surveys 3 groups Model theoretical criteria

Select sample Experiment size

Pre‐ survey Perform experiment

Step Gather target Follow three‐step Virtual Reality 4 groups program experience

Choose: Post‐survey analytical method: SPSS

Step Process & analyse 5 data

Choose: statistical tests

Step Results 6

Conclusions & Step recommen‐ 7 dations

Figure 1.1 Research design

9 1.7 Thesis outline This thesis is structured as follows. The next chapter encloses the technological aspects considered within this thesis. Chapter 2 describes the essentials of sensor technology, as it is an elementary component of the Smart Home concept. User acceptance is closely related to understandability and therefore it is necessary to elaborate more on the system behind the technology. Eventually, this knowledge is utilized to assess MyriaNed as representative sensor protocol of the wireless sensor networks. Based on its characteristics, it is discussed how MyriaNed can be applied within the Virtual Reality Smart Home experiment. When the technology is clarified, the way in which the user acceptance of Smart Homes can be tested is described. The experiment needs an underlying model in order to obtain results that can be analysed statistically. Therefore, Chapter 3 discusses potential user acceptance models. Eventually, this is necessary to develop a new user acceptance model in which the appropriate constructs and moderators are selected for this thesis. All elements that are discussed until then will be implemented within the setup of the experiment. Chapters 4 and 5 will comprise the entire development of the experiment, discussing each individual part. Chapter 4 focuses on the surveys that will be used to analyse the experiment. The survey questions are based on the constructs and moderators of the newly created user acceptance model. The development of the Virtual Reality Smart Home experience will be further elaborated in Chapter 5. Eventually, after conducting the experiment, the data are collected and described in Chapter 6 and analysed in Chapter 7. Chapter 8 concludes this thesis by discussing the main findings and recommendations for future implementations and research.

10 2 Sensor technology: The essentials 2.1 Introduction When entering the world of Smart Homes, there are a lot of technologies involved that may or may not be familiar to the public. Therefore, the concept of sensor technology needs to be clarified before talking about its implementation into Smart Homes. In order to get a better understanding of sensor technology, it is essential to first elaborate more on the system behind the technology. This will form the foundation for the further outline of this thesis. In this chapter, the different characteristics of a sensor and sensor network will provide the necessary knowhow in order to define why MyriaNed is the most suitable wireless sensor network (WSN) for the Virtual Reality experiment. Eventually, this technology will not be physically implemented, since the experiment will be held within the VR setting. Yet, the technology specifications are applied to create a highly realistic and representative reflection.

Due to the divergent terminology used within sensor technology, it is first essential to grasp the idea behind sensors and to distinguish the differences between sensors, transducers and actuators. This will be elaborated in Section 2.2. From there on, further detail will be provided on the sensor characteristics in Section 2.3, since most of the different characteristics might seem alike or at least highly related, although this is not always the case. Considering the research question, it is vital to know what kind of sensor types are used within Smart Homes. Section 2.4 will provide insight in that matter. Since sensors most of the time do not operate as individual devices, information about sensor behaviour within a network is presented. There are multiple network topologies applied, all having both advantages as well as disadvantages. Since the topology is crucial for the way in which sensors can communicate amongst each other, this will be clarified in Section 2.5. Within this section, the main criteria for choosing the appropriate WSN are explained. Comparing WSNs is not as unambiguous as it might seem. The terminology used and the denominations applied amongst engineers and scientists differ in such a way that it makes the matter unnecessary complex. Within the scientific literature, it is hard to identify articles that perform a wide comparison on multiple WSNs as a whole. This is due to the fact that WSNs consist of multiple components that operate from different perspectives based on the applications it needs to run. For instance, there are different hardware and software components, operating systems, topologies, communication standards, protocols, and applications involved (Gopalakrishnan Nair, et al., 2011). In fact, a wireless sensor network platform is just an umbrella term for all of those different components involved. The platforms are not commonly known as it is the composition and application of the platforms that defines them. The applied communication standards and protocols are more known since they are applied within multiple platforms. Therefore, it would be better to identify the key components necessary for this research purpose instead of looking at each individual WSN platform. Any company can build hardware for a communication protocol. Therefore, the list of platform manufacturers is endless, although the communication standards are limited (Berrios, 2008). In other words, there is no general WSN platform within any industry, whereas there are common communication standards and protocols applied worldwide. Section 2.6 will provide valuable insight in communication protocols on forehand, before a comparison and evaluation of these protocols will take place in Section 2.7. Afterwards,

11 necessary information will be provided on MyriaNed to find out how this technology can be implemented best. The chapter will conclude in Section 2.8. Appendix A, B and C will provide valuable background and comparing information to support this chapter.

The main sub‐questions central to this chapter are:

 What comprises sensor technology and what can be measured?  How is sensor technology applied within Smart Homes?  How can sensor technology support users of Smart Homes?  How can MyriaNed be virtually implemented within the VR experiment? 2.2 Sensor terminology: sensors, transducers and actuators In order to understand sensor technology, it is important to know what a sensor does. Logically, a sensor is an object that can sense something. This something might be the state of a physical object or process, for instance the humidity in a room, or a change in an electromagnetic field or conductivity. Within this thesis, the following definition of a sensor is applied:

‘’A sensor is a device that converts a physical phenomenon into an electrical or optical signal that can be quantified.’’

By seizing an external input and transforming it into an output, a sensor transmits non‐ electrical information from the physical world into an output in the world of electric and optic equipment (Kumar, 2013; Kenny, 2005; Dargie & Poellabauer, 2010). This process can be holistic and complex, for instance in the healthcare industry or in computers, but it might also be more simplistic and straightforward, like a domestic thermometer. The external input, which can be detected, comprises of phenomena like physical motion, pressure, humidity or heat. The input obtained from the physical environment is detected by the sensor, after which the device responds by transforming the input through an embedded component into an electrical signal. This electrical or optical output can then be converted into anything from a door that opens automatically up to the administration of a drug to a patient in a medical facility. The sensor itself consists of a sensitive component, that can detect the external, physical quantities and a conversion component, that can process this input and convert it into an electrical signal. The output signals are either analogue or digital, but either way they can be analysed and measured (Wang & Liu, 2011; Dargie & Poellabauer, 2010). More detail about this signal will follow in Section 2.5.

The embedded role and appliances of sensors in society are very diverse. Within multiple sectors, sensors are an indispensable asset in conducting industrial processes, whether these are in the physical, chemical or biological area of expertise. For instance, sensors are used in food processing, monitoring activities, computer systems or the automotive industry. In fact, even the human body carries natural sensors within its ears, eyes, nose, tongue and skin to capture phenomena like sound, light, scent, taste, temperature or pressure. The human brain receives electrical signals that are transmitted when the human sensors detect a stimuli (Wang & Liu, 2011; McGrath & Scanaill, 2014; Dargie &

12 Poellabauer, 2010; Ansari & Chandak, 2014). In the United States, the term sensor is most commonly applied, whereas when referring to the measurement system in Europe, generally the word transducer is used. Most of the time, the terms sensor and transducer are used interchangeably, which is incorrect. A transducer is a device that transforms one type of energy into another. When talking about the sensor input, aspects like light, pressure or temperature are also forms of energy, as well as the electrical or optical output signals. Therefore, confusion occurs in choosing the right denomination. The terms sensor and transducer are derived from the Latin words sentire (to perceive) and transducere (to lead across). This indicates that a sensor is just a sensing element in order to perceive a phenomenon, while a transducer is a sensing element with corresponding electronic circuits necessary for a conversion (Stefanescu, 2011; Kyriacou, 2009). Accordingly, often sensors are an integrated part of transducers. When people tend to use both terms in their report, although not common, they merely refer to sensors as being the actual sensing part, whereas the transducer is the device involved. This would make the transducer a device that carries a sensor.

Due to the complexity of finding unanimity on the distinction between the two terms, the term sensor will be applied within this thesis to cover the entire sensing process. This will be the sensing process as displayed in Figure 2.1. It is ineffective to use both terms, since it will cause confusion and continuously safeguarding the distinction will only distract from the core of this thesis. Furthermore, McGrath and Scanaill (2014) indicate that this difference in terminology perspective is even visible between scientists and engineers. However, this dissimilarity is merely on academic grounds in regard to the development of applications. The dichotomy in definition has only minor impact on the actual ability to apply sensor technology within its designated context. Therefore, it is safe to solely use the term sensor technology instead of transducer technology (McGrath & Scanaill, 2014).

Within the context of sensors, actuators are an important and necessary asset. As stated, sensors convert a physical signal into a readable, electric signal. An actuator is a device which is applied to convert that incoming signal back into a physical phenomena. Like the term suggests, an actuator is an instrument that actuates or starts something. For instance, when starting a car, the electric motor inside the vehicle transforms electrical energy into mechanical energy, which in this case evolves into kinetic energy or physical motion. Besides electrical energy, actuators might also operate on hydraulic or pneumatic power, like hydraulic cylinders, a valve or relay. Actuators are necessary to provide feedback or to control the physical environment (de Silva, 2007; SCME, 2009; Dargie & Poellabauer, 2010; Nugent, et al., 2014; Varmah, 2010). In Figure 2.1, besides the sensing process, the positioning of the controller and actuator is displayed. In order to understand the individual elements, Figure 2.2 in Appendix A.1 provides two examples.

Figure 2.1 The sensing process and the link to the actuation phase

13 One of the main features of a sensor is its ability to sense physical phenomena. There are three ways in which a sensor can detect a physical quantity: by physical contact, noncontact or sample removal. Sensors that require physical contact have to be placed upon or in the detectable quantity. This can be done when sensing in liquids, gases or the human body. In the noncontact approach this direct contact is not necessary, the examined quantity can stay in its natural condition, since it will not be disturbed. When examining behaviour in a Smart Home, ideally the residents can move freely through the house, without any hinder from devices or obstacles. In such a case, passive infrared (PIR) sensors are useful. Remote sensing is also part of the noncontact principle, mostly applied to observe phenomena on earth. In order for sensors to detect a physical quantity by utilizing sample removal, part of the object needs to be collected. This approach is mostly used within the healthcare and environment domain. For instance, to monitor the glucose level in blood, a representative tissue sample is obtained from the human body. This removal is rather invasive, however sometimes a urine sample might be sufficient to obtain the information necessary. Besides using sensors, laboratory instruments can be applied within this latter context as well (McGrath & Scanaill, 2014). 2.3 Sensor characteristics Besides the way in which sensors can detect physical quantities, there are plenty of other common characteristics, that are crucial for sensor purposes and performances within any industry. The most important characteristics are defined below (Kenny, 2005; Flynn, 1990; Irish, 2005; Carr & Brown, 1998; Christ & Wernli, 2013; van der Horn & Huijsing, 1998; NI, 2015).

It is important to know the sensitivity of a sensor, since it provides information about the minimum input that is required for a sensor to have a measurable output, thereby distinguishing the relation between the physical input signal and the electrical output signal. Whenever the input parameter unit alters, this should have consequences for the output signal. In general, the level of sensitivity is determined by the change in output signal of the sensor per unit change in the physical quantity. An example is provided in Appendix A.2. When drawing this relation in a graph, a steeper line resembles a higher sensitivity. This sensitivity could either be constant or non‐constant over the entire range of the sensor, resulting in a linear or nonlinear relation between the units.

Every sensor has a certain range, being a maximum and minimum value in which the sensor can operate accordingly. Different sensors might have various ranges that are measureable. Under some circumstances, a sensor can work outside of its range. However, in such a case additional adjustments are required to obtain representative results. Normally, when the physical input quantity exceeds the range of the sensor, this results in an inaccurate measurement or even causes damage to the sensor. The range defines the region in which input signals can be converted into an electrical signal by the designated sensor. The upper limit of the range represents the maximum quantity that the sensor can measure, whereas the lower limit characterizes the minimum quantity. Taking the example of a barometric sensor, its range might be from 20 to 120 kPa. Within the context of the range of sensors, the term span is often utilized. The span of this barometric sensor is 100 kPa.

14 Another important asset of sensors is the accuracy of the instrument. Whenever a physical quantity like heat is detected and converted into an electrical signal by using a thermometer, it is generally assumed that the output as presented on the display is the actual temperature measured. However, there is always some sort of uncertainty or error between the measured and actual quantity. The maximum deviation possible between the indicated value on the screen of the sensor and the actual value is called the accuracy of the sensor. The actual value is the Full Scale Output (FSO) measured by a primary standard and therefore considered the true value. The accuracy is usually defined as a percentage of the ideal FSO, but it can also be presented in an absolute value. If a developer of a light sensor guarantees that the sensor has an accuracy of four percent of the FSO, this indicates that the output value gained from the sensor is within four percent of the actual, real value in the environment. The higher the accuracy (being a lower percentage), the lower the uncertainty or discrepancy between the indicated and true value. If a sensor has a relatively wide range, it is more likely that the accuracy of that sensor will be relatively low.

Within the calibrating process of sensors, besides accuracy, the precision of the instrument is evaluated and adjusted when necessary. The precision of a sensor determines the reproducibility of sensor results if the exact same conditions are measured again. Also, precision can be defined by the significance of the sensor’s displayed number, meaning the number of digits behind the comma that a sensor can measure. A thermometer that can measure the temperature as 21,6°C is more precise than another thermometer measuring 22°C. By reproducibility, ideally the same results are displayed over and over again on the sensor screen, although in reality this consistency is not always the case. This can be due to for instance the measuring methods or the circumstances in which is measurement is performed. The difference between accuracy and precision is enlightened in Appendix A.3.

Another aspect which is most often mentioned within sensor terminology is the resolution of a sensor. The resolution indicates the minimum signal fluctuation that is detectable by the instrument. This smallest variation should be notable within the last digit of the output display of the sensor. Note that there is a difference between sensitivity and resolution. The resolution provides information about the discrete level of value which is displayed by the sensor, also known as the code width. It is the ability to see the smallest difference in readings. The sensitivity of a sensor, on the other hand, indicates the minimum change necessary to provide a measurable output. The difference between sensitivity and resolution is more distinct in the example of Appendix A.4.

The last relevant characteristic highlighted is the response time of a sensor. This characteristic indicates the time required for a sensor to respond to a changing input quantity by converting it into a perceptible output value. There is always a certain time lag between the change in parameter input and the newly adjusted output signal. The response time of a barometric pressure sensor might for instance be 1 ms. Response time is sometimes mistaken for latency, another common characteristic. However, latency is the delay period that a message is travelling from one point to another. In the case of a sensor, the response time resembles the entire time it takes for the sensor to display the changing input, whereas the latency is the time the incoming input is transferred to the converter. Latency is more often used in network situations, in which messages are

15 spread. Logically, the response time is always higher than the latency. Despite the characteristics discussed, there are plenty more features that entail the sensor principle. Obviously, it is essential for a sensor to fulfil the task required, but this task can be manifold. Choosing the correct sensor according to its characteristics really depends on the situation in which it will be applied. 2.4 Sensor types for Smart Homes There is a wide range of sensors on the market, varying from chemical, electrical and mechanical sensors, up to optical and thermal ones (SCME, 2009). Several of these sensors are used within Smart Homes. Before heading into those, it is important to identify the dichotomy in which sensors can be classified: passive and active sensors. Passive sensors detect energy within their environment without requiring an external source. These sensors are used to record an object’s natural radiation. A well‐known example is a passive infrared sensor, in which the infrared light that is radiated by objects can be captured (Dargie & Poellabauer, 2010). In Smart Homes, PIR sensors are used to detect user activity since humans emit infrared light as well. Even the human senses have passive sensors, as the nose detects smell and sends this information to the brain (Kumar, 2013). Active sensors, on the other hand, do require an external power source. They have to emit a specific signal into the environment in order to receive information back from the transmitted signal (Kenny, 2005). For instance, a fishing boat uses sonar to send sound waves into the water to scan the area. Whenever the sound wave hits a fish, the signal is reflected back to the boat. Through this, the sensor itself has to send out a signal to an object in order to get information, whereas a passive sensor absorbs the information signals from an object directly. A more appealing example occurs when using a camera. Sensors within the camera are active whenever the user applies flash, since it sends out a signal in the form of light in order to illuminate an object. Through the object, energy is reflected back to the sensor within the camera. When the photographer does not need a flash, the camera uses natural energy from the sun which is captured by the sensor in order to make a photo.

In order for a home to become a Smart Home, certain technologies have to interact with each other and their environment. Sensor technology is an indispensable part of the home automation. Within a Smart Home, multiple sensors are almost invisibly integrated into the home. The most common sensors applied for Smart Home purposes are listed in Table 2.1. The term remote sensors is derived from the more commonly applied remote sensing, defining those sensors that do not need physical contact to gather information. Although the sensors itself may come in different forms, they are based on the type of sensors and measurands as in Table 2.1 (Kavitha, et al., 2012; Ding, et al., 2011; Dargie & Poellabauer, 2010; Lewis, 2005; Bouchon‐Meunier, et al., 2015). These sensors can be used to solely monitor residents inside a home, but in order to really assist dwellers in their daily activities, monitoring is not enough. In the Section 1.3, a distinction has been made between four categories of Smart Home technology. Considering the different areas, sensors technology is mostly involved in the latter two, being the semi‐automatic and the fully automatic applications and devices. The semi‐automatic devices are flexible or functional objects that are placed within a Smart Home environment, whereas the automatic applications are more part of the Smart Home. The sensors displayed in Table

16 2.1, as part of the sensor and actuator network, are therefore considered part of the fully automatic applications.

Table 2.1 Most common sensor types used within Smart Homes

Type of sensor Measurand Example Environmental sensors Temperature, humidity, flow, Humidity sensor, thermistor pressure Remote sensors Acceleration, position, motion, PIR sensor, optoelectronic velocity, proximity sensor Contact sensors Strain, force, vibration, personal Piezoresistive sensors, RFID identification

Piezoresistive sensors are sensors that change the resistance when a strain, force, pressure or acceleration is applied. These sensors can be utilized to track body movements by attaching these sensors to the human body. These piezoresistive sensors can either be wearable sensors or their piezoresistive effect is part of a wearable pressure or acceleration sensor (Lewis, 2005; Orengo, et al., 2010; Dargie & Poellabauer, 2010). Radio‐frequency identification (RFID) can either be used as a contact sensor, like when taking a fingerprint, or as an active remote sensor in retinal scans. The optoelectronic sensor is applied within certain remote sensors that deal with light. Cameras are also involved in this category. Video cameras and RFID can provide valuable insight in residents’ daily activities, however there is a huge downside in using these devices. Storage and information extracting challenges can quite easily be overcome, but privacy challenges are far more complicated (Caine, et al., 2005; Ding, et al., 2011). In fact, privacy concerns are one of the main barriers for assistive technology in home automation and Smart Home appliances. In those cases, people really have to assess whether their privacy concerns are more highly valued and significant compared to their needs and potentially enhanced comfort. The feeling of security and safety that come along with the privacy concerns, are essential factors as well (Magnusson & Hanson, 2003; Ehrenhard, et al., 2014; Courtney, 2006). For the user acceptance of Smart Homes and the technology involved it is crucial to understand the impact of using RFID and video cameras within this environment. Therefore, it is vital to identify whether it is absolutely necessary to implement these types of sensors within the Smart Home environment. 2.5 Sensor networks and criteria In most cases, sensors do not operate as individual features in a Smart Home environment. Sensors can cooperate and communicate in a system or network, in which the individual sensor units are called sensor nodes. A node is an autonomous device, equipped with a designated sensor. A sensor node always has two components: a hardware and software component. The hardware component is utilized to detect the external physical phenomena. Under normal circumstances, this sensitive hardware feature is clearly visible on a sensor, as it protrudes from the device. The software component is necessary to convert the input signal into an output signal, by interpreting and transforming the physical phenomena. Through these two components, the sensors can be programmed according to the users’ needs (Kumar, 2013). In general, after detecting the quantity and conditioning the signal, the resulting analogue signal is converted to a digital signal by using a analogue‐to‐digital converter (ADC). Of course, when an analogue signal is required, this is not necessary. Later on, in the actuation

17 phase, the digital signal is converted back into an analogue signal. This is done by applying a digital‐to‐analogue converter (DAC) (Dargie & Poellabauer, 2010).

When obtaining data from an environment, more than one node is necessary to fulfil that task. In fact, maybe a dozen or even a hundred nodes are required to access the information requested. The sensor nodes are interconnected and collaborate in such a way that they form a sensor net of nodes, also known as a sensor network. The individual, autonomous sensor nodes communicate their information to each other or directly back to a central point within the network. The sensor network can either be cabled or wireless. In a larger network, with multiple nodes, it is more practical to apply a wireless sensor network. Based on various studies, the following definition of a wireless sensor network (WSN) is provided within this thesis (Iyer, et al., 2008; Papageorgiou, 2003; Kumar, 2013):

‘’A wireless sensor network is a group of tiny wireless sensor devices, called sensor nodes, which are deployed in an ad‐hoc fashion to cooperate on sensing a physical phenomenon in order to perform a common application.’’

The wireless sensor nodes transmit data by utilizing their on‐board wireless radios and processing facilities. Within a WSN, different types of sensor nodes may be involved, for instance for monitoring environmental condition in a living room, by using temperature, pressure and humidity sensors. Since a wireless autonomous sensor node is required to collect and distribute data, in addition to the sensing component the sensor needs storage capabilities, a communication module, a battery, and a process controller (Dargie & Poellabauer, 2010).

A WSN consists of three elements: a node, a data gateway and external software or an external system. The data collected by the nodes is transmitted to a certain head node, better known as the base station or the gateway. This gateway captures the data received and connects the sensor network to other wired or wireless networks in which the data can be analysed by using software. The way in which the communication network is structured can alter according to the network topology. The network topology is the layout pattern in which nodes are linked to form a network in which they communicate. There are some basic network topologies, described in Table 2.2 (Townsend & Arms, 2005; Lewis, 2005; Sharma, et al., 2013; Sharma, et al., 2013). Since most of the topologies contain multiple links, it would be best to create a wireless network in which cable use is minimized. In the context of Smart Homes, it would be unthinkable to create and live in a home situation in which various cables are tangled and scattered throughout the living room. Therefore, it is essential for a Smart Homes to have a wireless sensor network, in order to provide comfort and ease to its residents.

18 Table 2.2 Different network topologies applied within a (wireless) sensor network

Topology Definition Form Advantage Disadvantage Application Point‐to‐ Direct link Direct link Small areas Between two point between two Reliable schools or POS

nodes Fast systems Bus Nodes Easy to connect Depends on main Ethernet connected to Easy to expand cable networks one main cable Low cost Difficult to repair Local area called the bus Small network networks (LAN) Ring Nodes are Easy to install Adding nodes Office building connected to Easy to manage If one node fails, School campus one another in a Locate defect the entire closed loop Handling traffic network fails

Tree Star‐configured Large networks When cut, all Cable TV network Manageability communication connected to a Easy to expand is gone bus backbone Difficult to maintain Star Nodes Low start‐up Gateway Fibre‐optic connected to a cost dependence cable central gateway Easy Expensive Home system node extendable Router and Robust laptops

network Mesh Nodes have Handling traffic Expensive Internet direct Large networks Redundant Internet of connection to at Robust connections Things least two other Security nodes Hybrid Combination of Reliable Design Wide area various network Low power complexity networks topologies Flexibility High costs to (WAN) connect two topologies

In the current market, there are multiple wireless sensor networks. In order to define the best suited WSN for the virtual experiment setting, despite the aspects discussed, there are a couple of other criteria that need to be considered for the use in a Smart Home. A decent WSN requires a low energy consumption, high scalability, self‐management capabilities, high network response, high reliability, low costs, easy maintenance, and high security (Townsend & Arms, 2005; Patel, et al., 2011; Sharma, et al., 2013; Dargie & Poellabauer, 2010; Mahalik, 2007). For a Smart Home, low energy consumption is a very important factor, together with easy maintenance and high reliability. Whenever a WSN is installed, users should be able to depend on the system without failure. In that case, self‐management can ensure that the system can repair itself to a certain level without intervention of a user or mechanic. 2.6 Communication protocols within a WSN By applying wireless sensor networks, aspects like cable costs, maintenance, connection failures, and installation costs can be narrowed or even eliminated (Khan, et al., 2012). This can provide multiple benefits to a sensor network. Since the wires are disposed, the nodes have to communicate differently. The way in which nodes can transmit and receive data is defined within the physical layer of the OSI model. The Open Systems

19 Interconnection (OSI) reference model or protocol stack defines how data can be transmitted from one network or computer to another. Besides the physical layer, there are a total of seven layers that describe this process, starting from the physical up to the application layer, see Figure 2.3. The physical layer describes how the links between networks can be activated and maintained, including the transmitting medium (Simoneau, 2006). Every sensor node has an onboard wireless radio that is utilized for transmitted radio frequency (RF) signals. Since the sensor nodes are only small and energy consumption is very important within WSNs, the operating frequency of the physical layer of the network needs to be low (Townsend & Arms, 2005). There are a number of standards developed by the Institute of Electrical and Electronics Engineers (IEEE). These standards, as displayed in Table 2.3, are common radio protocols within the physical layer of networks (Townsend & Arms, 2005; Mahalik, 2007; Frank, 2013). Besides the physical layer, part of the second layer (data layer) of the OSI model is defined within the IEEE standards. This part of the data layer is called the Media Access Control (MAC) sub‐layer, in which protocols are used in order to ensure that opposing transmitted signals of two different nodes do not collide within the same channel (Dargie & Poellabauer, 2010).

Table 2.3 Radio standards developed by the IEEE Standards Application Frequency Data rate Range Energy IEEE 802.11 Wi‐Fi 2.4 GHZ and 5 GHz 11 and 54 Mbits/s 50‐100m High IEEE 802.15.1 Bluetooth 2.4 – 2.485 GHz 1 Mbits/s 10m Medium IEEE 802.15.4 ZigBee 868 MHz, 915 MHz, 20 ‐ 250 Kbits/s 10‐100m Very low and 2,4 GHz

IEEE 802.11 can be applied in wireless local area network (WLAN) communication that requires high bandwidth and data rate, however for the use in low‐power sensor networks the standard is precluded due to the high power demands. IEEE 802.15 standards are known as wireless personal area network (WPAN) standards. At forehand, Bluetooth might seem appropriate since the required energy consumption is lower, however relatively speaking the level of power is still high compared to its maximum transmission range. Furthermore, IEEE 802.15.1 operates from a star topology, only supporting up to seven nodes within a single network, making the standard pretty limited (Townsend & Arms, 2005). This single hop approach, in which the nodes communicate directly to the gateway node, is rather constrained compared to multi‐hop networks, in which nodes can transfer information through each other before sending it to the gateway node (Dargie & Poellabauer, 2010). Although many more IEEE standards are developed, the most relevant is the IEEE 802.15.4, especially designed in order to encounter the previously formulated problems for wireless sensor applications. Despite the fact that the standard requires low power, it can still transmit at a long range at the most common, license free bandwidth worldwide, 2.4 GHz. A special feature within this standard is the ability of the radio of the node to be set into sleeping mode, reducing the amount of energy. When requested, the node can be awakened and synchronized immediately to the network (Mahalik, 2007; Townsend & Arms, 2005). IEEE 802.15.4 is highly associated with ZigBee, the first communication protocol based on this standard for low‐power sensor networks. The IEEE standard only defines the first two layers within the OSI model, whereas ZigBee extends the standard by the completing the upper layers

20 of this OSI model like depicted in Figure 2.3 (Lönn & Olsson, 2005; Townsend & Arms, 2005).

Considering the emphasis of this thesis, it is not relevant to Application layer Applications Customer discuss the other layers involved. More importantly, ZigBee is a wireless technology Presentation layer Application Profiles and communication protocol Application Session layer used within home automation Framework ZigBee and Smart Home industry. Network and There are more protocols Transport layer Security Layers based on IEEE 802.15.4, like WirelessHART and ISA100.11a Network layer MAC Layer for the process and industrial Data link layer automation sector (Nixon, IEEE 802.15.4 Physical layer PHY layer 2012). However, besides IEEE standards, there are more standards that can function as OSI Model ZigBee Stack PHY and MAC layer. Within the home automation networks, Figure 2.3 ZigBee stack derived from the OSI Model there are many other radio frequency wireless solutions that compete with ZigBee through other protocol stacks. These protocol stacks might not even have seven layers, since they operate on different hardware and software components in the lower and upper layers (Fischione, 2009; Frank, 2013; Gopalakrishnan Nair, et al., 2011). Due to a different composition of the technology, other characteristics are presented in the application phase. The other RF wireless solutions for home automation networks are Z‐ Wave, INSTEON, MyriaNed, EnOcean, ONE‐NET, Wavenis and DASH7. The protocol stacks of these solutions are remarkably different compared to the OSI reference model. More importantly, they contain different features and characteristics which can be distinguished when applied in Smart Homes (Frank, 2013; Herman, 2014; Nixon, 2012; Schuts, et al., 2009; Gomez & Paradells, 2010). 2.7 Smart Home communication protocols – a comparison Within the home automation industry, the first communication protocol for linking electronic devices was developed in 1975 by the name of X10. The technology utilizes the wired home power line in order to send messages to other devices, providing only limited control (Gill, et al., 2009). Ever since, standards and protocols have been improved and nowadays far more advanced technologies are presented on the market. The most important ones for Smart Homes and home automation are briefly discussed in Appendix B before evaluating them on their prime features (EnOcean, 2008; Gratton, 2013; De Sanctis, et al., 2012; Mathur & Newe, 2014; Frank, 2013; ONE‐NET, 2012; Schuts, et al., 2009; Rathnayaka, et al., 2011; Rathnayaka, et al., 2012; Gomez & Paradells, 2010).

The reason for the extensive amount of wireless networks is that all protocols believe that they solve the network problem in home environments and others are lacking to fulfil the comfort needs of residents. When evaluating wireless network protocols, it is essential to keep in mind the perspectives of the sources used. Most of the sources

21 available provide different technical specifications. Besides, some features displayed in the literature, like scalability or throughput, are rather theoretical maximums instead of practical parameters. Therefore, objectively evaluating wireless network protocols without possessing the actual systems, is hard to accomplish. After extensive filtering and critically assessing the literature, the most reliable characteristic quantities found are presented in Appendix C (Van Dijken, 2010; Haase, 2013; Martin, 2007; Kaur & Sharma, 2013; Ploennigs, et al., 2010; Nabi, et al., 2012; van Gijsel, et al., 2012; van Snick & Veys, 2014; Sivasankari, et al., 2014). Still, note that discrepancies might occur. For instance, the known data rate of ZigBee for 2.4 GHz is 250kbps. However, this is the absolute maximum throughput. Wielding this data rate would result in an ultra‐high energy consumption. Testing ZigBee PRO, an upgraded variant, in a more realistic setting with moving nodes resulted in a throughput of around 30kbps (Hamilton & Sampath, 2011). This depends on the transmission time and packet sizes of data send each time. For instance, the internal radio of MyriaNed is sleeping 99% of the time, providing a relatively low throughput, but a low‐power consumption as well. When monitoring temperature however, providing information two times per second is fine since no drastic changes will occur in the mean time, leaving the throughput relatively insignificant. Since an average room does not exceed ten meters in length, the range of Smart Home protocols can be limited. All protocols claim to be robust, secure and reliable, without complexity, which all four are important factors for Smart Home appliances and implementation (van Straten, 2012). Therefore, these are hard to compare. Latency, interoperability, cost, and power consumption, on the other hand, are easier to distinguish, see Appendix C. Latency and data collision are closely linked (Pountourakis & Baziana, 2007). Some relatively less important criteria for home automation purposes are throughput, scalability and the prior discussed indoor range (van Straten, 2012). These characteristics are already sufficient to perform the necessary appliances for almost all wireless networks.

Evaluating the important factors in Appendix C, the most appropriate protocols for the VR experiment are ZigBee, MyriaNed, and EnOcean. Considering the costs, ZigBee suits best, since MyriaNed and most likely EnOcean, due to the energy harvesting technology, are more expensive. However, ZigBee is rather limited in their flexibility to operate simultaneously with other protocols already present in a home environment. Comparing the ratios of costs to lifetime and power consumption, MyriaNed and EnOcean are more interesting options. As discussed in Chapter 1, the current barriers for Smart Home implementation and user acceptance are the lack of common protocol and interoperability, costs, flexibility, and security. Although MyriaNed and EnOcean barely differ, Kaur & Sharma (2013) point out that the security of the EnOcean protocol is pretty basic, whereas Van Dijken (2010) indicates that MyriaNed security is highly reliable. When transmitting messages using radio frequency technology, it is important that the sensitive data is secured in an optimized way, especially considering the current user acceptance barriers. Therefore, MyriaNed will be selected as representative wireless network protocol. 2.8 MyriaNed MyriaNed is a WSN protocol developed by the research and development department of a company called Devlab. Devlab is operating from within the Eindhoven University of Technology campus. The name MyriaNed is derived from the Greek word Myriad, which

22 means ten thousand, referring to its ad‐hoc scalability. Its main applications are within the building automation industry, transportation, energy efficiency, elderly support and monitoring environment (DevLab, 2013). The MyriaNed protocol operates without a single point of failure and works great when applications run distributed. The idea behind the WSN was to develop an ad‐hoc protocol with high energy efficiency, a long node and battery lifetime, and high scalability (DEMANES, 2013; Schuts, et al., 2009). Besides, it needed to be robust, self‐adaptable and self‐configurable. Through the bottom‐up, gossiping approach, each node is considered equal and has the same influence on the system. Its epidemic communication provides robustness since information is flooding to all participating nodes. This way of communication ensures that scalability can be very high, where various other routing approaches and networks only provide limitations when extending the network. Through experiments, MyriaNed showed to provide constant quality even when the number of nodes are scaled over one thousand (van Mierlo & van Lieshout, 2015). Despite a larger network, information can still be send in an ordinary fashion. Other protocols often find themselves struggling with the large number of nodes in order to keep the network stable, resulting in a major energy loss only to keep the network up and running. At such a point, MyriaNed can still use its total energy to send data. No individual nodes have to be addressed when altering the network behaviour (Chess, 2014; Zijderveld, 2012). A recent study indicates that when MyriaNed is implemented to achieve lighting control in offices, the control system can reduce 30 to 70 percent of the power consumption (Agueci, 2015). This percentage depends on the available daylight and the number of occupants. Considering this, Smart Homes could also really benefit from these experiences.

MyriaNed can both be implemented in existing and new environments. When using the protocol for the experiment inside the Virtual Reality Smart Home, the nodes of the network need to be within a twenty meter range of each other. This might not seem to propose a threat in a home setting, however the signal emitted by the MyriaNed node has difficulties travelling through concrete. Therefore, the nodes have to be placed efficiently within the home environment in order to deal with this potential limitation. Information is send two times a second and the travel time per node is half a second, so when travelling through twenty nodes it will have a delay of ten seconds (de Ruiter, 2013; van Kraaij, 2015). This delay might provide problems in other industries, however in the home automation industry this is acceptable. If there are eight nodes in a living room that measure humidity, this means that new information about the humidity will be sent every four seconds. No drastic changes will happen in such a short period. Although Devlab, in association with Van Mierlo Ingenieursbureau, developed numerous sensor types, not all of those are already available on the commercial market (van Mierlo & van Lieshout, 2015). The most common sensor types provided by MyriaNed are to detect temperature, light, humidity, acceleration, pressure, PIR and motion. For the purpose of a Smart Home, these sensors are sufficient to set up an appropriate sensor network within the home environment. However, MyriaNed is compatible to measure all those aspects, but it cannot measure them at the same time yet by a single node (van Kraaij, 2015). Therefore, when installing the sensors in the Virtual Reality Smart Home, different types of sensors need to be placed within a single room. The actuators that come alongside the sensors can perform almost any task desired. This can vary from measuring blood pressure for health purposes up to security applications (EmenEm, 2013).

23 There are a couple of projects in which MyriaNed has been applied in the home environment. Researchers of the ALwEN project, Ambient Living with Embedded Networks, monitored the activities of elderly in a home with MyriaNed. The people who lived in the house performed general daily activities. The so‐called Big Brother‐house had 65 MyriaNed nodes and seven cameras to validate the results. The objective of this sensor home project was to observe if older adults can still live a longer independent life, in a healthy state without any illnesses, like dementia or epilepsy (DevLab, 2011; Almende, 2011). In the Orbis Elderly home in Sittard, a 150‐node MyriaNed sensor network was installed in order to track data. Furthermore, data was collected from accelerometers. Almende was also part of The Independent Living project, which is linked to the Big Brother‐house and the Orbis Elderly home. The project aimed to succeed in three areas: safety, social cohesion and health monitoring. User experiences were monitored, users were offered applications to improve social interaction and user demands were taken into account (Almende, 2010; Almende, 2011). 2.9 Conclusion Within this chapter, information has been provided regarding sensor technology and MyriaNed. Becoming familiar with the specific terminology is crucial for understanding sensors, wireless sensor networks and Smart Home technology. Sensor technology can be applied within multiple diverging industries and applications. Considering the Smart Home sector, sensors are used inside the home environment to monitor or track residents in order to improve health conditions or comfort issues. In order to achieve these goals, various sensors can be placed throughout a house, from the living room up to the bathroom. Light and motion sensors can detect whether there is a person inside a room, determining the proper light settings for that particular moment. Through this, not only comfort, but also health care support can be created, like fall detection and prevention or health monitoring through body sensors. Since sensors do not operate as individual devices for most of the time, an entire network of wireless sensors can be applied to provide the most optimized situation. Nowadays, it seems like the market is flooded with wireless sensor networks and protocols. Nevertheless, still new protocols are entering the market since the developers believe that their product will provide new assets and unique selling points in comparison to those already available. When Devlab developed MyriaNed, they indicated that the gossiping mesh interaction could really improve a wireless network. In the case of Smart Homes, MyriaNed distinguishes itself from the other protocols, especially in its scalability and power consumption. The sleeping mode together with the communication method provide energy savings that do not influence its performance. MyriaNed’s high security and reliability within a Smart Home is of great importance, since privacy and consistency form barriers that need to be addressed before the entire concept can benefit. MyriaNed’s limitations concerning its travel time and transmit through concrete are surmountable and do not pose impediments for this purpose.

Implementing MyriaNed within the VR experiment requires humidity, motion, light, temperature, passive infrared and pressure sensors. 3D accelerometers are not necessary for this implementation, since these register how speed changes over a period of time. These sensors are necessary at for instance doors of a room or cabinet. The data obtained is used to determine the speed at which automatic doors should open. Therefore, in the

24 phase of implementation, these accelerations are already known. In the Smart Home, it is crucial to safeguard the privacy of the residents. Therefore, cameras are forbidden under any circumstance. The PIR sensors, on the other hand, can easily track body heat energy without visual confirmation. Using motion sensors, accompanied by PIR sensors, would suffice to detect any presence in a home. Since the signal has difficulties transmitting through concrete, all sensors have to be placed strategically in the rooms. When this is done in an efficient way, the travel time and throughput will form no obstacle to detect measurands. As the ambient sensor nodes of MyriaNed are rather small, its presence will hardly be noticed within the house, thus creating a natural habitat. In such a setting, performing activities of a daily living can really be assisted by the technology at hand. Since the nodes will not be implemented physically but virtually, it has to be certain that there are no discrepancies between the virtual capabilities of the sensor node and the actual capabilities. Table 2.4 provides an overview of the sensor characteristics discussed in this chapter.

Table 2.4 Overview of the main sensor characteristics Characteristics Definition Sensitivity The minimum input that is required for a sensor to have a measurable output Precision Reproducibility of sensor results if measured under the exact same conditions Range The maximum and minimum value in which a sensor can operate Response time Time required for a sensor to convert an input change into an output value Resolution The minimum signal fluctuation that is detectable by a sensor Scalability The potential number of nodes that can be included in a sensor network Costs The amount of money necessary to obtain a sensor network/node Network topology Layout pattern in which nodes are linked to form a communication network Self‐management The way in which a sensor network is independent of user intervention Maintenance Whether it is necessary to maintain the sensor network manually Throughput The amount of information per unit of time that can be transmitted Accuracy The maximum deviation between the indicated and actual value of a sensor Transfer function The relation between the input and output signal of a sensor within a graph Energy consumption The energy necessary for the sensor technology to work accordingly Latency The delay period that a message is travelling from one point or node to another Data collision The risk that simultaneously sent data from two different nodes collide Interoperability The ability of a system or technology to cooperate with another system Reliability The degree to which it is possible to depend on a sensor network without failure Security The degree to which the safety of the sensor network is guaranteed

25 3 User acceptance model for Smart Homes 3.1 Introduction Nowadays, with the current market changing from a supply push to a demand pull orientated economy, operating according to a user‐centred approach is highly recommended. Smart Homes are not yet widely implemented, since Smart Home technology, like MyriaNed, lacks user acceptance. Only when Smart Homes become familiar to the potential clients and people understand and see the benefits that the technology can offer them, in proportion to the costs and complexity, Smart Home technology can be accepted by its users. After the acceptance, in a user‐centred market the demand will increase and eventually the implementation of these sensor networks in residences will result in an increase in Smart Homes (Balta‐Ozkan, et al., 2013; Davis, 1993; Sponselee, et al., 2008).

The leitmotiv running through this entire thesis comes down to enhancing the user acceptance of Smart Homes, since it is believed that increasing the user acceptance will lead to a higher intention to use, which contributes to a wider implementation of Smart Homes. This is ought to be accomplished by setting up an experiment in which user experience with Smart Home sensor technology is observed and analysed, in order to find out if this personal experience can change peoples’ opinions about intending to use Smart Home technology. Besides defining the appropriate sensor technology, another major feature is how to quantify and validate the experiment, in order to attain what it is intended to be achieved. Therefore, this chapter will have a closer look on how to validate the Virtual Reality experiment.

For decades, researchers have conducted studies on ways to examine and improve the user acceptance of technology and new systems. As there is still no generally applied, holistic solution, this implies the complexity of the matter. The complexity is mainly due to the fact that the user acceptance process involves individual, autonomous human beings. Unlike a technological system, their reactions cannot be predefined and will change over time. Therefore, it is important when developing a user acceptance model, to look from a participants’ point of view. As highlighted in the following quote (Rowling, 2000):

“Understanding is the first step to acceptance, and only with acceptance can there be recovery.”

Ironically, the quote by novelist J.K. Rowling shows an aspect of great importance for the user acceptance of Smart Homes: understanding. For one, understanding of the system, the potential users do not want to get tied up in red tape, modifications and complexities. Systems need to be easy in use and useful (Balta‐Ozkan, et al., 2013; Davis, 1989; Sponselee, et al., 2008; Surendran, 2012; Venkatesh, et al., 2003; Mayer, et al., 2011). Besides understanding, people want to apply sensor technology in Smart Homes as personal support to fulfil their needs. Consequently, they need to realize what benefits occur, so understanding of the needs and benefits is vital as well (Demiris, 2009; Ding, et al., 2011; Augusto, 2010; Sponselee, et al., 2008). By being able to utilize MyriaNed in

26 Smart Homes as support, it will be possible to overcome future challenges and people can get a better understanding and become more aware of for instance their own energy consumption, helping them to improve their way of life.

In general, the basic concepts underlying a user acceptance model, according to Venkatesh et al. (2003), Davis (1986) and Grønland (2010), are displayed in Figure 3.1.

Individual reactions to using Intention to use information Actual use of information information technology technology technology

Figure 3.1 Basic concepts for user acceptance model

The first technology acceptance and adoption models were developed by Rogers, with his Innovation Diffusion Theory (IDT) and Fishbein & Ajzen, with their Theory for Reasoned Action (TRA) (Rogers, 1983; Fishbein & Ajzen, 1975; Ajzen & Fishbein, 1980). Both models are used as a foundation for the most well‐known and widely applied Technology Acceptance Model (TAM) of Davis (Venkatesh, et al., 2003; Kamel, 2004; Davis, 1986). Ease of use and usefulness underlie the intention to use according to TAM (Davis, 1986; Davis, 1989; Davis, et al., 1989). In order to predict the user acceptance, which is directly linked to actual use, the most important predicting factor is intention to use. There are several models constructed and extensions developed in order to create a better fit for research to predict intention to use and actual use of technology (Venkatesh, 2015; Sponselee, 2013). For instance, Sponselee et al. (2008) argue that actual use is predicted by more than only intention to use and they propose this as one of the shortcomings of TAM.

These aspects are vital elements in conceptualizing a model to examine the user acceptance of Smart Homes. Instead of creating a model from scratch, it is useful to understand the most common models applied within this area of expertise. This will be accomplished by the means of analysing user acceptance models for high‐tech technologies. Four models will be clarified and their research designs and methods will be further elaborated. Eventually, this knowledge will be used to draw up a new model that will be applied within the experiment. Firstly, the Technology Acceptance Model (TAM) will be discussed in Section 3.2, which functions as a fundament for all of nowadays technology acceptance models. In Section 3.3 the second model, the Unified Theory of Acceptance and Use of Technology (UTAUT) model will be clarified, which is composed out of eight other acceptance models and mostly applied in organizational structures. Years later, as an extension of this UTAUT model, another model has been constructed in which the focus lays on smart products and smart environments. Further elaboration on that model will be in Section 3.4, before turning to the final model discussed in this chapter in Section 3.5. This model is developed by Dutch scientists and is called TAUM, the Tele‐care Acceptance and Use Model, based on an iterative process between benefits and use. To conclude, Section 3.6 will present the new drawn user acceptance model for this thesis, based on all discussed models.

27 The main sub‐questions that are central to this chapter are:

 How can the intention to use of Smart Homes be measured?  Which constructs are necessary in order to examine the user acceptance of Smart Homes? 3.2 Technology Acceptance Model 3.2.1 TAM: Model and construct definition

Perceived Usefulness

Attitude Toward Behavioural Actual System External Variables Using Intention to Use Use

Perceived Ease of Use

Figure 3.2 Technology Acceptance Model (TAM) (Davis, 1986)

The nowadays widely applied TAM had two major objectives when developed in 1986: creating a better understanding of the process that will lead to user acceptance and to provide a fundament for testing user acceptance of new systems before implementation (Davis, 1986). Although, during the years, the model has been modified and extended, the original model is still considered a vital instrument to make prior assessments (Bagozzi, 2007; Venkatesh, et al., 2003; Venkatesh & Davis, 2000; Venkatesh & Bala, 2008).

TAM consists of the six constructs as depicted in Figure 3.2. Within this organizational‐ oriented model, the most important constructs are perceived usefulness and perceived ease of use. Perceived usefulness is defined as the degree to which an individual believes a particular system would help in gaining job or comfort performance. The perceived ease of use construct, on the other hand, is created to identify the degree of ease associated with the system use, as to which extent using the system will be free of effort (Davis, 1986; Venkatesh, et al., 2003; Venkatesh, 2015). Both constructs are influenced by external variables, like cultural, political and social factors. Some of the social factors mentioned are skills and facilitating conditions (Surendran, 2012). When considering Smart Homes, social factors are essential elements when testing the user acceptance, whereas cultural and political factors are less relevant. For instance, imagine someone is already experienced in creating do‐it‐yourself sensor appliances for home automation systems, or a participant in the experiment already has a WSN in his residence or he or she works at Philips. This could affect their attitude towards the technology, compared to someone who is completely unknown to the matter and still uses a feature phone. Therefore, these social factors as defined by Surendran (2012) should not be ignored when developing the user acceptance model for Smart Home technology.

Perceived ease of use can also be influenced by for example training and documentation. As a matter of fact, like displayed in Figure 3.2, even perceived ease of use has an impact

28 on perceived usefulness. They both influence the attitude towards using a new system: when a system is perceived easy to use and useful, it can create a more positive attitude towards using. Perceived usefulness also influences the behavioural intention to use. If a system is perceived as useful and is thought of as an instrument to enhance job performance, this would have a positive effect on the intension of use. This would be different for perceived ease of use, since easiness is not a direct reason for using a system. Only when there is the belief that the system could provide a positive contribution to the current situation, and therefore be helpful, this would be the case. TAM considers behavioural intention to use, which is also influenced by attitude towards using, to be the only factor to directly influence the actual system use (Davis, 1986; Venkatesh, et al., 2003). 3.2.2 TAM: Statistical methods In order to validate the model, certain statistical analyses had to be applied. The relations between the constructs within TAM are theorized to be linear (Davis, 1986), as they also have been in the TRA model of Fishbein & Ajzen (Fishbein & Ajzen, 1975). The model was tested with a sample size of 120 participants, in which 112 useful cases remained.

Attitude toward using was determined by five measurement items. These measurement items are the questions necessary to provide valuable insight and to validate the constructs in the model. For attitude toward using, a seven‐point semantic differential rating scale is applied (Davis, 1986), which is identical to the one Ajzen & Fishbein (1980) used. Perceived usefulness and perceived ease of use were measured in ten items, also using the seven‐point semantic differential rating scales (Davis, 1989).

To evaluate the construct validity, Davis (1986) applied the multitrait‐multimethod technique and matrix, commonly used in that timeframe. Through this method, high levels of convergent and discriminant validity are verified for the applied constructs, indicating that these constructs are appropriately operationalized. By analysing the reliability of the measurements scales with Cronbach’s alpha, a minimum reliability of 0.80 for perceived usefulness, perceived ease of use, and attitude toward using is necessary. Cronbach’s Alpha indicated that all constructs exceeded 0.90, whether this was pooled or individually judged (Davis, 1986). By performing an ordinary least squares (OLS) regression analysis, the hypothesized relationships of the TAM model were tested. Perceived usefulness is the most important predictor of actual use in this model as it has direct influence on behavioural intention to use, even higher than the influence of attitude toward using on latter construct.

29 3.3 Unified Theory of Acceptance and Use of Technology 3.3.1 UTAUT: Model and construct definition

Performance Expentancy

Effort Expectancy

Behavioural Use Behaviour Intention Social Influence

Facilitating Conditions

Voluntariness Gender Age Experience of Use

Figure 3.3 Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh, et al., 2003)

UTAUT was developed by Venkatesh et al. (2003) to formulate a unified model that integrates elements of eight prominent and competing models. These eight models include the previously mentioned IDT, TRA, TAM and TAM2. The UTAUT model is mostly applied in organizational structures (Grønland, 2010). The overall goal of the model is to create a better understanding of the dependent variable of use behaviour. This is achieved by evaluating the user acceptance and usage behaviour determinants of the eight models. It was shown that four constructs have a direct link to either behavioural intention or use behaviour, as depicted in Figure 3.3.

Performance expectancy entails the extent to which a person thinks that his job performance will be enhanced when he or she will use a new system (Venkatesh, et al., 2003). The construct is based on five more or less similar constructs of the other models, being perceived usefulness, extrinsic motivation, job‐fit, relative advantages and outcome expectations. Gender and age are theorized to have a moderating influence on performance expectancy. Besides gender and age, there are two other so‐called moderating variables applied in the UTAUT model, being experience and voluntariness. Venkatesh et al. (2003) state that performance expectancy is the strongest predictor of intention and remains in both voluntary and mandatory settings significant at all points.

The second construct, effort expectancy, defines the level of convenience when using the system. The compared models provide three constructs that contribute to effort expectancy: perceived ease of use, complexity and ease of use. Resemblances between these constructs’ definitions and measurement scales cause a valid convergence into this construct. Logically, the natural process of getting used to a system takes more effort at the beginning of the implementation. Therefore, this construct is relevant at the start, but its importance will lower steadily over time (Venkatesh, et al., 2003). The relation between the construct and behavioural intention will be influenced by the salient moderators gender, age, and experience (Grønland, 2010).

30 In UTAUT, the construct social influence is highly related to peer pressure. Individuals are directly or indirectly influenced, as relative others might believe that the person should utilize the system. Other models defined this phenomenon as subject norm, social factors or image (Venkatesh, et al., 2003; Grønland, 2010). All moderators influence the relation between social and behavioural intention. Like effort expectancy, social influence is highly significant at the early stages and its importance will fade away over time (Venkatesh, et al., 2003).

The last construct of the model is the facilitating conditions, a broad notion outlined in other models as perceived behavioural control, facilitating conditions and compatibility. The construct is clarified as the extent to which an individual believes that additional support is provided when using the system. According to Venkatesh et al. (2003), facilitating conditions has a direct relation with use behaviour unlike the other three constructs. This was tested based on perceived behavioural control, which is part of the constructs that emerged into facilitating conditions. Venkatesh et al. (2003) show that the relation between facilitating conditions and use behaviour only grows more significant with an increasing experience and age. This is due to the belief that when use behaviour of the system will be continued, a hindrance or foster by resources would directly influence the relation (Venkatesh, et al., 2003).

Some constructs of the existing models are not applied in UTAUT. One of those constructs is attitude toward using, which contains the overall reaction of individuals regarding the usage of the system. Venkatesh et al. (2003) state that attitude toward using is indeed an interesting construct. However, empirical evidence shows that attitude toward using is only applicable when performance and effort expectancies are not part of the model. This is due to the fact that variables might collide since they are already measured in UTAUT through other constructs like effort expectancy (Venkatesh, et al., 2003). To complete the model, behavioural intention is indentified to have a significant positive influence on technology use behaviour. 3.3.2 UTAUT: Statistical methods A partial least squares (PLS) methods was applied to determine reliability and validity of the measures in the model comparison. The minimum set on forehand was .70 and all consistency reliabilities exceeded that factor. The closer to 1, the more reliable the measures. Constructs like social influence could be measured at three different times, since the UTAUT model was developed in a longitudinal field study at four organizations. The received data of the 215 participants was pooled in order to obtain one overall result (Venkatesh, et al., 2003; Dutch Cochrane Centre (DCC), 2014). The sample size posed a limitation because of the large number of measurement items per variable. Therefore, only the highest four loading items were analysed using the pooled data.

31 3.4 Unified Theory of Acceptance and Use of Technology for Smart Environments 3.4.1 UTAUT‐SE: Model and construct definition

Gender Age

Effort Expectancy

Performance Behavioural Expectancy Intention

Social Influence

Personal Personal Importance Innovativeness Relevance in IT

Figure 3.4 Unified Theory of Acceptance and Use of Technology for Smart Environments (UTAUT‐SE) (Mayer, et al., 2011)

This model, developed by Mayer et al. (2011), is derived from UTAUT. As discussed in Section 3.3.1, UTAUT is mostly utilized in organizational structures, whereas this extension called UTAUT‐SE was developed especially for smart products in the Smart Home environment and for voluntary use. TAM was not suited for this research as it neglects the social context in technology adoption. Mayer et al. (2011) argue that computing devices that merge with the physical world, like ambient sensor networks, result in divergent perceptions of people in contrast to the more classical devices. These perceptions could even stress to an evoked fear for the technology control. This change in attitude and technology presents the main reason for UTAUT‐SE as depicted in Figure 3.4: the user acceptance of smart products in the home environment, since older models do not include this perception. Models concentrating on Smart Home environment acceptance are rather scarce (Mayer, et al., 2011). Vastenburg et al. (2007) state that in general, consumers have a positive attitude toward Smart Homes in which perceived ease of use is the most important factor. An insufficient level of control, privacy and usability are most important for the user acceptance of Smart Homes (Freudenthal & Mook, 2003).

Use behaviour and facilitating conditions are excluded from the model, because the research lacks a working prototype as a result of which use behaviour cannot be determined by observation. Like TAM and UTAUT, Mayer et al. (2011) argue that based on previously conducted research perceived usefulness is the strongest predictor of

32 behavioural intention. However, Mayer et al. (2011) also point out that the emotional response to the technology is of vital importance in predicting behavioural intention. Performance expectancy in UTAUT‐SE is also related to effort expectancy and social influence, due to the strong prediction factor. Voluntariness of use is excluded, since smart technology in a home environment is voluntarily, unless the residents’ medical condition requires extra monitoring through smart technology. Experience is excluded as well, because the study of Mayer et al. (2011) is not longitudinal and therefore no differences in experience can be evaluated over time. As smart products in this research were not physically available, actual use is not included in the model.

Mayer et al. (2011) also introduces three new moderators, importance, personal relevance and personal innovativeness in IT, because these are used to define involvement, which Mayer et al. (2011) believe to be an important aspect influencing technology user acceptance. Importance comprises the continuous desire for support when accomplishing an activity. Under personal relevance, a persons’ dedication and interest level in an application domain is evaluated. Furthermore, it also includes the general relevance of an activity to a user. The main difference between the importance and personal relevance is that when an activity is significant for a user, this does not automatically mean that the user wants help with it. The last moderator, personal innovativeness in domain of information technology, is an interesting moderator when the technology is new to people. Personal innovativeness will play an important role in an individual’s adoption behaviour. 3.4.2 UTAUT‐SE: Statistical methods Through an online questionnaire, which covered five clarifying application scenarios, 175 respondents participated. The online questionnaire was developed based on the existing technology acceptance scales, which were operationalized according to the different literature findings. The smart products that Mayer et al. (2011) discussed, like a Smart Kitchen, are not yet widely available, which is why they utilized a scenario‐based approach, consisting of both a textual and graphical explanation. Every scenario had the same questions, with minor adjustments to the smart product. A seven‐point Likert scale was utilized and the measurement items were reviewed by three industry and three academic experts.

T‐tests were applied to compare the scenarios in order to determine if the scenarios were rated in a consistent way. As this turned out to be, the scenarios could be further analysed as a whole. In order to test the research model, PLS was applied as a Structural Equation Modelling (SEM) technique (Mayer, et al., 2011).

33 3.5 Telecare Acceptance and Use Model 3.5.1 TAUM: Model and construct definitions

Designer Care receiver Caregiver

Needs & Benefits Use Dependence

Accessibility Personal Variables Facilitation Conditions Figure 3.5 Telecare Acceptance and Use Model (TAUM) (Sponselee, 2013)

TAUM, displayed in Figure 3.5, focuses on the user acceptance of Smart Home technology by elderly people. According to TAM and TRA, intention to use is the main determinant for actual use, in which actual use is referred to as the optimum acceptance. Sponselee et al. (2008) believe that technology will be perceived as useful, if technology supports the users’ goal. Whereas TAM is considered a measurement model for a short period of interaction, TAUM was developed as a model which focuses on long‐term interaction and actual user acceptance. Sponselee et al. (2008) criticize that TAM does not cover all relevant aspects to predict actual Smart Home technology acceptance and use. Other studies show that the user acceptance of a new technology is also influenced by costs, information supply, and experience (Sponselee, et al., 2008; Chan, et al., 2008; Alam, et al., 2012; Mayer, et al., 2011). TAUM was developed based on the concept of TAM, to get a better insight in why telecare technology or Smart Home technology has or has not been accepted in personal (care) situations at home.

The constructs of TAUM consist of needs & dependence, benefits and use. Firstly, the needs of the end‐user and the other stakeholders need to be established. Sponselee et al. (2013) expect that if the needs are thoroughly met, the usefulness will increase, resulting in an increase of the technology usage. Perceived usefulness is dependent on the personal needs and can be more specifically defined as effectiveness. The needs can vary from non‐ essential needs (comfort) to vitally important needs (necessary for survival) and they may change over time. As discussed in TAM, only usefulness is considered a key predictor for intention to use, whereas several others beg to differ and discuss the importance of job‐ fit, extrinsic motivation, relative advantage, outcome expectations, and performance expectancy as most important predictors for (intention to) use. Perceived usefulness is measured merely by executing a survey. Benefits, like sense of safety can be measured more specifically, but at the same time it is also further determined by more difficult aspects, like appreciation and evaluation (Sponselee, et al., 2008). There might be a discrepancy between subjective and expected benefits, so they are considered important to measure in this model.

Benefits may change over time, which is why TAUM shows an iterative process between benefits and use. This evaluation process and the change in benefits over time may lead to a changing conclusion, as previously considered benefits at the start might vanish when actually applying the technology. The process is influenced by the different stakeholders:

34 the designer, caregiver, and care receiver (Sponselee, et al., 2008). Other moderators of great importance are accessibility of the technology, comparable to the perceived ease of use in TAM, the personal variables of the end‐user, and the facilitating conditions, concerning the support in the technology usage. Social influence is not considered as a sole construct or moderator, but is seen as an element of indirect experience, which means that after usage an evaluation is formed. This may lead to the prior discussed change in benefits, since there will be decided if the technology is still accepted by the end‐user. Sponselee et al. (2008) believe that many external variables can be categorized under one or more of the factors in TAUM. Voluntariness, subject norm and social factors are not considered specifically within this study. This is due to the fact that a home setting is voluntary (Venkatesh & Davis, 2000), but also because Sponselee et al. (2013) do not expect a strong social influence on telecare acceptance and Smart Homes. Note that this model is only tested with elderly people and that it is a longitudinal study (Sponselee, et al., 2008; Sponselee, 2013).

3.5.2 TAUM: Statistical methods The study started with 81 people, ages varying from 51 to 90 years old. The participants filled in a survey at three different moments in time: at the start of the experiment, after one month of usage and after a year of usage. At the third measuring point, only 55 respondents participated, resulting in 35 useful cases. Due to this longitudinal study, the change in use and benefits over time can be taken into account.

The constructs are itemized and these measurement items are linked to different point‐ scales in the survey. All the items are first subjected to a reliability test, being Cronbach’s alpha. A broad variance of tests were used to examine the individual relationship of each effect. Eventually, the results show that using the telecare system provides beneficial effects in the area of independence and independent living. The benefits itself differ dependent on the needs of the participants. In total, significant links are displayed between needs, system functionalities, beneficial effects, and system usage. Sponselee et al. (2013) believe that this emerging TAUM model can provide valuable insight for Smart Home user acceptance for future studies. 3.6 Conclusion The evaluated models all show different aspects which are interesting and important for the user acceptance model applied within this thesis. However, some models are too complicated to use and/or do not entirely fit the purpose. For these reasons, a new model will be developed with constructs and moderators of UTAUT, TAUM, UTAUT‐SE, and TAM.

In the Smart Home environment, improving comfort is the underlying key factor affecting the user acceptance. In particular, three constructs come forward in the models to have a strong relation with intention to use or actual use. Perceived usefulness and perceived ease of use are important predictors of intention to use and occur in all four models, whether literally or combined with other constructs. When deciding about the perceived usefulness of the technology, costs and energy consumption considerations form important conditions. Furthermore, facilitating conditions are an important factor to consider in the new acceptance model as well, because the support that the system offers and the costs of that support have been stated to have a direct influence on the

35 actual use. It could be useful to know up front whether the Smart Home technology company has an adequate helpdesk, thereby influencing someone’s deliberation on buying the system. Especially, when facilitating conditions comprise more than costs, for instance compatibility, it is most likely that the construct has a direct relationship with intention to use rather than actual use.

The moderators of the new model will consist of personal characteristics, which influence the relationship between the independent and dependent variable(s) (Cohen, et al., 2003). As seen in the evaluated models, age, gender and experience can be responsible for different reactions towards the constructs and therefore toward the Smart Home technology. However, experience can be interpreted in different ways and therefore, this potential moderator will be split into two moderators, being social influence and technological knowledge. Social influence is highly relevant when a new system is presented, however Smart Homes already participate on the market for quite some decades now. The slow start has prevented them in gaining worldwide notice. Since social influence will decrease over time, this does not necessarily have to be tested as an individual construct. However, it does have an influence on perceived usefulness, so it should be taken in as a moderator. Technological knowledge is important to test a person’s affinity with Smart Home technology and the expected effort required to understand MyriaNed, which explains the influence on the constructs perceived ease of use and facilitating conditions.

The model as displayed in Figure 3.6 combines the previously discussed models. The constructs are coalesced with the key barriers of the Smart Home implementation: safety and security issues, privacy issues and affordability (Bal, et al., 2011; Chan, et al., 2009; Kim, et al., 2013; Abi‐Char, et al., 2010; Busnel & Giroux, 2010). Through this model, it will be tested whether user experience with Smart Home technology can enhance the user acceptance of Smart Homes, which is tested by intention to use. This study is not a longitudinal study in which actual use is measured, therefore the most important predictor of actual use, being intention to use, forms the dependent variable. Facilitating conditions contains more aspects and therefore it will lead to intention to use instead of actual use as describe in the other acceptance models. The specific clarification of this model will follow in Chapter 4.

Requisites

Perceived Usefulness

Perceived Ease Intention to Actual Use of Use use

Facilitating Conditions

Personal Characteristics

Technological Social Gender Age Knowledge Influence

Figure 3.6 The proposed user acceptance model for Smart Homes

36 4 From acceptance model to survey 4.1 Introduction Defining a model based on different studies can be helpful to create a framework through which a specific research can be conducted. Although constructs might seem similar to those presented by other models, the usability of the model for a specific purpose will be embodied by the accurate characterization of each individual variable. In this chapter, the specific constructs and moderators as displayed in Figure 3.6 will be further clarified. This model will eventually be utilized to test the user acceptance of Smart Home technology and the effect of experiencing this technology through a Virtual Reality Smart Home experiment. In Section 4.2, all the different constructs and moderators are defined. These definitions are important when developing a survey to set up the individual questions. The variables will be operationalized in order to develop two separate surveys, being a pre‐ and post‐survey, which will support the Smart Home experience. A survey is a commonly used tool within quantitative research, since the developer can obtain useful and extensive information from a lot of participants without requiring a considerable amount of money. Besides, surveys can be administered online, through email or social media, on paper, by phone or face‐to‐face. Since the experiment will be held at the Eindhoven University of Technology, paper surveys will be provided. As the diverging target group differs in age and nationality, both an English and a Dutch version need to be created. This has to be taken into account for the actual Virtual Reality program itself as well. By safeguarding the privacy of the respondents as participating is anonymous, it is adopted that participants feel more liberated to answers honestly. Through this, it is more likely that accurate data will be obtained. To obtain uniformity, it would be best to formulate a limited amount of open‐ended questions in contrast to the closed‐ended questions. Therefore, the majority of answer possibilities of the questions within the surveys are based on a Likert scale, in which choices are limited. The surveys are presented in Appendix D and E. Section 4.3 will provide the necessary support to accompany the elements applied within the surveys. Besides construct‐ and moderator‐ related questions, there are questions included concerning the experience itself, the tools applied and the observations made by the researchers. This all will conclude in Section 4.4. Eventually, the surveys will generate a lot of data that can be used within the further outline of this thesis.

The main sub‐questions central to this chapter are:

 How can the constructs and moderators of the user acceptance model be operationalized for the survey?  How can the Virtual Reality experience and observations be included in the survey?

37 4.2 Defining and operationalizing the model Although some of the models discussed in Chapter 3 are developed for organizational structures, the various constructs applied can be used in other settings as well (Venkatesh, et al., 2003). Since performance expectancy and effort expectancy are highly associated with perceived usefulness and perceived ease of use in UTAUT, it seems only one of both combinations can be suitable for testing user acceptance of Smart Homes. Since the applied user acceptance model concerns Smart Homes, the denomination of perceived usefulness and perceived ease of use characterise the situation more adequately compared to performance expectancy and effort expectancy. Performance is highly related to job qualifications and since these terms cover the same items, the first set of constructs is more appropriate.

Table 4.1 provides an overview of the construct and moderator definitions applied within the user acceptance model. Perceived usefulness is measured by a person's belief that using the Smart Home technology enhances the home comfort and therefore indirectly contributes to an increased quality of life. Due to the fact that quality of life comprises more aspects and increasing comfort is directly linked to Smart Home objectives, increasing comfort in the home environment perfectly presents the overall goal.

Although the sensor network that will be implemented in Smart Homes is wireless and human interference is hardly necessary, the perceived part of perceived ease of use is of great importance. The construct is defined as the degree to which a person believes that using the technology would be free of effort. It is about how the user perceives the Smart Home technology, which in the case of the experiment is MyriaNed. When WSNs work automatically after being programmed correctly, the system will not be used to such an extent that the technology itself needs to be adjusted constantly. However, when changing conditions provide changing needs, interaction with the technology might be high. Therefore, the perceived ease of use is vital to take into consideration.

The final independent variable consists of facilitating conditions. In the case of UTAUT, facilitating conditions are directly linked to use behaviour. However, the facilitating conditions within this model comprised more elements that are not only applied when the technology is actually used. Whereas guidance and support when using the system are still part of the construct, other elements like affordability of the support and compatibility of the system are included as well. Since these considerations are vital when implementing Smart Homes, these factors already contribute to a person´s thinking process when purchasing the system. In addition, if someone already possesses a certain smart technology or wireless sensor network, it would be best if the new technology is compatible with the ones that are already there. These factors added to the facilitating conditions construct are necessary ingredients for Smart Home acceptance. Therefore, the hypothetical relation between this construct and intention to use will be analysed.

The personal characteristics in the model of Figure 3.6 are gender, age, technological knowledge and social influence. Gender and age are highly important within this study, since elderly might have a different attitude towards the technology, as highlighted by Sponselee et al. (2013). It is common sense that in general the attitude of elderly toward applying and adopting a new technology or systems is more resistant compared to others. Unless the perceived usefulness and ease of use are high, there might arise unforeseen

38 problems. Likewise, the attitude of men and women toward new technology are expected to differ significantly, according to other studies (Venkatesh & Morris, 2000; Mayer, et al., 2011; Grønland, 2010). Therefore, these moderator factors can influence the relations between constructs as presented in Figure 3.6.

As explained in Subsection 3.5.1, experience is essential when considering the user acceptance of Smart Home technology (Gardner & Amoroso, 2004; Lovecraft, 1973). Two types of preliminary experience with innovative technology are implemented in the user acceptance model, being technological knowledge and social influence. Technological knowledge focuses more on the personal knowledge of the participants and social influence on knowledge gained through others. As highlighted in Section 3.6, in the case of Smart Homes social influence will be included as a moderator. It is necessary to determine whether social influence has an effect on the relationship between perceived usefulness and intention to use. Technological knowledge consists of the experience and knowledge the participants have of the Smart Home technologies and their attitude towards these technologies. It is adopted that different knowledge levels can influence the relationship between perceived ease of use and intention to use, and between facilitating conditions and intention to use. As discussed by Mayer, et al. (2011) voluntariness of use is more of an issue with technology systems that are implemented in an organizational environment and not by implementation in Smart Homes, where users can choose whether they want to install and use the technology or not. The personal relevance as initiated in TAUM is covered within perceived usefulness, as something useful has relevance for an individual.

Eventually, all the independent variables and moderators are utilized to define the dependent variable intention to use. Based on the other models discussed, it is believed that intention to use is causally related to actual use. This means that an enhanced user acceptance of Smart Homes leads to an increased implementation of this dwelling type. Due to the underlying thought of this thesis, this implementation is presented in the model, although the experiment will only enclose the constructs up till intention to use. However, based on that potential causal relationship between the two, the outcome of the model can be used to see how actual use can be enhanced. This will be further elaborated in Chapter 8.

Table 4.1 Defining the constructs and moderators of the user acceptance model

Constructs & Moderators Definition Perceived Usefulness Degree to which a person believes that using Smart Home technology would enhance his or her comfort. Perceived Ease of Use Degree to which a person believes that using Smart Home technology would be free of effort Facilitating Conditions Degree to which an individual believes a network exists to support use of the Smart Home technology and believes that objective factors in the environment make Smart Home technology easier to accept Intention to use Degree to which an individual believes that he or she is intended to use Smart Home technology Age How old a person is Gender Whether the person is a male or female Technological Knowledge Degree to which a person already encountered Smart Home technology or another similar system by themselves Social Influence Degree to which a person encountered Smart Home technology through others

39 Now that the individual constructs are defined, the entire model should be further operationalized before it is possible to set up the surveys that are part of the Virtual Reality experiment (Venkatesh, et al., 2003). Eventually, each individual construct should be represented within the surveys in order to test the entire user acceptance model. By testing intention to use, not only should there be questions included about the independent variables perceived usefulness, perceived ease of use and facilitating conditions, but also about the dependent variable intention to use itself. As actual use is not explicitly part of the experiment, no questions will concern that construct.

Although MyriaNed will be used as Smart Home technology, it would not be wise to address the sensor protocol by name within the surveys. Some respondents might not even be familiar with Smart Home technology, let alone MyriaNed. Furthermore, even those who are familiar with sensor technology probably only know the more known protocols like ZigBee, Z‐Wave or X10. Therefore, naming MyriaNed would only cause unnecessary confusion. At the end, within the experiment it is about the appliances of the technology instead of the technology itself. Some of the definitions as formulated in Table 4.1 are not only affiliated to Smart Home technology, but also to ‘normal’ technology or other innovative technology. When translating the constructs and moderators into questions, it is necessary to define those technologies at forehand in order to clarify each technology category. This can be done by adding an accompanying page as introduction and reminder at the first page of both surveys. The goal of this thesis is not explained anywhere within the surveys, since this could only lead to social desirable answers.

Besides the constructs, the four moderators affecting the relations need to be operationalized, since they will be included in the surveys as well. Whereas age and gender are rather straightforward, the other two are more complex. For the sake of analysing the surveys, it is vital to represent all construct in both a pre‐ and post‐survey. In the pre‐survey, personal information is gathered about the respondents before experiencing the virtual Smart Home. After experiencing, a post‐survey will provide similar questions. Through this, it is possible to distinguish whether experiencing the Virtual Reality Smart Home has influenced respondents’ opinion. The results will be compared to see if changes occur. Whether these changes are a direct consequence of the experiment is hard to tell, since other factors may influence this process. However, when both pre‐ and post‐survey are formulated similarly, the entire experiment is conducted under the same circumstances each time, and no extra technology information is given during the experiment, it is more likely to say the potential change is due to the experiment (CCV, 2011; van Vliet, 1973; Baarda & De Goede, 2006). Furthermore, the respondents will perform every part of the experiment individually in a sequence of events, ensuring no external contact is made. As the importance of technical knowledge and social influence are explained through the user acceptance model, it would be hardly impossible to create two identical groups with similar knowledge levels, social circumstances, ages and gender. Consequently, there will be only one group conducting the experiment (Baarda & De Goede, 2006; CCV, 2011). As explained, the experiment will consist of a pre‐survey, directly followed by experiencing a virtual Smart Home, and finally completing a post‐survey.

The moderators only need to be included in the pre‐survey, as these concern general information about the personal characteristics of the respondents. This information will

40 not change over this short period of time and therefore only have to be asked once. Since the medium of the experiment is rather new in this line of research, it would be interesting to see what reactions respondents give to the Virtual Reality setting and equipment. In order to get people’s opinion on the Virtual Reality and the effect of the experiment, some additional questions should be added to the post‐survey. To know how people react to the VR experience really gives insight on whether this medium could be used on future occasions for Smart Homes and in other (similar) areas. Besides the opinion of the respondents, it would also be useful to observe peoples’ reactions and need for guidance in relation to the Smart Home technology and the Virtual Reality controllers. In order to operationalize these experiences and observations for the survey, both aspects need to be defined. The Virtual Reality experience is defined by the effect of the experiment on people’s opinions, and the observations are based on how people respond to the experiment and the controllers. 4.3 Survey questions, measurement scales and sample size To finalize the surveys, the definitions as presented in Section 4.2 are converted into questions. In order to stimulate the progress of the experiment, it is important that the surveys are not too exhaustive and tedious. To create a coherent experiment in which the three elements come together, the number of questions per survey should be limited. In that light, creating a rather similar pre‐ and post‐survey could provide the necessary stability and structure to the experiment. At the end, the surveys are the underlying elements supporting the experience. The number of questions evolving from the definitions of Section 4.2 should be achievable and feasible, without compromising on quality or thoroughness. The pre‐survey is presented in Appendix D and the post‐survey in Appendix E. In the general section of the pre‐survey of Appendix D, age and gender are established through singular questions. The definitions as formulated in Table 4.1 form the guidelines for the preparation of the moderator questions of technical knowledge and social influence. In order to measure the construct perceived usefulness, eight questions are formulated based on the definition of Table 4.1. Similarly, ease of use and facilitating condition are respectively measured through five and three questions. The three questions covering the dependent construct of intention to use finalize the pre‐survey. The additional page provided at the start of both the pre‐and post‐survey provides necessary information as discussed in Section 4.2. At the end of the pre‐survey, some introductory information is given to prepare the participant for the VR experience.

The accompanying information at the beginning provides guidance when filling in the surveys. Since participating is voluntarily and anonymous, it is intended to lower the possibility of receiving socially acceptable and desirable answers. The distinction between ‘normal’ technology, innovative technology and Smart Home technology at the first page makes sure that no interpretation differences will occur when filling in the surveys. Therefore, as a reminder, this is repeated in the post‐survey of Appendix E. In the post‐ survey, Virtual Reality experience and observation questions are included as discussed in Section 4.2. Obviously, the observant page will not be included in the post‐survey of the respondents. During the virtual experience, observations are made by the authors of this thesis. Table 4.2 provides an overview of the questions admitted within the surveys.

41 Table 4.2 Overview of number and actual questions per construct and moderator

Constructs & Moderators # Pre # Post Survey questions / statements Age 1 ‐  What is your age? Gender 2 ‐  What gender are you? Technological knowledge 3 ‐  Are you familiar with the dwelling type Smart Home? 4  Do you use (innovative) technology? Social influence 5 ‐  Do you know any households that use Smart Home technology? 6 ‐  In the last two years, did you buy any innovative technology like a tablet, smart television, smart meter or smart sensor network? 7 ‐  Have you ever thought about purchasing Smart Home technology, like a smart television or a wireless sensor network to control lightning? Perceived usefulness 8 1  Do you think that applying Smart Home technology could provide more spare time? 9 2  Do you have any privacy concerns regarding Smart Home technology? 10 3  Do you have any security concerns regarding Smart Home technology? 11 4  Do you have any cost concerns regarding Smart Home technology? 12 5  Do you have any energy concerns regarding Smart Home technology? 13 6  Would perceivable benefits like conveniences or financial benefits overthrow your concerns as answered in questions 9 until 12 / 2 until 5? 14 7  Do you think that using Smart Home technology can enhance your comfort within your home environment? 15 8  Overall, do you think Smart Home technology is useful for your home environment? Perceived ease of use 16 9  I believe that I can understand Smart Home technology easily. 17 10  I would say I easily adapt to using Smart Home technology. 18 11  I like to explore Smart Home technology. 19 12  It would be easy for me to become skilled in Smart Home technology. 20 13  Overall, I think that Smart Home technology would be easy to use in my home environment. Facilitating conditions 21 14  I believe additional support is necessary to enhance my attitude towards using Smart Home technology. 22 15  Free technical support with Smart Home technology would increase my interest. 23 16  Smart Home technology should be compatible with other technology I use in my home environment. Intention to use 24 17  I would like to know more about Smart Home technologies. 25 18  Would you like to have Smart Home technology installed in your home environment? 26 19  Would you consider purchasing Smart Home technology in the near future? Virtual Reality experience ‐ 20  Did the Virtual Reality experience help you in understanding what Smart Home technology could mean for your home environment? ‐ 21  Did the Virtual Reality experience influence your opinion about implementing Smart Home technology?

42 ‐ 22  After experiencing, I have a better understanding of Smart Homes. ‐ 23  How did the Smart Home technology make you feel during the experience. ‐ 24  Do you think that using Smart Home technology can provide you energy conservations or other financial benefits? ‐ 25  Experiencing Smart Home technology changed my attitude towards Smart Homes. Observations ‐ ‐  Did the participant need help or guidance in the experiment regarding the controllers? ‐ ‐  Did the participant need help or guidance in the experiment regarding the Smart Home technology? ‐ ‐  How does the participant respond to the Smart Home technology during the experiment? ‐ ‐  How does the participant respond to the controllers during the experiment? As discussed in Section 4.1, both the pre‐ and the post‐survey are translated into Dutch, in order to reach as many respondents as possible. The answer possibilities for each question are drawn up by using the Likert Scale for measurement. A Likert scale is commonly used within surveys to measure knowledge, influence, perception, value, attitude or behavioral change (Likert, 1932). In this case, a five‐point scale has been selected, since it will provide a decent spread in options for each question. A seven‐point scale might be more accurate, but at the same time it could also provide confusion and inconvenience and therefore a five‐point scale might be easier to understand (DeVellis, 2012). In total, six different Likert scales are applied. This is due to the fact that some of the answer possibilities had to be customized for a particular question. Table 4.3 displays the question numbers and their corresponding Likert scales. A preliminary test amongst potential participants in different age, gender, knowledge and social groups provided valuable insights in the intelligibility, comprehensibility and answer coherence of the survey in order to see whether the surveys are clear, unambiguous and understandable.

Table 4.3 Likert scale per questions in pre‐ and post‐survey

# Pre‐questions # Post‐questions Likert scale 3,9 ‐ 12 2 ‐ 5 Not at all ‐ Slightly ‐ Somewhat ‐ Moderately ‐ Extremely 5,6 ‐ None ‐ One or two ‐ Three ‐ Four ‐ Five or more 7 ‐ Not at all ‐ Occasionally ‐ Sometimes ‐ Often ‐ Very often 8,13 ‐ 15,25,26 1,6 ‐ 8,18,19,24 Not at all ‐ Not likely ‐ Undecided ‐ Most likely ‐ Definitely 16 ‐ 24 9 ‐ 17 Fully disagree – Disagree – Undecided ‐ Agree ‐ Fully agree ‐ 20 ‐ 22, 25 Not at all ‐ A little bit ‐ Sometimes ‐Considerably ‐ Definitely Ultimately, it is essential to have an indication of the minimum sample size that is necessary to gain useful and representative results out of the experiment. Since Smart Homes can provide valuable and diverse advantages for every individual, a heterogeneous sample is desired (Baarda & De Goede, 2006). Elderly might respond differently to questions concerning ease of use compared to younger people. Likewise, Apple users might attach more value to compatibility compared to other technology users. The required accuracy does not have to be extremely high, since results from the random sample size only provide a valuable image of the entire experiment. The results have to be representative for the entire population. The experiment is rather complex, since it includes an experience as well as two surveys. Normally, experiments require a smaller sample size compared to survey‐based conducted research (Baarda & De Goede,

43 2006). Since the experiment is a mix of both, the sample size will be an average of those two as well. Due to the potential number of respondents and time available, the sample size will be chosen on pragmatic grounds. Absolute minimum number of respondents in the most simple scenario needs to be 30 participants, but since the case of this thesis is more complex, at least 100 respondents are necessary to obtain an adequate sample size (Baarda & De Goede, 2006). 4.4 Conclusion In this chapter, a pre‐ and post‐survey are drawn up based on the user acceptance model obtained from Chapter 3. By defining the individual constructs and moderators, the model can be translated into different surveys for the experiment. Through this, the constructs perceived usefulness, perceived ease of use, facilitating conditions and intention to use can be operationalized and valued. Since moderators are seen as elements influencing relationships, information has to be gathered about these general characteristics. This can be achieved by including the variables age, gender, technological knowledge and social influence in the pre‐survey. Besides the questions that are highly relevant for analysing the user acceptance model, it is useful to include questions concerning the Virtual Reality experience. The setup is rather new in this line of research and therefore it is interesting to see how people react toward and feel about the technology. In order to identify how people react to the technology, observations are vital. These observations will be registered through four basic questions in which reactions and guidance dependencies are recorded. In order to safeguard the readability of the surveys, MyriaNed as a protocol name will not be included.

Since the sample of respondents will consist of both foreign and Dutch participants, the surveys need to be translated in both English and Dutch. Ideally, the sample needs to be as divergent as possible, in order to capture all groups of society. Therefore, it is also important to develop an easy‐to‐understand survey, which can only be interpreted in one way. When translating, these conditions need to be safeguarded, in order to prevent discrepancies between the versions. To create representative outcomes, the participants need to perform the experiment on their own and no interaction between respondents during the experiment must occur. In order to provide valuable results that can be interpreted further outside the scope of this research, it is vital that the sample size of the target group is sufficiently large. Practically, this means that at least 100 respondents need to participate within this research and complete the experiment. Ideally, this group of respondents varies proportionally in age, gender, knowledge level and social coherence. The final element necessary to conduct the experiment consists of the Virtual Reality experience. The entire layout of the experience will be attended to in Chapter 5.

44 5 The Virtual Reality Smart Home experience 5.1 Introduction For people to really grasp the idea of a Smart Home, experiencing such a home can provide valuable assets in order to form a personal opinion. As highlighted in the scientific relevance of Subsection 1.2.2, there is limited research on user acceptance of Smart Homes. The only methods applied to validate a model are surveys or longitudinal research. Only few concentrate on letting users experience the technology and research if they would be interested in the technology. Within the Netherlands, there are a couple of prototype Smart Homes open to the public. In these homes, people get the opportunity to meet high‐tech technologies and supportive systems (Smart Homes, 2015). Using and altering these homes for scientific research would be impossible, not only based on costs but also on time, materials and knowledge. In the same ideology, creating a Living Lab could provide benefits without any doubt, however identical problems occur. Shifting the experimental surrounding into a virtual world, in which the same realistic context is safeguarded, can eliminate almost all potential obstacles. However, the knowledge about the technologies and systems implemented is still necessary. In this light, the Smart Home concept and MyriaNed sensor technology are elaborated within the first three chapters of this thesis. As the distinction between Virtual Reality and reality fades away, results obtained in the virtual world become interesting and relevant to such a degree that scientific research can really benefit from these innovations (Bergvall‐Kåreborn, et al., 2009). In order to develop a realistic setting, it would be better to create a homogenous image of the entire context. In the case of augmented or mixed reality, real life images are interspersed with virtual images, creating a changing view in which the virtual setting will irrevocably be compared to the real images (Ott, et al., 2007; Fahn, et al., 2013). This would impair and undermine the setting as a whole and therefore solely using Virtual Reality is most adequate.

In order to create an optimal experience, a Virtual Reality Smart Home has to be designed. When developing the Smart Home, it is important to keep in mind the overall purpose of the virtual context. This will form a guideline for the technological implementation of MyriaNed in a number of activities, which will be explained in Section 5.2. Developing the model itself requires knowledge of many different software programs, equipment and tools in which a virtual Smart Home is built from scratch. The entire process of designing the Smart Home will be described and depicted in Section 5.3. Eventually, the participants will navigate through the environment. Although moving freely, unwarily they will follow a preset route along certain activities. Further elaboration on this Virtual Reality tour will be discussed in Section 5.4. After modelling the context, interactions need to be implemented in order to create the real‐life setting in which automatic technology will provide comfort. Besides modelling, it is necessary for this purpose to program different elements and interactions inside the house. Within that stage, the entire model has to be transferred into the Virtual Reality world by utilizing a Powerwall, 3D glasses and a wand. As highlighted by WorldViz (2015), there are a couple of advantages attached to using a Powerwall. Besides costs, the image quality achieved through the Powerwall is relatively high. Especially, when considered only a small physical system footprint is required. In terms of ease of use, the light 3D glasses feel comfortable and no additional wires are necessary. However, besides the pros there are some cons

45 when applying the Powerwall. Due to the fact that participants are standing in front of a large screen, the sense of immersion is limited in comparison to for instance the Oculus Rift. An Oculus Rift is a more or less extended version of 3D glasses in which the Virtual Reality setting is immersed inside a headset and therefore displayed right in front of a person’s eyes. This will widen the field of vision in a more realistic way and no additional Powerwall is required. Although some prototype models are already on the market, the Oculus Rift will be released in 2016 (Oculus VR, 2015). Another disadvantage of the Powerwall is that the equipment is sensitive to ambient light, as a result of which the experimental area must be darkened at all times. Besides, only one person can participate and interact with the virtual Smart Home at the time, since only one headgear can be tracked by the cameras involved (WorldViz, 2015). However, the latter is not a limitation for this thesis, due to the fact that participants conduct the experiment individually to reduce external influences. In order to get an understanding of how to program the activities, Section 5.5 provides a piece of code used and elaborates more on that subject, after which a conclusion will follow in Section 5.6. Appendix F up to I provide valuable information to visually support this chapter. Table 5.1 depicts the entire process as adopted to develop the Virtual Reality experience and the different software used.

Table 5.1 Process of creating a Smart Home in Virtual Reality

Determine Export Smart Export Smart Program in DevelopmentCreate Smart Design Virtual Reality Interactions appliances for Home to 3ds Home to Vizard with Home in Revit Smart Home Max WorldViz Python

The main sub‐questions that are central to this chapter are:

 What is necessary to set up a realistic Virtual Reality program?  How will MyriaNed be implemented within the virtual Smart Home?  How is the virtual Smart Home designed to create a realistic setting?

46 5.2 Technological implementation and activities After finishing the pre‐survey, participants need to get an overall idea of the technologies involved by experiencing a Smart Home. Therefore, every element inside the virtual Smart Home needs to fulfil that purpose. The virtual experience will be enhanced when the respondents have to perform activities that refer to daily routines. In order to do so, people need to navigate through the environment. There are many ways in which people can be navigated through a model. Navigation can be done automatically by transporting the person from one room into another, or by jumping from one room into another, up to an environment in which the respondents can move freely according to their own preferences. By jumping through the model, the focus can be placed on the activities that need to be performed instead of the walking and navigating. The degree of freedom is very limited and it is doubted whether participants will gain insight in the close to real‐life conditions as displaced within the model. Virtual Reality can only be used as a realistic tool when realistic conditions are included. Therefore, walking freely would be most appropriate, however this would be most challenging when modelling and programming the Smart Home. Keeping the overall purpose in mind, compromising on certain aspects would really downgrade the realism of the entire model and experiment. By voice‐guiding the respondents through the program, there will still be a large freedom of movement while at the same time all respondents will perform the same sequence of routines. This inventive method will not compromise on the experience of walking freely through the Smart Home. As also older people will participate, a Dutch and an English version are required. Electronic voice guidance is provided by Acapela text to speech, which represented the most realistic, fluent and natural voice, in order to guide the participants smoothly through the program (Acapela group, 2015).

In Chapter 2, it is defined what type of MyriaNed sensors can be applied in order for the experience to be realistic. In addition, it is relevant to distinguish which appliances are implemented in the Smart Home. Although possibilities are endless when applying Smart Home technology, it would be best to confine the number of options. The MyriaNed appliances can be categorized into certain focus areas that will be highlighted during the experience. Obviously, comfort needs to be the overall theme since increasing comfort is the main goal within the concept of Smart Homes. Enhancing the comfort can be achieved through many different aspects, like luxury, convenience, security and energy conservations (Alam, et al., 2012; Aldrich, 2003). MyriaNed can provide valuable appliances for any of these areas (van Mierlo & van Lieshout, 2015). Some ground rules are set up in order to decide which of the endless possibility with MyriaNed should be displayed in the program. Criteria are that it needs to be realistic and it has to visualize one of the four aspects discussed. Furthermore, it is important that the appliances show the broad variety of possibilities in order to encourage the participants to think about how Smart Home technology would suit their preferences and home environment. Appendix F shows technological possibilities that should be embraced by the Smart Home activities in order to address all themes and to create a broad variety of appliances in the home environment. Furthermore, the corresponding sensor types are displayed in order to accomplish the activities (van Mierlo & van Lieshout, 2015).

Appendix F defines possibilities that can be included within the activities that need to be formulated. In order to create the feeling of convenience, security, luxury and energy

47 savings in the experience, the daily activities need to be as logic and natural as possible. Therefore, the following activities are selected:

 Kitchen activity: Sensors embedded in the kitchen area can provide multiple solutions in the fields of convenience and energy saving. Primarily, residents attend to their kitchen when they need something out of the refrigerator or when they need to perform a certain activity on the countertop, like doing the dishes, preparing diner, washing their hands, or putting away food, cutlery or crockery. In order to support any of these activities, sensors can be implemented to control fridge and lighting activities. Depending on the manoeuvre, light can be adjusted accordingly. Cutting an apple requires different lighting when compared to cleaning the counter. By applying a sensor‐equipped cutting board, it is assumed that when an object is placed on the board and its pressure sensors, something needs to be cut. This activity requests a focus light to enlighten the working area and to foster the working conditions. Besides supporting light, energy conservations can be achieved through simple, but useful automations. It is estimated that where lighting takes up to 17% of the energy use, fridges cause 8% of the energy bill (Layton, 2009; HOMEMAKERSonline, 2012). Sensors can be applied to eliminate inattentiveness when closing the fridge. When the fridge stays unattended for a short period of time, it will close itself automatically, based on proximity. Nowadays, some freezers already have an sound alarm when this happens, but automatic closure is possible as well. By using a motion sensor within the fridge door, convenience as well as energy savings can be obtained.

 Living room activity: Entering or exiting the kitchen or living room puts certain actions in motion in order to provide convenience and energy conservations. Through motion and PIR sensors, presence is detected and based on that attendance, the heating and lighting are turned on. This lowers the energy consumptions in a huge way, since heating based on user presence will make sure that heating and cooling will only occur when it is actually required. When a resident is leaving the room to go upstairs for half an hour, no energy will be wasted downstairs. In order to provide extra luxury in the Smart Home, the heat is radiated through the floors. By adjusting the blinds in a particular room, a motion detection does not necessarily has to mean that lights need to be turned on. Sun light will enter, providing enough luminosity to brighten the entire room by day. Based on the light sensors that are attached to the windows, it is decided whether the blinds need to be raised or the lights in the living room need to be turned on. The actual activity within this room is provided through accessing a Smart Wall. Through the Smart Wall, all kinds of entertainment and work‐related activities can be launched. In fact, the wall can replace an ordinary television as well as a computer. However, this depends on the ease of use and the desire to use a variety of applications through a single medium. By showing its potential, participants are triggered to define their own preferences and to understand the possibilities of support. Since no keyboard will be provided, voice guidance will be applied to navigate through the different television channels or Internet pages. In order to launch the Smart Wall, pressure sensors are inserted within certain areas of the couch that will activate the medium.

48  Outside activity: In order to provide convenience and security for residents of the Smart Home, motion and PIR sensors are placed strategically throughout the house. Based on presence through body heat, the wireless sensor network can indicate whether there are residents present per room or within the entire house. When leaving the house temporary, for instance by going into the backyard, the same sequence of actions will occur as when someone walks from one room to another. Whenever a resident is walking outside the proximity of for instance the living room, all automatic modifications are tuned back to normal. This normal state indicates that the Smart Wall will shut down, alongside heating and lighting activities. However, when residents leave the house for a longer period of time, for instance to go to work, visiting friends or go jogging, some additional security measurements will be activated. Electric devices and the central heating will be shut down, windows that are left open will be closed, and the doors will be locked. This can be achieved through a non‐contact, sensor‐embedded key in combination with the presence of the resident, which is detected by the entrance door.

 Bedroom activity: Although some activities might be minor, it is not about the activities itself, but about the entire experience achieved through these activities. In the bedroom, light will turn on automatically, since human presence is detected by the sensors embedded in the room. Whether this light is sun light coming from outside by raising the blinds or by turning on the artificial lights, is determined by the light sensors on the windows. As elaborated in Section 2.9, cameras are prohibited as it will undermine the privacy of the residents. Especially in private areas, it is key to safeguard the privacy by retrieving only the necessary information in order to provide support. A motion sensor is implemented to detect the movement and the PIR sensor makes sure that the light stays on through sensing the body heat. Again, dependent on the preferences of the resident, light can be altered or turned off when requested, either through a handclap or by a diminished proximity field around the bed.

 Bathroom activity: People attend their bathroom repeatedly on a daily basis, performing multiple activities. Although sensors and their applications can provide a wide range of conveniences, it is chosen for the bathroom to focus on the more luxurious appliances. In order to relax while taking a shower or bath, certain actions can be set into motion when touching the pressure sensors attached to the water tap. When the water is rising in the bath tub, light and humidity sensors make sure that lights will alter to preset colours and relaxing background music will be played. Again, the artificial lights are only triggered when the light sensors on the armatures indicate that an insufficient amount of light is entering through the windows. The music played can be a pre‐set radio frequency or tune transmitted through one or two speakers in the bathroom. While normally an increase in humidity results in a fogged mirror, a humidity sensor can prevent the mirror from fogging. Furthermore, whenever the resident is finished taking a bath or shower, the humidity will return to normal, when can occur mechanically as well as naturally by automatically opening a window. Through adequate ventilation, mould growth and other unwanted bacteria can

49 be prevented. Besides the humidity, also the colours and music will be turned off as the activity has finished.

By performing those activities singlehanded, the participants will see how each activity can become smart through wireless sensor network appliances. As respondents will not see the small sensors but only note their appliances, participants can experience that through the invisible and wireless implementation their familiar environment will not be disturbed, although benefits and additional comfort will be realized. 5.3 Smart Home design In order to provide the participants with a Virtual Reality Smart Home experience, the environment in which the technology is implemented is of great importance. The Smart Home is developed in Autodesk Revit 2015 and further adjusted and furnished in 3ds Max Design 2015. It was necessary to export the Revit model to 3ds Max, since a direct translation to the Vizard Virtual Reality Software Toolkit is not possible. Besides, 3ds Max has certain benefits over Revit regarding its modelling, texturing and lighting features. Several design decisions were made during the development of the model. Realism is of major importance as it will be vital for the experience of the participant. Furthermore, the setting needs to bring the technology experience upfront and it should show realistic activities as explained in Section 5.2. In order to focus on the technology, the layout of the Smart Home is kept modern and neutral.

Figure 5.1 presents the layout of the Smart Home as designed within the Autodesk programs. The home encompasses two bedrooms, one bathroom, a combined kitchen and living room, as well as a long hallway, a simple front yard and a more advanced backyard. The entire home is enclosed by a hedge to represent a normal urban area. Appendix G show an impression of the outside (Appendix G.1) and the inside (Appendix G.2) of the Smart Home.

Figure 5.1 Floor plan of the Smart Home with two bedrooms (upper right), a bathroom (upper left), a kitchen (lower left), and a living room (lower right)

50 Figure 5.2 Green public areas, red private areas and yellow semi‐private areas. The Smart Home has an L‐shape, in which there is a clear distinction between the more public (kitchen, living room and outside), private (bedrooms) and the semi‐private areas (bathroom), as shown in Figure 5.2. The emphasis will lie on the public areas, due to the fact that the largest activities mentioned in Section 5.2 occur in these room. As stated by Allameh et al. (2014), more and more of the daily life will take place in the public zones of a Smart Home instead of the private zones. The outside of the Smart Home has a modern architectural design, without distracting features. Appendix G.1 shows that the windows in the private rooms are smaller compared to the ones in the public rooms, which all have adjustable blinds. The living room is separated from the backyard by a large glass wall, creating lots of natural lighting in the room. The outer space is kept open and is modelled as a terrace, with a lounge set and a barbeque area. The garden is enclosed by a hedge, to both create privacy and to facilitate a more easy to model and realistic surrounding.

The rooms are accordingly designed to suit the activities and to let the participants feel like home. The hallway is small and entails a painting on the wall, a coat stand, and a shoe cabinet as shown in Figure 5.3. This hallway symbols the beginning and the end of the interaction activities in the Smart Home. In order to create a domestic feeling a stairway is designed. However, in order to control the route of the participants, the stairway is not

51 designed to enter as it is an optical illusion. The kitchen has an open space layout just like the living room. Since these are the first two activity rooms that the participants will enter, it is important that they have enough space to get adjusted to the controllers in order to perform the activities. Some furniture is modelled like paintings and plants to create liveliness and to Figure 5.3 The hallway with the falls stairs contribute to the domestic feeling. All are placed at the sides of the room to preserve the open space. The kitchen is quite large and probably will attract direct attention as seen in Figure 5.4. This is even more encouraged by the participants’ point of view when entering the room and it ensures a naturally route. A lot of walking is involved in the sequence of kitchen activities, again to get adjusted to the controllers, as will be shown in the walking route of Figure 5.10. After the kitchen, the living room activity is next, in which the same design principles are applied as the kitchen. However, within this room less walking is involved in order to motivate the Figure 5.4 The view of entrance of the kitchen participants to focus on the Smart Wall activities, where they have to make real body movements in order to complete the activities like bending through the knees and talking. The living room is also developed to create the comfortable ambiance of a home, as seen in Figure 5.5. The garden represents the transition to the more closed and private parts of the home. Sliding doors are utilized to navigate the participants in and out the garden on both sides of the house. Besides the lively terrace, for the most part the garden consists of grassland. The hallway and both the bedrooms have a more sober and plain appearance and are less extendedly furnished to create the private atmosphere. A representation of the identical bedrooms is shown in Figure Figure 5.5 The living room with the large windows 5.6. The bathroom is situated at the end of the hallway, another transition between public, private and semi‐private as shown in Figure 5.2. It is extensively designed to contribute to the atmosphere of the activity that takes place there. Besides a bathtub, a shower, cabinet, washbasin, toilet, spotlights, and a mirror are modelled within this area. The wall of the bathroom contains realistic and colourful tiles to enlighten the room. This is depicted in Figure 5.7. Next to the interior, a rather simple exterior is designed to support the domestic and urban feeling. Outside in both the back and front yard the security activities take place.

52 Figure 5.6 One of the two identical bedrooms Figure 5.7 The bathroom with the wall tiles In order to support the activities, sensors are placed according to the characteristics of MyriaNed throughout the Smart Home. Appendix H indicates the exact position of each and every sensor. For the purpose of the figure, the sensors are divided between object and room sensors, indicated by a coloured square or pentagon. As suggested by the name, the room sensors operate for the entire room. For instance, when taking the living room, the PIR, motion and temperature sensor detect measurands for the entire room. On the contrary, an object sensor only works for a specific object, like the pressure sensors within the headrest of the couch or the motion sensor of the sliding doors. In total, based on the dimensions of the rooms, two temperature, six PIR and six motion sensors are applied as room sensors. For the object sensors, five pressure, two humidity, ten motion, and thirteen light sensors are utilized as object sensors. This makes a total of forty‐four sensors applied in order to create the Virtual Reality Smart Home with their corresponding activities and interactions, as depicted by Appendix H. 5.4 Virtual Reality tour After participants have filled in the pre‐survey they are guided to the experiment site, where the Virtual Reality Powerwall is set up, portrayed in Figure 5.8. Figure 5.9 shows the 3D glasses which the participants will wear, the wand and the 3D glasses that can be worn by the researchers. The white nodes on the 3D glasses and the wand can be tracked by the Powerwall setup. The two cameras at each side of the wall can track the movement of those nodes. Consequently, when wearing the tools, participants walk within the provided space in front of the Powerwall and their movements will be registered in the Virtual Reality Smart Home. In addition, participants can either jump or bend, go through their knees or try to grab an object,

Figure 5.8 Powerwall & tracking cameras Figure 5.9 3D glasses with tracker, wand and 3D glasses without tracker

53 which all can be monitored and registered through the program. The wand is applied as controller, since all interactions in the Smart Home that are not automatically programmed are performed by using this wand. In the Figures of this Section sometimes a hand is shown, indicating the representation of the wand in the Virtual Reality program. The buttons on the wand Figure 5.10 Routing Virtual Reality experience represent certain hand movements or gestures like grabbing, touching or opening. After a brief explanation to the respondent about the wand, the control buttons and the 3D glasses, the program is launched.

The storyboard of the experience tour is displayed in Appendix I, accompanied by the text that is spoken by the voice guide. The route of the tour is depicted in Figure 5.10. The program starts in front of the Smart Home, where is explained that the performed activities will focus on comfort, energy saving, security and luxury, after which the Figure 5.11 Start VR tour in front of the Smart Home participant can enter the Smart Home. This stage is represented in Figure 5.11. In the hallway, the Smart Home tour will begin, the participant is asked to continue to the kitchen. Allameh et al. (2013) declare that the kitchen and the living room will be the centre point for Smart Home’s daily life. Therefore, the kitchen forms the starting point of the activities, see Figure 5.12. The participant is informed about the fact Figure 5.12 Kitchen activity starts and blinds opening that the blinds are opening automatically, which they can perceive and also the floor heating is turned on, which is not perceivable since in reality one cannot see the floor heating up either. Then, the participant is asked to walk to the fridge and open it in order to take the kiwi and to place it in the fruit bowl behind them, framed in Figure 5.13.

When placing the kiwi, it is pointed out Figure 5.13 Kitchen activity of automatically opening that the fridge is closing automatically as and closing the fridge the fridge is left unattended. From the same fruit bowl, the respondent is asked to take the green apple

54 and place it on the chopping board, which will trigger a focus light to give the best possible chopping light as show in Figure 5.14. When the kitchen activity is finished, the participant is asked to walk to the couch in the living room and push on the left side of the headrest. This action is displayed in Figure 5.15. The Smart Wall will be launched and music will be played to let the participant get Figure 5.14 Kitchen activity with the focus light acquainted with Smart Home Cinema. Since the Smart Wall is also capable of supporting work‐related activities, the participants are asked to tap the right side of the same headrest, shown in Figure 5.16. Google is launched and the participant will search for the news by asking the Smart Wall to search for the ABC News website. In order to tap the headrest, the participants will have to bend their knees and they might have to Figure 5.15 Launching the Smart Home cinema take a real step to the couch, which increases the realism of the Virtual Reality program. After this luxury activity, the living room activity is completed and the participants will continue their way to the backyard where the first security appliance awaits. When stepped outside and after turning around, it is shown that the blinds are closing and the Smart Wall is shut down, as depicted in Figure 5.17. In addition, the Figure 5.16 Internet search in the living room floor heating is turned off, all because there is no movement anymore in the rooms. The tour will continue through the other sliding door of the back garden, as depicted in Figure 5.18. Through that door, the original hallway will be reached again.

Figure 5.17 Outside activity when turning around Figure 5.18 Crossing the backyard to the next sliding door

55 The participant will move through the bedroom door on the right, where an energy saving solution is provided. The bedrooms show some lighting options, in which the lights go on when someone walks into the bedroom. This stage is shown in Figure 5.19. Afterwards, participants need to continue their way down the hall and then take the door on their right, which will bring them to the Figure 5.19 Bedroom activity with lights going on bathroom, to experience the second luxury activity.

When entering the bathroom, the participants are kindly asked to move forward towards the bathtub. When the water tap is touched the bathtub will fill itself. Simultaneously, relaxing music is played and the light colours change to a soothing and preset colour. This is displayed in Figure 5.20. As participants try to bend over the tub to open the taps, they already experience the feeling Figure 5.20 Bathroom activity when opening water tap of sitting inside the bathtub. When the tub is drained, the music stops automatically and the lights return to normal. Furthermore, the humidity needs to lower and therefore the windows will open instantly, see Figure 5.21. Participants then will walk to the mirror, located above the washbasin, which is a nice extra feature to make the tour informative but fun at the same Figure 5.21 High humidity detected and windows open time, as they can see their own hand reflection. After the bathroom activity, the participants return to the hallway, leaving the Smart Home through the front door. There, they will experience the final security activity, in which the door is locked and the bathroom windows that were open are closed, since there is no movement anymore in the Smart Home. This security measure is depicted in Figure 5.22. The tour will be Figure 5.22 Windows close and door locked for security concluded by thanking the respondents for their participation while they are walking towards the enclosing hedge. When finished, the participant can take off the 3D glasses and step outside of the Virtual Reality world.

56 5.5 Interaction through programming In order to create a real‐life experience through Virtual Reality, the activities need to be executed individually through real‐life actions. Therefore, launching a Smart Wall or placing an ingredient on a chopping board needs to be performed singlehanded. Besides modelling the environment, the interactions and sensor‐driven activities need to be programmed. Programming will be done in Python code by applying Vizard 5, being the Virtual Reality software of the WorldViz platform (WorldViz, 2015). Besides the activities, the voice guidance will be included in the programming as well. The environment as modelled in 3ds Max Design is exported to Vizard. Eventually, an import tool named vizconnect is applied to adjust the Virtual Reality environment of 3ds Max in order to define the appropriate perspective, hand gestures, walking speed and spin rate at which the participants will explore the model.

Codes are written step by step in order to guide the participant to and through every activity. Since it is necessary to inform the computer about every single aspect of the routing and activities, multiple variables and functions are defined. Variables are necessary to hold values or are used when accessing the command list of Vizard. The command list has to be applied when performing or loading a compatible model, file or action into the Virtual Reality setting. Functions are useful when conducting and organizing a large script, by dividing that script into manageable parts. Bits of code can be admitted to such a function. Besides code, additional parameters or arguments can be included between parenthesis. When the information within the function is required, the function can easily be called in order to perform a certain preset activity. This type of interaction is commonly used in the Smart Home script when applying for instance revolving or sliding doors, adjusting lighting control, launching activities or closing blinds.

soundintro = viz.addAudio('EN1 - Welcome.wav') soundintro.loop(viz.OFF) soundoutro = viz.addAudio('EN20 - Thank you.wav') soundoutro.loop(viz.OFF)

A piece of the code used is provided as an example above. In order to import the audio files that have been created with the Acapela text to speech program, a variable is generated. The variable evokes the command list of Vizard through which a wav‐file can be added. In this case, it is necessary that both the introductory and final audio are activated when a participant is walking within a certain proximity field. Obviously, the files should not play simultaneously, but one after the other, the first time when the participant is entering ('EN1 - Welcome.wav') and the second time when he is leaving ('EN20 - Thank you.wav'). The second and fourth line of code are included to prevent the audio file from eternal looping.

57 countEnter1 = 0 def IntroAudio(e): global countEnter1 countEnter1 = countEnter1 + 1 if countEnter1 < 2: soundintro.play() elif countEnter1 >1: soundoutro.play() vizact.ontimer2(5,0,outside)

Like explained, the software does not automatically assume that both files need to be played after one another. Therefore, this needs to be explicitly defined within the code. A global variable has been applied to direct the process. These variables are used when the content needs to be accessible throughout the script. However, the content can be altered for a specific purpose within a function. In this code, a global variable is used to count the number of times a participant is entering a specific area or proximity. A function is created by the name of IntroAudio, indicating the first set of audio files utilized. The argument e as applied for this function indicates an empty argument, since no additional parameters are necessary. When defining the content of the function, note that the if statement indicates a certain condition for executing a request. In this case, the welcome variable as set above can only be called upon when a participant is entering the proximity less than two times, resulting in one single entry. Likewise, the second time a respondent enters the field, another condition is adopted. The empty parenthesis behind the variables indicate that the variable method can be invoked whenever the condition is met. The elif statement means else if, indicating that when the first if condition is false and the elif condition is true, the first should be ignored and the elif code should be executed. The last line of code is based on a specific module that can be imported into Vizard. The vizact module can be invoked to perform a certain action. In this case, the module is used to call the IntroAudio function on a specific time interval, being once after five seconds. Hence, when the participant enters the proximity for the second time, the audio file Thank you will be played, and after five second, the function outside will be executed. The function outside is not defined in here, since it includes all the security activities and interactions as discussed earlier, resulting in an excessive script for this overview. Eventually, a proximity field has to be created for this specific purpose. All the information as provided so far has to be attached to that field.

prox = vizproximity.Manager() prox.setDebug(False) prox.addTarget(vizproximity.Target(viz.MainView)) verts1 = [(3,15),(3,10.7),(-3,10.7),(-3,15)] sensor1 = vizproximity.Sensor( vizproximity.PolygonArea(verts1,offset=[0,0]), None ) prox.addSensor(sensor1) prox.onEnter(sensor1, IntroAudio)

58 By creating a proximity area through the imported vizproximity module, the exact layout of the polygon can be defined through coordinates retrieved from 3ds Max. The coordinates of the entrance path of the front garden are defined. As indicated in the last line of code, when the created proximity field is entered, the function IntroAudio as defined earlier, will be called, executing its content. The code as shown above is just a minor part of the entire script, indicating the complexity and detail level at which must be operated in order for the Smart Home virtual tour to run. Note that this small piece of code only ensures that the voice guidance will start and end when the participant is walking over the garden path. The entire script will not be provided within this thesis, as it only serves as the mean to achieve the Virtual Reality Smart Home experience as discussed in Section 5.3 and 5.4. 5.6 Conclusion In order to design a Virtual Reality Smart Home tour, certain aspects are necessary to make it a realistic representation. First the themes and activities which address the Smart Home technology, followed by the Smart Home design and finally programming of the interactions and sensor‐driven activities. Four themes are set up, being convenience, security, luxury and energy saving. These themes are part of the greater underlying concept of enhancing comfort. Five activities are developed to show as many Smart Home technology aspects as possible and take all place in different areas, being the kitchen, living room, bedroom, bathroom and outside. All these activities are accomplished by implementing 44 sensors. Multiple other appliances are possible, however this shows a broad variance in themes and possibilities to encourage participants to contemplate on what Smart Home technology implementations would be suitable for their home environment. The Smart Home is modelled by different programs in order to achieve the best possible realistic depiction of a Smart Home. The home was designed as a modern and high‐tech environment, furnished with plants and paintings in order to provide a domestic and urban atmosphere. Sensors are ambient and seamlessly embedded within the environment to support the comfort of the residents. By programming the interactions, the participant is immersed into the Virtual Reality world to create an even more realistic experience. Using the wand for performing certain activities only contributes to this feeling. The total experiment will take up around twenty to thirty minutes per respondent, in which the experience requires half of that time. Understandably, the total duration of the experiment depends on the capability, handiness, understanding, reading ability and condition of the individual participants.

59 6 Data collection and sample description 6.1 Introduction This chapter analyses the data retrieved from the experiment, by applying IBM SPSS Statistics 22. Only if a respondent completed all three elements of the experiment, results can be analysed. SPSS is commonly used as a program where entering, editing and analysing data can be done in an extensive and clarifying way. Through this, the user acceptance model developed in Chapter 3 is analysed. The analysis results of the pre‐ and post‐data are compared to see whether the experiment changed the opinion of the respondents about Smart Homes and its technology. Therefore, the pre‐ and post‐survey are of vital importance since it provides the underlying proof of a potential difference in user acceptance after experiencing the Smart Home technology in Virtual Reality. Firstly, the process of collecting data and the prevention of errors will be described in Section 6.2. More information about the participating sample is outlined in Section 6.3, in which the descriptive statistics of the data are addressed. The first step in analysing the user acceptance model is to test the reliability of the survey questions per construct. The reliability will test whether the questions are all measuring the same variable. Only if that is the case, questions can be combined into constructs. When considering the experiment, the same questions in both the pre‐ and post‐survey need to be reliable in order for the question to be suitable. After this is verified, the sample statistics can be further elaborated. This section will also highlight how participants experienced the concept of Smart Homes through Virtual Reality and the VR setting as an instrument. Besides, the outcomes of the observations made during the experiment are also addressed in this Section. These observations concern the reactions of the participants and the support needed with the controllers and the Smart Home technology. This chapter will conclude in Section 6.4.

The main sub‐questions central to this chapter are:

 How are the data obtained and what is the composition of the sample?  How do the participants perceive the Virtual Reality experience? 6.2 Data collection Since the Virtual Reality setup is permanent and immovable for the purpose of this thesis, potential respondents had to come to the Eindhoven University of Technology campus. This limitation was turned into an opportunity by randomly inviting people working at the campus to participate. The participants were approached by addressing them on the campus and introducing the research. Furthermore, potential respondents were made aware of the experiment through posters on campus, social media and email. Respondents could make an appointment or walk in during each day. Accordingly, due to the flexible and easy accessible outline, a large and diverse population could conduct the experiment. The data for the study were gathered within a limited time frame, since the experiment was held for three weeks starting from the 10th until the 30th of June 2015. Taking part in the experiment, the participant was isolated and he or she completed the pre‐survey in an adjacent room of the VR experience room, in order to prevent interaction or foreknowledge. In total, the experiment took about 25 minutes per individual. As stated in Chapter 4, participation was anonymously as well as voluntary. Participants did not receive any reward for participating, since this could influence their

60 decision making process. In addition, this could lead to socially desirable answers or respondents randomly or quickly filling in their surveys in order to obtain the reward. During the experience, the participant was observed by the two researchers in order to record reactions about both the Smart Home and the VR technology. Furthermore, the participants were given the possibility to ask questions during both the surveys and the Virtual Reality experience to prevent obscurities. Support in addition to a Virtual Reality voice guide was provided verbally to help the participant in a non‐physical way when required during their Smart Home assignments.

The developed user acceptance model contains four constructs, being perceived usefulness, perceived ease of use, facilitating condition, and intention to use and four moderators, age, gender, technological knowledge, and social influence, each of which are defined and represented in the pre‐ and/or post‐survey by one or more questions, as displayed in Table 4.2. Although constructs and moderators are based on scientific models that are verified by various studies, the model and thereby the survey questions still need to be tested on reliability. When a construct is reliable, the questions that are part of the construct are all measuring the same aspect. As the constructs and moderators of the user acceptance model are measured by multiple questions, it is necessary to combine these to analyse them correctly. Therefore, it has to be checked whether it is statistically possible to formulate one variable per construct, which is composed of all the survey questions concerning the particular construct. This reliability analysis can be performed by applying Cronbach’s alpha. The Cronbach’s alpha examines mutual correlations through which an assessment can be made on the reliability of the scales used. As depicted in Table 4.2, the number of questions per construct or moderator differ. A variable can only be perceived reliable when correlations are proven for the pre‐ as well as the post‐survey. If not, it is necessary to delete the particular question in both pre‐ and post‐survey. In order to obtain a reliability analysis in which both pre‐ and post‐survey results can be compared, it is important to discuss both survey variables side by side in this chapter. It is decided to exclude question 6 concerning the buying behaviour of innovative technology and question 7 regarding thinking about purchasing Smart Home technology of the pre‐survey from the moderator social influence. This is due to the fact that these questions are ambiguous. Social influence is only analysed through question 5 of the pre‐survey, being the number of households one knows.

In order to optimize the accuracy of the measurement of the experiment, it is important that errors in the experiment are prevented. Experimental errors can be divided into two types, systematic and random errors. The most important difference between the two is that systematic errors cause all results to be incorrect and cannot be eliminated by a statistical method, and random errors are minor experimental or accidental errors which cannot be controlled (National University of Singapore, 2015). Incorrect calibration is counteracted by calibrating the Powerwall each day and by testing the Virtual Reality setup before letting participants execute the program in order to prevent systematic errors. The experimental conditions were kept constant throughout the time span of the experiment. A potential bias of the observer is counteracted by observing with two observers and by deliberating after each observation. There were some random errors that occurred during the experience. The first was a program error in which participants would jump to the roof instead of continuing the program. However, this interruption

61 sometimes even occurs in video games, which made it more difficult to prevent. Whenever it happened, the observers made sure the bug did not distract the respondent by telling them how to respond or by restarting the program. The second one occurred when the participant’s height was below 1.65 meter, since the system had difficulties tracking someone’s head and hand movement at that height. By holding the wand higher, this error could be more or less prevented. Furthermore, due to the fact that the pre‐ and post‐survey contained the same questions, it was possible that the participant remembered certain answers. However, this is hard to prevent, since the pre‐ and post‐ survey needed to be identical in order to analyse the user acceptance model. 6.3 Sample and descriptive statistics In the end, 134 participants performed the experiment. One of the participants could not complete the experiment due to nausea, bringing the total sample size to 133 useful cases. The distribution and characteristics of the sample will be further elaborated in this Section. All variable constructs within the user acceptance model are tested on reliability with Cronbach’s alpha. The reliability coefficient should be higher than 0.6 to be considered acceptable and a value of 0.7 or higher represents an adequate reliability (Yong, et al., 2007; Nunnally, 1978). Nunnally (1978) indicates that for preliminary research, a Cronbach’s alpha of 0.5 is also reported as sufficient. This thesis has passed this stage, therefore a Cronbach’s alpha of at least 0.6 or higher is adopted. Before the reliability per variable can be assessed, some questions need recoding in order to create uniformity in the submitted answers and to obtain a positive reliability value. No missing variables were found, due to fact that every survey was checked for missing answers directly after it was filled in by the respondent.

Almost two third, 63,2%, of the participants in the experiment were men, being a total of 84 men and 49 female respondents. Figure 6.1 shows the age distribution, which is not normally spread. In the caption of Figure 6.1, the pre‐Q1 refers to question 1 of the pre‐survey. Age was measured on a ratio scale. The ages up till 34 years are overrepresented and comprise of 53.4% of the total sample size. Considering both gender and age, the percentages are explained by the fact that the experiment was conducted at the Eindhoven University of Technology, where young people are educated. At this university, one Figure 6.1 Age ‐ Pre‐Q1 out of every five students is female, and within the group of professors this percentage is even lower (TU/e, 2006; DUB, 2013; Stichting de Beauvoir, 2012). However, the average age of the sample is 39, with a standard deviation of 17.3, indicating a large range of ages. This is true, since the youngest participant was 14 and the oldest 93. Furthermore, the participants over 65 years comprise only 10.5% of the sample size. This is explained by the fact that participants over 65 years have more difficulties participating in the Virtual Reality experiment, because it is set up at the university.

62 Technological knowledge is measured by two different aspects. The first question asks about the familiarity with Smart Homes and the other one about the current IT use, as depicted in Figure 6.2 and 6.3.

Figure 6.2 Technological knowledge – Familiar Figure 6.3 Technological knowledge ‐ Current with Smart Homes Pre‐Q3 use IT Pre‐Q4 Both questions indicate a normal distribution. A normal distribution will show that the responses of the participants are clustered around a mean value of the construct. The answers are symmetrical and gradually spread toward each extreme answer. When an ideal normal distribution is shown in a graph, a curve in the shape of a bell is formed to identify the ideal symmetry. In general, rather normally distributed data are necessary in order to perform certain statistical analyses. However, the data per construct can also be spread out more on the left or right of the graph. This does not necessarily indicate that there is something wrong, as long as the curve shape is rather close to the normal distribution (Baarda & De Goede, 2006). Generally, the participants are slightly to somewhat familiar with the dwelling type Smart Homes and 69.7% of the participants own three or more innovative technologies. It is assumed that when someone is using more innovative technologies, their technological knowledge will be higher compared to someone who is only using a feature phone.

When combining the questions, the Cronbach’s alpha of technological knowledge returns insufficient with a coefficient of .511. Both questions of the moderator encompass a different Likert scale and they also test a different area of technological knowledge. Question 3 tests the specific knowledge about Smart Homes, whereas question 4 tests the overall use of innovative technology. Furthermore, it is possible that participants do possess more innovative technologies, but are not familiar with Smart Homes. Instead of excluding a question, it would be best to treat both questions individually in the remainder of the analyses.

63 Social influence is measured by the question if participants know any households with Smart Home technology. Due to the fact that social influence is measured by only one question, no reliability analysis is necessary. Figure 6.4 shows the frequency of social influence. Most respondents know a maximum of two households that use Smart Home technology. Social influence considering Smart Home technology is rather limited. 60 participants, being 45%, even indicated not to know any households with Smart Home technology. Besides, only 12% of the 133 participants know three or more related households. Figure 6.4 Social influence – Pre‐Q5

When taking a close look at the data distribution of the construct perceived usefulness in the pre‐ and post‐survey as depicted in Figure 6.5 and 6.6, both are more or less normally distributed. Therefore, perceived usefulness can be regarded as a continuous variable when performing statistical analyses. Variables could also be used when non‐normally distributed, however in that case variables should be scaled on an ordinal level. The Cronbach’s alphas of perceived usefulness are both sufficient in the pre‐ and post‐survey, with a reliability coefficient of .694 and .703.

Figure 6.5 The data distribution of perceived Figure 6.6 The data distribution of perceived usefulness in pre‐survey – Pre‐Q8‐15 usefulness in post‐survey – Post‐Q1‐8

64 Similarly, the data distributions for perceived ease of use are checked in both the pre‐ and post‐survey. The results are displayed in Figure 6.7 and 6.8. Again, both constructs show a rather normal distribution of their data. The fact that the distribution is slightly spread more to either left or right does not result in a non‐normal distribution. Since both reliabilities are sufficient and the data is normally spread, perceived ease of use can be applied under normal conditions when conducting the analysis. Considering the Cronbach’s alpha of perceived ease of use of both surveys, the coefficients show an excellent reliability. While the pre‐survey indicates a reliability of .874, the post‐survey displays a Cronbach’s alpha of .851.

Figure 6.7 The data distribution of perceived ease of Figure 6.8 The data distribution of perceived use in pre‐survey – Pre‐Q16‐20 ease of use in post‐survey – Post‐Q9‐13 The construct of facilitating conditions presents a Cronbach’s alpha of .496 in the pre‐ and an alpha of .568 in the post‐survey. For the research conducted within this thesis, these coefficients are considered insufficient. Consequently, at least one of the questions involving facilitating conditions needs to be removed from both surveys. This construct is tested through three individual questions about whether additional support would enhance the participants’ attitude, whether free support would increase interest and whether Smart Home technology should be compatible with other technologies. When taking a closer look at the type of questions, the first two questions deal with the support provided with the technology. Both questions have the same characteristic approach. The compatibility question does not measure facilitating conditions in the same manner, since it focuses on a different aspect. This can also be derived from the change in Cronbach’s alpha when one of the items is deleted. Deletion of the additional support question would result in an Cronbach’s alpha of .094, excluding free support in an alpha of .265 and removing compatibility in an alpha of .667 in the pre‐survey. The post‐survey shows comparable results. A reason for this could be that participants who believe the system should not necessarily be compatible with other technology they use, might still want free and additional support, and vice versa. Therefore, the reliability differs in such a way that the compatibility question has to be excluded, being question 23 in the pre‐ and 16 in the post‐survey. When deleted, the Cronbach’s alpha restores itself to an acceptable coefficient, leaving the facilitating conditions construct based on two questions. Figure

65 6.9 and 6.10 show a normal distribution for the construct facilitating conditions. Compared to the other distributions shown in this Section, the data of facilitating conditions seem similarly and equally distributed for both pre‐ and post‐survey. This slight difference might be explained by the fact that only two questions are still considered part of that construct. The means and standard deviations as illustrated in Figure 6.9 and 6.10 are nearly the same.

Figure 6.9 The data distribution of facilitating Figure 6.10 The data distribution of facilitating conditions in pre‐survey – Pre‐Q21,22 conditions in post‐survey – Post‐Q14,15

Figure 6.11 The data distribution of intention to use Figure 6.12 The data distribution of intention to in pre‐survey – Pre‐Q24,25,26 use in post‐survey – Post‐Q17,18,19 For intention to use, the results are depicted above in Figure 6.11 and 6.12. The data sets of both pre‐ and post‐survey show an acceptable spread. The means indicate that for the majority of the answers, a rather positive response is given. The three questions based on intention to use are whether participants would like to know more about the technology, whether they would like to install it in their own home and whether they would consider purchasing the technology in the near future. Just like the perceived usefulness, perceived

66 ease of use, and facilitating conditions, intention to use has a sufficient Cronbach’s alpha for both the pre‐ and post‐survey, being .797 and .735. The normal distributions of facilitating conditions and intention to use define that both constructs can be admitted in a parametric statistical analysis that is necessary in order to analyse the entire user acceptance model.

In the post‐survey, questions 20 up to 25 are concerned with the topic of the Virtual Reality experience. The answer ratio of those questions can be seen in the pie charts of Figure 6.13 up to Figure 6.17 and Table 6.1.

Figure 6.13 VR experience helps understanding Smart Figure 6.14 VR experience influences opinion about Home technology for home environment – Post‐Q20 Smart Home technology – Post‐Q21

Figure 6.13 shows that more than 90% of the participants perceived that the Virtual Reality experience helped in understanding what Smart Home technology could mean for their own home environment. This is a very interesting statement, bearing in mind that understanding is closely related to the user acceptance, as described in Section 3.1. In addition, it attracts attention that the respondents were very opinionated when answering this question, as the answer option undecided was hardly filled in. Figure 6.14 indicates that 62,4% believed that the VR experience influenced their opinion about implementing Smart Home technology, which are 83 out of the 133 respondents. 30 respondents are not at all or hardly convinced that the VR experience has influenced their opinion, being 22,6%.

According to Figure 6.15, 79% of the participants have a better understanding of what the term Smart Homes entails, after experiencing one. 15,8% of the participants gave notice that they hardly had a better understanding of Smart Homes after experiencing. A potential reason for this percentage could be that some participants already experienced a Smart Home or are working on the subject of Smart Homes, what could be the case since the experiment also Figure 6.15 Virtual Reality experience creates better involved students and professors. This understanding of Smart Homes – Post‐Q22

67 could explain why the Virtual Reality experience did not influence their opinions.

Figure 6.16 Whether using Smart Home technology Figure 6.17 Experiencing Smart Home technology can provide energy or financial benefits ‐ Post‐Q24 changed my attitude ‐ Post‐Q25

Question 24 and 25 of the post‐survey concerning the VR experience, are displayed in Figure 6.16 and 6.17. Based on what the participants experienced in VR, Figure 6.16 displays that more than 75% believes that applying Smart Home technology can provide them energy conservations or other financial benefits. Apparently, the Virtual Reality experience appeals to the respondents’ minds to such an extent that they are convinced that Smart Home technology can be financially beneficial. In addition, Figure 6.17 indicates that by experiencing Smart Home technology, more than 60% perceived a change in their attitude towards Smart Homes. On the other hand, almost 22% of the respondents did not or only perceived a little change in their attitude. Note that not one participant filled in the answer option not at all in question 24 of the post‐survey regarding financial benefits. Therefore, only four answer option are shown in Figure 6.16.

In questions 23, the participants were asked Table 6.1 Feelings during Virtual Reality experience – about their feelings during the experiment, Post‐Q23 concerning the Smart Home technology. It was possible to check multiple boxes of Feeling Percentage per feeling feelings. Therefore the percentages given Positive feeling Exited 56,4 % in Table 6.1 are a score per feeling and not Interested 75,2 % in total. Overall, responses about Comfortable 35,3 % experiencing the Smart Home technology Secure 12 % resulted in mainly positive feelings. 75.2% In control 19,5 % Fortunate/Happy 30,1 % of the participants felt interested, more Negative Anxious 5,3 % than half even felt exited. Experiencing the feelings Nervous 9,8 % technology did not really make the Controlled 12,8 % participants feel secure or in control. Out of Confused 8,3 % the negative feelings, being controlled and Weird 13.5 % feeling weird were the main response Other 4,5 % answers that were given. Six participants addressed that they experienced also other feelings than the ones given. They felt rich, overwhelmed, out of control, nothing special, had an ‘aha!’‐moment or a feeling of recognition.

68 In addition to the VR experience questions, four observant questions were included that needed to be filled in by the observers. The results of those questions are displayed in Table 6.2 and 6.3.

When comparing both Table 6.2 Observation help with Virtual Reality set up and controllers observation questions about and Smart Home technology guidance during the Help Smart Home experiment, it is clear that Help controllers technology participants needed more help concerning the Frequency Percent Frequency Percent controllers in contrast to the Not at all 60 45.1 % 85 63.9 % technology involved. The A little bit 27 20.3 % 34 25.6 % observance scale in Table 6.2 Sometimes 26 19.5 % 13 9.8% requires some explanation, Considerably 15 11.3 % 1 0.8 % regarding the terms. Not at all Definitely 5 3.8 % ‐ ‐ was checked when a Total 133 100.0 % 133 100.0 % participant did not need any additional explanation or guidance during the program. When a participant needed extra instructions or directions one or two times, a little bit was checked, three or four times sometimes, five or six times considerably and with seven or more times definitely was checked. Table 6.2 shows that 15.1% needed a lot of help with the controllers, which is almost one out of every seven respondents. Fortunately, the majority only needed minor additional instructions and almost half the participants did not request any additional help. Considering the help with regard to the Smart Home technology, the number of times help was required is considerable less. Only one person required a considerable amount of help with the Smart Home technology and none needed definite guidance. It could be that there is a relation between the age of the participant and the level of help they needed. The age generations under 34 years old probably all grew up with video games and are therefore perhaps more used to applying controllers.

More than half of the Table 6.3 Observation Response to Virtual Reality set up and participants gave a positive controllers and Smart Home technology reaction to both the Virtual Response Smart Reality tools and the Smart Response controllers Home technology Home technology, which is Frequency Percent Frequency Percent shown in Table 6.3. However, it is possible that these results Negative reaction 17 12.8 % 1 0.8 % are influenced by social Neutral reaction 49 36.8 % 18 13.5 % desirability, as participants will Positive reaction 67 50.4 % 114 85.7 % rather give a positive reaction Total 133 100.0 % 133 100.0 % instead of a negative reaction in order to prevent disappointment. Furthermore, there is also the possibility of a measurement error, when the observers are biased or do not interpret the situation correctly. The errors have already been thoroughly discussed in Section 6.2.

During the observation of the experiment, another negative feeling was experienced by some participants, however this had nothing to do with the Smart Home technology. 15.8% of the participants felt uneasy, of which nine participants felt really nauseous, four

69 were a bit nauseous and eight people felt unstable or unbalanced. This is a common problem when performing Virtual Reality in front of the Powerwall, due to the fact that walking for the biggest part does not take place in the physical environment, but is done by using the controllers. The human brain cannot always process this phenomenon correctly, as it is fooled by the unnatural movement and the realism of VR. The way in which this affects the respondents varies per person. At the beginning of the experiment, it was pointed out to the participant that this could happen and that they should notify the researchers if they felt nauseous or uncomfortable. It is even possible that the percentage as mentioned is even higher, if people concealed the fact that they were nauseous out of socially desirability, or when they experienced some nausea later on. 6.4 Conclusion All constructs and moderators of the user acceptance model showed a sufficient Cronbach’s alpha and were normally distributed. Since the experiment was conducted at the Eindhoven University of Technology, the majority of the respondents were men and under the age of 34. Almost 70% of the participants owned three or more innovative technologies and most respondents had at least heard of the dwelling type Smart Homes, although 45% of the participants did not know any household with Smart Home technology.

Overall, looking at how the respondents perceived the Smart Home technology in the VR setting, they experienced a more positive feeling regarding the technology. However, on the downside 15.8% felt nauseous or unstable during or after finishing the experiment. When this would not have been the case, perhaps even more participants would have had an overall positive feeling. The technology is still rather new and undergoes many developments, therefore this obstacle might be overcome over time. The game community contributes highly to these improvements and they also focus on innovative ways to interact with the VR world through different tools (Zyda, 2005; McMahan, et al., 2012; Ma, et al., 2014). As some participants needed help with the controllers in the experiment, these developments could help improving experiences when these tools are applicable for the Powerwall.

In general, results show that participants have a much better understanding of Smart Home technology and Smart Homes after experiencing the Virtual Reality Smart Home. More than half even believe that experiencing has influenced their opinion about Smart Home technology and indicate that it changed their attitude towards the technology. Considering the fact that 75.9% of them think that Smart Home technology could provide them energy conservations or other financial benefits, this will have a positive influence on their opinion and attitude.

70 7 Data analysis 7.1 Introduction Within Chapter 6, the data collection and the descriptive statistics of the sample are described. In this chapter, the data analysis is conducted. In Section 7.2, the relationship between the three independent variables, perceived usefulness, perceived ease of use and facilitating conditions with the dependent variable intention to use are examined. Through a linear regression, it is verified whether the relationships between the constructs are significant and how much of the total variance is explained by the variance of the independent variables. This has to be evaluated for both the pre‐ and post‐survey model. Next, Section 7.3 shows a comparable linear regression analysis, in which the significance of the moderators and the interaction effects are analysed. In addition, this section elaborates on the verified relationships between the moderators and the constructs. As intention to use is also tested through three individual questions in both the pre‐ and post‐survey, Section 7.4 compares those results through a Paired Samples T‐ test. In Section 7.5, the conclusions drawn from the analyses are presented and the final user acceptance model is shown. Appendix J and K are provided to support the analyses of this chapter.

The main sub‐questions central to this chapter are:

 Which statistical analyses are necessary to analyse the data?  What are the results of the pre‐ and post‐survey comparison?  How can the results of the analyses be interpreted? 7.2 Analysing the relationships By discussing the sample and descriptive statistics, more knowledge is obtained about the sample size and the characteristics of the constructs and moderators. These analyses are just utilized as a foundation and precondition for applying statistical methods in order to examine the entire user acceptance model. At first, it is necessary to distinguish whether there is a relation between the constructs. The statistical method of linear regression attempts to predict whether there is a relationship between a dependent variable and one or more independent variables (StatisticSolutions, 2015; Laerd Statistics, 2015). Since the pre‐ and post‐survey are analysed side by side instead of simultaneously through one statistical method, linear regression is appropriate to identify the individual relationships between constructs. Structural Equation Modeling (SEM) might seem appropriate in this case as well, since SEM considers several equations simultaneously, whereas ordinary regression analysis only predict one at the time. However, SEM is applied in the rather more complex models with interrelations between independent variables or more dependent variables, with either very large or small sample sizes and when data is not normally distributed for most of the time (Alavifar, et al., 2012; Monecke & Leisch, 2012). Therefore, applying a linear regression would suffice for this user acceptance model. When performing such a statistical analysis, the result will show a so‐called coefficient of determination, also known as R square, indicating how well the data retrieved fits the statistical model.

It is suggested that research should aim for a coefficient of determination of 1, however this is only possible to reach within an ideal and controlled environment. Chin (1998) indicates that in the type of research in which this thesis is conducted, lower values are

71 applied, which is in line with Hudson (2000). These levels are respectively .19, .33 and .67, for weak, moderate and strong relations. The explanatory power of the model is determined by the R Square value of the dependent variable intention to use.

Figure 7.1 shows the linear Perceived regression for the pre‐survey, Usefulness .289 *** 2 consisting of the independent R : .464 variables perceived usefulness, Perceived Ease .522 *** Intention to of Use use perceived ease of use and .251 *** facilitating conditions and their Facilitating relationships to the dependent Conditions variable intention to use. The Figure 7.1 Linear regression of user acceptance model based on model displays an explanatory the pre‐survey power and R Square of .464, indicating a moderate to strong relation and therefore can be considered substantial. The statistical information behind the R Square of the model of Figure 7.1 is provided in Appendix J.1. The value indicates that 46.6% of the total variance is explained by the variance of the independent variables. However, this is the R Square without including the moderating effects. In Section 7.3, the R Square including the moderating effects will be discussed. An ANOVA shows that the cumulative effect of all three independent constructs are significant at a p < .001 level, which is also indicated by the three asterisks (***) behind the values of Figure 7.1. The unique effects of all independent variables are significant and positive, as displayed in Appendix J.2. This means that an increase in value of one of the three independent variables will elicit an increase in the dependent variable intention to use.

Figure 7.2 shows the linear Perceived regression model of the post‐ Usefulness .431 *** 2 survey. The post‐model indicates R : .493 an R Square of .493, which can Perceived Ease .435 *** Intention to also be found in Appendix J.3. The of Use use .269 *** R Square including the Facilitating moderating effects is again given Conditions in Section 7.3. Compared to the Figure 7.2 Linear regression of user acceptance model based on coefficient of determination of the the post‐survey pre‐survey model, the value has increased a little bit and is still sufficient according to Chin (1998) and Hudson (2000). An ANOVA of this model shows that the cumulative effect is still significant at p < .001. Likewise the results for the pre‐survey model, Appendix J.4 shows the unique effects of the independent variables, which are all positive as well as the effects in the pre‐survey. In line with the R Square of the model, the unique effects of the post‐model even increased regarding to the pre‐survey. 7.3 Analysing the moderator effects The relations as described in Section 7.2 show that the underlying basis of the model as developed in Chapter 3 is well‐founded and legitimate. However, besides the relationships between the independent variables and the dependent variable, there are some moderators potentially influencing these relationships. Based on other research, it is assumed that the moderator gender influences the two relations between on one side perceived usefulness and perceived ease of use and intention to use on the other side.

72 Furthermore, age is considered to influence all three relations, whereas technological knowledge excludes the relation between perceived usefulness and intention to use. On the contrary, the moderator social influence is claimed to just only influence that particular relationship that has been left out by technological knowledge. In order to examine whether these assumptions are correct, all potential interaction effects are tested on all relations. In the end, it can be checked whether the assumed moderating effects have proven to be right after all or new relations are revealed.

In order to test the moderating effects, interaction variables are created for each possible relationship. First, the variables are standardized to ensure that all variables contribute equally to the analysis. Second, a numeric expression is created by multiplying one of the constructs by one of the moderators. This multiplication is performed and repeated for every relationship. For instance, in order to test the moderating effect of social influence on the relation between facilitating conditions and intention to use, social influence is standardized by multiplying it to facilitating conditions in SPSS, to create an interaction variable. By displaying all the potential relationships into one linear regression table, the insignificant relations can be removed. If a moderating effect is significant, the moderator itself cannot be removed, regardless of its significance (Hair, et al., 2009). Therefore, despite the fact that age, current IT use and familiar with Smart Homes are insignificant, they need to be included in the analysis because of the significant relations of facilitating conditions with age and with familiar with Smart Homes, and of perceived ease of use with current IT use and familiar with Smart Homes. Eventually, only the identifiable relations remain. Table 7.1 shows the results from analysing the moderating effects in the model, based on the pre‐survey data. Appendix K.1 shows the moderating effects through a linear regression model.

Table 7.1 The remaining moderator effects on the relationships between the constructs of the pre‐survey

Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta t Sig. (Constant) ‐.059 .064 ‐.919 .360 Standardized Perceived Usefulness .304 .066 .304 4.581 .000 Standardized Perceived Ease of use .553 .074 .553 7.434 .000 Standardized Facilitating conditions .302 .061 .302 4.911 .000 Interaction Usefulness x Social Influence ‐.142 .061 ‐.158 ‐2.329 .022 Interaction Ease of use x Familiar with SH ‐.183 .057 ‐.218 ‐3.229 .002 Interaction Ease of use x Current IT use .241 .063 .265 3.846 .000 Interaction Facilitating Conditions x Age .132 .065 .128 2.018 .047 Interaction Facilitating Conditions x ‐.183 .064 ‐.173 ‐2.860 .005 Familiar with SH Standardized Age .088 0.64 .088 1.371 .173 Standardized Familiar with SH .084 .068 .084 1.227 .222 Standardized Current IT use .032 .069 .032 .464 .644 Standardized Social Influence .142 .072 .142 1.977 .050 aDependent Variable: Intention to use (standardized)

73 The majority of the moderating effects are insignificant, resulting in only five remaining moderator effects. These effects are not based on gender, indicating that none of the relations are influenced by whether respondents were male or female. Consequently, gender needs to be excluded from the model, leaving only age, technological knowledge (familiar with Smart Homes and current IT use) and social influence as moderators for the pre‐survey data relations. As displayed in Table 7.1, the remaining five moderating effects show sufficient significance. By including these relations into the model, the explanatory power of the model has increased from an R Square of .464 to .601, resulting in a much stronger relation, see Appendix K.1. The moderator social influence affects the relation between perceived usefulness and intention to use. In fact, the moderator has a negative influence on the relation with a standardized coefficient of ‐.158. When there is a positive relationship between two constructs, a moderator with a negative standardized coefficient can weaken the strength of that relationship. Familiarity with Smart Homes of technological knowledge influences both the relation between ease of use and facilitating conditions with intention to use. In both cases, the standardized coefficients of Table 7.1 show a negative influence on the relationship of ‐.218 and ‐.173. Perceived ease of use is also influenced by the other element of technological knowledge, being the current IT use. However, this relationship is positively influenced by the moderator with a coefficient of .265. The relationship between facilitating conditions and intention to use is the only one that is influenced by the moderator age. The standardized coefficient of this relation is .128, indicating that an increase in the moderating effect of age on the relationship between facilitating conditions and intention to use will establish a positive increase of the dependent variable.

The individual relations are illustrated in the graphs of Figure 7.3 up to 7.7. The moderating effect of social influence on the relation between perceived usefulness and intention to use is significant as indicated by Table 7.1. Social influence dampens the positive relationship between perceived usefulness and intention to use, as seen in Figure 7.3. When the perceived usefulness is low, social Figure 7.3 Moderating effect of social influence on influence highly influences the relation between Usefulness and Intention to Use relationship with intention to use. However, when there is a high perceived usefulness, social influence is of less importance to the relationship. This indicates that for the intention to use to increase, the perceived usefulness of Smart Home technology is far more relevant for someone who is less likely to be socially influenced than for a participant who is highly influenced by his or her social environment.

Figure 7.4 shows the effect of the technological knowledge element familiar with Smart Homes on the relationship between perceived ease of use and intention to use. As displayed in Table 7.1, the effect of this moderator on the relationship between ease of use and intention to use is highly significant with a p‐value of p < 0.001. A low ease of use combined with a low familiarity of Smart Homes equals a low intention to use. When the Smart Home technology provides a high ease of use, a high familiarity is not as important

74 anymore and the intention to use will increase. This indicates that familiarity with Smart Homes does not have much impact on the relationship between perceived ease of use and intention to use, if it were not for the fact that when there is a high familiarity with Smart Homes, the relationship between the two already starts at a higher intention to use. From Figure 7.4, the negative standardized coefficient of ‐.218 can be Figure 7.4 Moderating effect of familiarity with Smart distinguished by the fact that the line, Homes on relation Ease of Use and Intention to Use and therefore the effect of low familiarity is steeper than the line of high familiarity.

Due to the fact that the technological knowledge moderator is based on two diverging aspects of technological knowledge, it is best to separate those in the user acceptance model under the denominator of technological knowledge. The effect of the other element, current IT use, is shown in Figure 7.5. Table 7.1 indicates that this relation is highly significant. As the participants use more innovative Figure 7.5 Moderating effect of current IT use on relation technologies, the relationship between between Ease of Use and Intention to Use ease of use and intention to use becomes stronger. This indicates that perceived ease of use is stronger related to intention to use when the current use of innovative technologies by participants increases. As illustrated in the graph, the current IT use has more impact on intention to use when there is a higher perceived ease of use. The slope of the red line is significantly steeper compared to the blue line.

The graph in Figure 7.6 displays the effect of the moderator age on the relationship between facilitating conditions and intention to use. Participants with an lower age show an almost constant effect regarding the relationship. This indicates that at a younger age facilitating conditions, like additional help and free support, are less important than at an older age. The graph shows indeed that at an Figure 7.6 Moderating effect of age on relation between older age, better facilitating conditions facilitating conditions and intention to use lead to a higher intention to use.

When being rather unknown to the dwelling type Smart Homes, the relationship between facilitating conditions and intention to use becomes more important. This is shown in

75 Figure 7.7. Help and support is more important when participants are less familiar with the dwelling type and therefore the facilitating conditions are of great influence in increasing the intention to use. Whenever a respondent was already highly familiar to Smart Homes, the effect on the relationship between facilitating conditions and intention to use is less significant, as it only results in a minor Figure 7.7 Moderating effect of familiarity with Smart change in relation. This is also Homes on relation between facilitating conditions and supported by the negative value of the intention to use standardized coefficient of the interaction variable.

From the data received from the post‐survey, the number of moderating effects are lower compared to the pre‐survey analysis. Table 7.2 provides an overview of the results of the applied analysis. Appendix K.2 illustrates the moderating effects through a linear regression model.

Table 7.2 The remaining moderator effects on the relationships between the constructs in the post‐survey

Unstandardized Standardized Coefficients Coefficients Model B Std. Error Beta t Sig. (Constant) ‐.044 .063 ‐.700 .485 Standardized Perceived Usefulness .372 .065 .372 5.732 .000 Standardized Perceived Ease of Use .481 .072 .481 6.664 .000 Standardized Facilitating Conditions .258 .063 .258 4.099 .000 Interaction Usefulness x Current IT use ‐.202 .069 ‐.186 ‐2.942 .004 Interaction Ease of Use x Current IT use .148 .056 .163 2.621 .010 Interaction Facilitating Conditions x ‐.119 .060 ‐.119 ‐1.975 .050 Social Influence Standardized Current IT use .041 .069 .041 .593 .554 Standardized Social Influence .088 .068 .088 1.303 .195 a. Dependent Variable: Intention to use (stand)

Compared to the pre‐survey analysis, the moderators familiar with Smart Homes and age are excluded from the post‐survey. Instead of familiar with Smart Homes having a moderating effect on two relations, the post‐survey results indicate that current IT use now influences two relationships. These two are the relations of perceived usefulness and perceived ease of use with intention to use. The final moderating effect is caused by the influence of social influence on the relation between facilitating conditions and intention to use. The standardized coefficient of this interaction variable is negative with ‐.119, whereas the interaction effects of current IT use have coefficients of ‐.186 and .163. All variables are significant since their p‐values are below .05. The significance of the interaction effect involving social influence reflect a value of 0.0496, indicating a barely,

76 but still significant value. Furthermore, due to the moderating effects, the explanatory power of the post model has increased from an R Square of .493 to a value of .564. This can be distinguished from Appendix K.2.

The moderating effects of the post‐ survey are displayed in Figure 7.8 up to 7.10. Figure 7.8 displays the moderating effect of current IT use on the relation between perceived usefulness and intention to use. The graph indicates that perceived usefulness is of more importance for participants with a low current IT use. The slope is far more steeper, which shows that when having a lower Figure 7.8 Moderating effect of current IT use on relation current IT use, the participants have a perceived usefulness and intention to use far more higher intention to use when they perceive the Smart Home technology as highly useful compared to participants with a high current IT use that perceive a high usefulness. Furthermore, it can be distinguished that the usefulness is of less importance for participants with a high current IT use. Those respondents already had a rather high intention to use and it increased only slightly when the Smart Home technology is perceived as more useful.

After experiencing the Virtual Reality Smart Home, it is shown that participants with a low current IT use have pretty much the same intention to use as before the experience, regardless of their perceived ease of use. This can be seen in Figure 7.9. However, participants with a high current IT use and a high perceived ease of use have a lower impact on the intention to use as before the VR Figure 7.9 Moderating effect of current IT use on relation experience. This is also shown by the between perceived ease of use and intention to use lower standardized coefficient (.163) compared to the pre‐survey data (.265).

Figure 7.10 shows that social influence dampens the positive relationship between facilitating conditions and intention to use. For participants with a low social influence, facilitating conditions are of more importance to increase their intention to use than for participants with a high social influence. It is remarkable that participants with a low social influence that experience Smart Home Figure 7.10 Moderating effect of social influence on relation facilitating conditions and intention to use

77 technology with high facilitating conditions, have a higher intention to use than participants who have a high social influence and the same high facilitating conditions. 7.4 Evaluation of the pre‐ and post‐model To test the change in intention to use between the pre‐ and the post‐survey a Paired Samples T‐test is utilized. Looking at Table 7.3, the questions of intention to use, which have been proven to be reliable and to measure the same aspect, are put next to one another in the Paired Samples T‐test. A Paired Samples T‐test is often used when the sample size is sufficiently large and representative, when pre‐ and post‐survey are identical and when the same group of participants show a change in behaviour as a result of the intervention (Fisher Box, 1987). The significance of Table 7.3 is less than .05, which implies that there is a statistically significant difference between all the pairs. It indicates that after the intervention of the Virtual Reality experience, all three means of the intention to use variables increased, as shown in Table 7.3. Overall, it can be suggested that the intention to use consisting of these three questions has increased after the intervention, which is in line with the Paired Samples T‐test of the complete and combined intention to use variable, presented in Table 7.4. Both Table 7.3 and 7.4 show a high correlation, which means that except from being significant, the relations are also very strong. The correlations are positive which indicates that if one is increasing the other will increase too. The intention to use is increased after the experiment, which suggests that experiencing a Virtual Smart Home indeed changes the intention to use of the participants.

Table 7.3 Paired Samples T‐test of individual intention to use questions

Paired Differences Std. Std. 95% Confidence Std. Error Corre Std. Error Interval of Difference Sig. Mean Dev. Mean lation Sig. Mean Dev. Mean Lower Upper t (2‐t)

Pair 1 Know more 3.917 .905 .078 about SHt – .528 .000 ‐.211 .789 .068 ‐.346 ‐.075 ‐3.078 .003 Know more 4.128 .656 .057 about SHt

Pair 2 SHt installed 3.451 1.083 .094 in home – .609 .000 ‐.489 .918 .080 ‐.646 ‐.331 ‐6.140 .000 SHt installed 3.940 .983 .085 in home

Pair 3 Purchase SHt 3.203 1.166 .101 in future ‐ .720 .000 ‐.241 .836 .073 ‐.384 ‐.097 ‐3.318 .001 Purchase SHt 3.444 1.047 .091 in future

78 Table 7.4 Paired Samples T‐test of combined intention to use questions

Paired Differences Std. Std. 95% Confidence Std. Error Corre Std. Error Interval of Difference Sig. Mean Dev. Mean lation Sig. Mean Dev. Mean Lower Upper t (2‐t)

Pair Intention to 3.524 .892 .077 1 use ‐ .644 .000 ‐.313 .701 .061 ‐.433 ‐.193 ‐5.155 .000 Intention to 3.837 .737 .064 use

7.5 Conclusion The linear regression showed that all independent constructs indeed have significant individual relations to the dependent construct of intention to use. Furthermore, intention to use is also specifically measured in the surveys through three questions, which all had a sufficient Cronbach’s alpha. Predicting intention to use in twofold can only contribute to a higher reliability of the user acceptance model. Both the unique effects as well as the R Square have increased in the user acceptance model of the post‐survey. The moderators are not all proven significant. The linear regression of the pre‐survey showed significant results for age, technological knowledge (both current IT use and familiar with Smart Homes) and social influence. The post‐survey’s linear regression only showed significance for the moderators current IT use and social influence. Gender has shown non‐significant in both the pre‐ and post‐survey model. No further conclusions and analyses concerning this moderator are therefore included. 60.1% (pre‐survey) and 56.4% (post‐survey) of the total variance is explained by the constructs perceived usefulness, perceived ease of use and facilitating conditions and the moderators age, technological knowledge (current IT use and familiarity with Smart Homes) and social influence. The explanatory power of the model is rather strong according to Chin (1998).

The interactions show that when having a low social influence, the perceived usefulness of the Smart Home technology gets more important compared to those that are highly influenced. Likewise, participants with a low current IT use are more influenced by the perceived usefulness than participants with a high current IT use. The results concerning social influence might seem salient, however those who are highly affected by their surrounding focus more on what others tell them instead of determining the usefulness of the product on their own.

Participants at a higher age are more influenced by adequate facilitating conditions for their intention to use. The same goes for participants with a low familiarity with Smart Homes, they are more influenced by better facilitating conditions in order to obtain a higher intention to use. Furthermore, facilitating conditions seem more important to participants with a low social influence, even to the extent that participants with low social influence who experience Smart Home technology with high facilitating conditions, have a higher intention to use compared to participants with both a high social influence and facilitating conditions.

The entire moderator of technological knowledge, of which current IT use and familiar with Smart Homes are part, affects the relation between perceived ease of use and

79 intention to use. The perceived ease of use is of more importance when the participants have a low familiarity with Smart Homes. Those participants even have a higher intention to use at a high perceived ease of use compared to the participants with a high familiarity with Smart Homes. It attracts attention that by looking at the relationship between perceived ease of use and intention to use influenced by current IT use, participants with a high current IT use are mostly influenced by the perceived ease of use of the Smart Home technology. A higher perceived ease of use for those participants evidently leads to a higher intention to use compared to participants who have a low current IT use.

Only the relationship between ease of use and current IT use is proven significant for both the pre‐ and post‐survey analyses. The relationship between ease of use and current IT shows almost the same graph (pre‐ and post‐survey) for participants with a high perceived ease of use and a high current IT use. The intention to use decreases slightly compared to the pre‐survey. This could indicate that after experiencing Smart Home technology, the intention to use of these participants has decreased. However, when comparing these results to the overall results of the Paired Samples T‐test on intention to use, it shows that the intention to use of both the individual survey questions and the combined variable have increased instead of decreased. Therefore, the influence of current IT use on intention to use is only minor compared to the overall result, which is also shown by the low correlations in both the pre‐ and post‐survey data.

The model proven by this thesis differs from the model defined in Chapter 3. Figure 7.11 shows the newly developed model, in which all variables displayed are found to be significant in either the pre‐ or the post‐survey analyses or in both. The means of intention to use show an increase in the Paired Samples T‐test, which is in line with the data retrieved from the Virtual Reality questions in the post‐survey. The knowledge about Smart Home technology has increased and more importantly, the willingness to install Smart Home technology and to purchase Smart Home technology in the future has increased. All these findings conclude that the knowledge and understanding of Smart Homes and Smart Home technology can be increased by a Virtual Reality experience. In addition, participants’ intention to use Smart Home technology has increased through the experiment and therefore the acceptance of Smart Homes as well.

Requisites

Perceived Usefulness

Perceived Ease Intention to of Use use

Facilitating Conditions

Personal Characteristics

Technological Technological Age Knowledge – Knowledge – Social Influence Current IT Use Familiar with SH

Figure 7.11 The validated user acceptance model for Smart Homes

80 8 Conclusions and recommendations 8.1 Conclusions The aim of this thesis was to identify the factors that affect intention to use and to analyse whether experiencing Smart Home technology can lead to an enhanced user acceptance and implementation of Smart Homes. Furthermore, the interaction between the respondents and the Virtual Reality tool was examined. The validated user acceptance model as created for this thesis lays emphasis on the variables necessary to examine the intention to use, rather than the actual use. Therefore, the actual enhancement of the implementation of Smart Homes is hard to distinguish, however based on the thesis there are several assumptions which can be adopted in this regard. For starters, several studies indicate that intention to use is the most important predictor of actual use. Intention to use encompasses the degree to which an individual believes that he or she is intended to use Smart Home technology. Within the admitted questions of the surveys concerning this construct, the consideration of purchasing Smart Home technology was included. This type of questions are interesting to ask, since it provides part of the connection to actual use. Participants are likely to see the benefits of the proposed product, however whether they are willing to spend money is even one step further in order to underwrite the implementation. Therefore, including this kind of questions contributes in fulfilling the aim of this thesis and thereby answering the research questions at hand. As a reminder, the research question is formulated below:

Which factors affect intention to use and to what extent can user experience with sensor technology through Virtual Reality enhance the user acceptance and contribute to the implementation of Smart Homes?

MyriaNed turned out to be the most appropriate and suitable representative of sensor technology for automatic appliances. Therefore, it is safe to say that results obtained in this thesis are not only bound to MyriaNed applications. There are differences when implementing wireless sensor networks, however MyriaNed provides an overall impression of the feasible possibilities. Despite the fact that MyriaNed is explicitly mentioned in the main research question, the participants of the experiment were not informed about this technology, since the technology is just a mere tool in order to provide comfort. Besides the technology, applying Virtual Reality in order to present the user experience provided an important asset in getting users acquainted with Smart Homes.

The scientifically substantiated claim that results obtained in Virtual Reality strongly correspond to results in real‐life situations supports the fact that Virtual Reality can be deployed as a power tool to expose potential relations of scientific models. In addition, it can be applied in order to create better understanding and to build a bridge between selling and buying parties in different sectors. As understanding is highly related to user acceptance, Virtual Reality might be the missing link for many companies or organizations in bringing their product on the market. Vice versa, consumers are more aware of what

81 they are potentially buying. In that perspective, possible appliances and benefits are endless.

Based on the data obtained from the 133 participants performing the experiment, the main research question can be answered. Every respondent completed a pre‐survey, experienced a Smart Home in Virtual Reality, and a post‐survey. The results of the experiment indicated that a user experience with sensor technology can enhance the user acceptance of Smart Homes. There was a highly significant improvement distinguished in curiosity, intention of installing, and considering a future purchase of Smart Home technology. Consequently, the entire intention to use construct showed a significant increase after experiencing the Smart Home in Virtual Reality. What these results mean for the implementation of Smart Homes will be discussed in Section 8.2. Based on the experiment results, it is assumed that perceived ease of use has the highest influence on intention to use. This result, which is in line with the findings of Vastenburg et al. (2007), might seem surprising, however due to the complexity of the technology and the appliances involved this can be clarified. Furthermore, as the technology is automatic for the most part, human interaction is minimal. Therefore, control is more or less out of their hands, causing respondents to indicate that the perceived ease of use should be high. The perceived usefulness has a considerable effect on the intention to use. Whereas there was a clear distinction between the two independent variables in the results of the pre‐survey, in the post‐survey these influences are similar. Despite small differences, overall perceived ease of use was still indicated as most important influencer. Facilitating conditions clearly had the least effect on intention to use. This might be explained by the fact that based on empirical results, in UTAUT facilitating conditions is directly linked to actual use. This could indicate that facilitating conditions have a stronger relationship with actual use compared to intention to use.

The results of the analysis indicate that older participants require more and better facilitating conditions to increase their intention to use, as they are more dependent of the additional support. Social influence affects the same relationship, as well as the relation between perceived usefulness and intention to use. Whereas the latter relation was already proven through scientific research, the interaction effect on the relation between facilitating conditions and intention to use was new. It turns out that those participants who are less influenced by their social environments, pay more attention to the content of the support network behind the Smart Home technology, and are therefore more affected by the facilitating conditions. On the contrary, those who are highly influenced are less concerned with the facilitating conditions regarding Smart Home technology, as relevant others influence their opinion about the technology. The only moderating effect verified for both the pre‐ and post‐survey model influences the relation between perceived ease of use and intention to use, being the current IT use. Participants are more intended to use Smart Home technology when they perceive it as easy to use and they already utilize more innovative technologies. The post‐survey indicated that the number of technologies used also influences the relation of perceived usefulness and intention to use. The results of the analyses indicated that the degree in which a person is familiar to Smart Homes affects the relationships of perceived ease of use and facilitating conditions with intention to use. Since the survey questions of technological knowledge already measured diverging areas of the moderator, technological knowledge has been split up into two separate moderating effects, being

82 the current IT use and the familiarity of people with Smart Homes. Both interactions affect the relation between perceived ease of use and intention to use. As indicated from the results, participants that lack knowhow on Smart Homes or possess multiple innovative technologies are more intended to use Smart Home technology if they perceive these technologies as easy to use. In contrast to the preconceived model of Chapter 3, which suggested that technological knowledge had a moderating on perceived ease of use and facilitating conditions, current IT use shows a connection to perceived usefulness. In this connection, it is shown that for those who use less innovative technologies, it is crucial to comprehend the usefulness of Smart Home technology. Besides age and social influence, familiarity with Smart Homes is crucial for the relation between facilitating conditions and intention to use. The proposed moderating effect of gender turned out to be invalid, as apparently there was no significant difference between males or females. What the results can mean for the implementation of Smart Homes will be discussed in Section 8.3. The final model verified within this thesis is drawn up in Figure 8.1.

Perhaps, due to the fact that most respondents were highly related to the Eindhoven University of Technology, their high education and knowledge levels affected the results shown within this thesis. However, since the Virtual Reality technology is rather new and most respondents never experienced anything comparable, it is more likely that their response to the technique is less influenced by their education or knowledge. In that regard, besides developing a model, the results obtained from observations and experiences during the experiment expose an equally appealing outcome. It is expected that Virtual Reality will play a far more significant role within society in the near future and people will get more into contact with VR on a daily basis. It is the responsibility of scientists and involving parties to show these results to companies and governments, encouraging them to use this technology to their own benefits and to broaden their spectrum in fulfilling the user needs of society. In the end, that should be the main goal in order to address the individual citizen.

Requisites

Perceived Usefulness

Perceived Ease Intention to of Use use

Facilitating Conditions

Personal Characteristics

Technological Technological Age Knowledge – Knowledge – Social Influence Current IT Use Familiar with SH

Figure 8.1 The user acceptance model regarding Smart Homes and Smart Home technology

83 8.2 Discussion The main drawback exposed when using Virtual Reality is the nausea people experienced while performing the activities. Almost one out of six respondents felt some kind of uneasiness. On the other hand, a large majority did not experience the least bit of unstableness or sickness. The nauseous effect can prevent the implementation of Virtual Reality technology on a larger scale. However, this might change over time. Daan Roosegaarde, one of the worldwide leaders on design and innovation, recently highlighted the fact that the human body is remarkably able to adapt to new circumstances. In the Dutch television program TV Show, he said the following regarding to this phenomenon (translated into English, (Roosegaarde, 2015)):

‘’It is interesting to see how technology makes the impossible possible. The first people on the very first train, they became nauseous, they had to puke, because their body was not yet used to all that, and now, yes, we just enter without any problems, so as a person, we are constantly changing and that, exactly that can be used, right, to make that world better, or more beautiful or poetic.’’

In order to be innovative, challenges need to be overcome. Although Virtual Reality is past the developing stage and already mature, the technology needs to be flawless to encourage multidisciplinary usage. As indicated, the human body must not be underestimated and when the technology evolves, it is most likely that these kind of inconveniences can be minimised or even get disposed. In addition, game designers are designated to solve the issue of motion sickness when playing video games. As the game community is highly related to Virtual Reality, their efforts of changing the perceived gameplay are crucial for Virtual Reality implementations. Potential solutions are to change the speed of the camera movement, the field of view, the saturation of colours and different perspectives applied. However, everyone perceives this differently. As highlighted after the experiment, the toleration level differs for each individual regarding the Virtual Reality sickness. Sickness gained in Virtual Reality settings slightly differs from motion sickness as the respondent standing in front of the Powerwall is often stationary, although he or she has a compelling feeling of moving since he/she moves visually through the virtual environment. The uncomfortable feeling affects the vestibular system and results in sickness or instability (LaViola, 2000; Corriea, 2013). As this downside effect does not only concern Virtual Reality appliances, it is likely to say that solutions will be introduced within the upcoming future. As VR can also be accessed through an Oculus Rift, real‐time motion can be added to minimize the sickness effect for the current time being.

When conducting a rather complex user acceptance model for both a pre‐ and post‐ survey situation, the question arises on how to compose one holistic model from the gained results. Despite the fact that the constructs of the model are all significant and valid for both the pre‐ and post‐survey results, the moderating effects differ considerably. Since the majority of the people never experienced Virtual Reality before, it could be that the pre‐survey model entails the most important information. However, after

84 experiencing, different moderating effects appeared significant. In composing the final model, should these results be added to one another, or should only the consistent results from both the pre‐ and post‐model be included? Eventually, the model should be a reflection of the entire experiment. When interpreting the models from other well‐ known technology acceptance theories like TAM or UTAUT, they all indicate the variables that need to be taken into consideration when putting a new product on the market. This suggests that the models are most useful in the preliminary phase, however since actual use is included, the tendency of those models is not only limited to that preceding phase. As actual use is not explicitly within the scope of the created model for this thesis, it can be discussed whether the same conditions apply. Since the total experiment consisted of both the pre‐ and post‐survey, it is assumed within this thesis that the moderating effects from both models are included in the final model. In adopting this theory, only gender is left out as moderator when conducting research on the user acceptance of Smart Homes and its involving technologies. 8.3 Recommendations and directions for future research The success of introducing Smart Home technology through a Virtual Reality home setting can become a catalyst for new areas of VR implementations. Specifically in combination with Smart Homes or other types of real estate, and in a wider perspective for the entire built environment and through the Building Information Modeling (BIM) process. In BIM, an up‐to‐date digital representation of a developed model provides insight in decisions concerning planning, design, construction and management. This is established in order to improve the collaboration and productivity and to make sure that all parties involved in the building process operate from the same information supply and communication is regulated. As the potential clients are a major part of the process, Virtual Reality can help significantly to enhance the user acceptance and to improve the communication between the parties. Through VR, construction or design decisions can become visible and tangible for the end user. As highlighted in this thesis, Virtual Reality can exhibit all kinds of possibilities to implement Smart Home technology into the home environment and can inspire and challenge clients to think about what Smart Home technology could mean for their home environment. By introducing Smart Home technology to clients at the start of a development project, the technology will gain user acceptance and can be adopted earlier in the building process. Customers are getting a home experience and thereby understand the consequences of Smart Home technology and even more importantly, experience the resulting comfort level. Through this, they get a better understanding of their own needs and desires and how this can be integrated into their home. Furthermore, this creates a pull‐market strategy rather than a push‐market strategy as discussed in the introduction of this thesis. The magnitude and importance of experiencing something at first‐hand shows how many benefits this can bring to both the buying and selling party.

By making the technology perceivable and tangible, the intention to use increases significantly. Due to the fact that survey results were almost identical for senior participants, VR can contribute in convincing elderly to apply Smart Home technology, which can extend their independent living. This might lower the threshold and accessibility of these technologies, as they can for instance weigh their privacy concerns to their perceived increase in comfort. The essence of applying Smart Home technology

85 becomes more and more visible. Heijmans, as one of the first real estate constructors, is currently developing new buildings in which home automation is included (Heijmans, 2015; van Berlo, 2015). Although this development is very promising, this push‐market strategy differs from the pull‐market approach highlighted within this thesis. By letting end‐users experience the home automation, more specific and tailored results can be obtained and less dispensable costs are made by the company. However, considerations have to be made since the VR experience process is rather time‐consuming. Nevertheless, the fact that Heijmans undertakes this step indicates the relevance of the topic at hand and involving VR would provide a valuable, extra dimension for this discussion and the implementation of Smart Homes.

Besides Smart Home technology, far more general appliances are imaginable for a Virtual Reality implementation. Clients who are thinking of buying a new house can experience their potential residence at the office of the real estate developer or agent before it is built and make changes without major cost involvements. In addition, the architects themselves can experience the houses they designed on forehand. Walking through a real‐life model gives a far better overview compared to a 3D‐computer model. Due to the different perspective, a house can be seen from a different angle and minor errors might show up, giving the architect the possibility to make adjustments. Same goes for all other companies involved, when implementing Virtual Reality in BIM, every party can get a good image of the expected end results. Blueprints are often read incorrect or show practical mistakes, when keeping in mind the end results and by falling back on a Virtual Reality model, solutions can be found more efficiently and more quickly, reducing delay and failure costs. Furthermore, building and communication mistakes can be reduced, both in design and in the actual development process, because a general consensus is reached. Virtual Reality can also be used to get an overview of an entire neighbourhood, for example the municipality or an urban planner can use it to decide if building plans and renovations are suitable for a certain neighbourhood. Furthermore, development companies can use Virtual Reality as communication or marketing tool to sell their plans to clients and the municipality. All in all, VR can become a major asset to Smart Homes and the entire building process and environment in general.

Whereas sensor developers mainly focus on developing the technology for the market, their role within the implementation is crucial as well. Typically, these companies state that they are less concerned about the implementation possibilities rather than the technological finesse, since their only objective is to sell their technology. However, in order to obtain widespread attention, they should provide interesting companies with numerous potential appliances. Assigning the marketing department to attract potential buyers is not enough, as it is about the tangible applications customers want to perceive and use. Sensor possibilities are endless, however these noticeable products are required to trigger people to evolve their own ideas about the technology implementation. By assigning a R&D department to the core business of a sensor company, the gap between the technology and the involvement of customers can be diminished. Obviously, it is up to the business itself whether this department should be included or outsourced. For MyriaNed, developer Devlab highly cooperates with its engineering agency member Van Mierlo, to bring promising applications on the market. Multiple test setups and research areas are designated in order to demonstrate and improve MyriaNed implementations. As it is difficult for sensor technology to clearly distinguish itself as best solution from all

86 the others available in the market, it should be their implementations that provide them unique selling points over the other developers. Just like the Smart Homes, the centre point of the discussion should be shifted from the technological point of view towards to social perspective. In this light, resellers or distributers can also provide an adequate solution. Perhaps, they are more specialised in presenting the product to the customers. For a long time, it was questioned which type of company would bring the Smart Home technology to the next level of widespread implementation. Would it be the developers, resellers, engineers, housing corporations or building companies that undertake the first step? For MyriaNed, it would be best to introduce and connect the technology to large players in the market, like Philips or BAM. The developers can take advantage of the current growth and interest in domestic technology and encourage Smart Home prototype developers to use their product. Not only could sensor technology be implemented into newly built environments, also the entrance in the renovation market would be highly recommended. The large real estate vacancy rate might be diminished by renovating current houses and offices into individually tailored homes. As experiencing Smart Home technology can stimulate people to overthrow their privacy concerns, it might be possible that less preferred domestic characteristics are compensated by high‐ end supporting technology.

Although this study provides interesting results in applying Virtual Reality for user acceptance research, some challenges for future research remain. The model created in this thesis excludes actual use, since the study is not longitudinal and actual use of Smart Homes is difficult to examine as there are major costs involved. A key challenge would be to include VR in a longitudinal study in which actual use is also included. The setup could be identical, only the experiment needs to be extended from three (pre‐survey, experience, post‐survey) to possibly four or five test moments. By identifying the variables that have a direct or indirect effect on the actual use of Smart Homes, a more holistic overview of the situation can be provided.

Regarding the experiment, for future research it would be critical to check the ratio of the sample size to see whether every social group is more or less included within the representation. Since the experiment was held at the Eindhoven University of Technology, the majority of the respondents consisted of students, professors or other university‐related staff members. The composition of the sample size would perhaps have been more in proportion if the experiment would have taken place in the city centre and random people were invited. Due to the immobility of the VR setup this is hard to realize, but not impossible.

Future research would also be encouraged in the field of VR, or related areas like augmented or mixed reality. Adding real‐time daylight or shadows in the VR environment could enhance the realism and experience. Graphically speaking, most respondents were already impressed by the current quality of the representation. Therefore, the graphics should not be the main focus point when improving the environment. Although render images of Autodesk programs present far more accurate reflections, the main critical points are within the lighting and shadowing capacities. Remarkably, no one in the experiment commented on the fact that the shadows stayed in place although the sliding doors were opening, however in communicating with clients and participants, architects would like to include those effects to provide valuable information. This might enhance

87 implementation possibilities within this field or other fields of the built environment. Besides research on creating an even more realistic representation, also research on the interaction with Virtual Reality would be recommended. This interaction is twofold, as the translation between the programs used can be improved as well as the usage of interaction tools inside the VR world. Since the VR platform is rather new, difficulties need to be overcome in exporting models from 3D‐software into the virtual world. Research in this field could stimulate a more natural usage of Virtual Reality. Developing interaction tools for VR appliances is outside the scope of the research areas, however research can be conducted on which tools would contribute the most to close the gap between virtual and real life. In the end, the fact that respondents were able to move freely in front of the Powerwall, grabbing elements and relocating them, contributed highly to the interaction and has proven vital for the entire experience. In this light, using Oculus Rift instead of the Powerwall is worth looking into as respondents’ entire field of vision is enclosed within a Rift. Furthermore, specific research could be conducted on how to oppose nausea during the Virtual Reality encounter. Specific studies on motion sickness regarding the game industry can provide valuable insights on this matter.

Another field of future research lays in the area of the type of technology used within this thesis. Instead of examining the user acceptance of fully automated appliances, semi‐ automatic appliances and networks of devices can be investigated. Although being currently in the phase of user acceptance, using a smartphone or tablet to control smart technologies significantly differs from the automatic technology as applied within this study. Creating a VR Smart Home in which these types of technologies are implemented is definitely conceivable. Through this, more interactions will be included and the experience will become even more interactive. As the Smart Home concept is very promising, more research should be conducted regarding the energy performance and costs of a Smart Home. For the corporate sector, a cost‐benefit analysis would be very helpful to compare and contrast various Smart Home aspects. Finally, it is imaginable that future research will focus on other areas in which VR can be implemented, like smart cities, psychology, medical and health environments, simulations or BIM‐related activities.

88 9 References Abi‐Char, P. R., Mhamed, A., El‐Hassan, B. & Mokhtari, M., 2010. A Flexible Privacy and Trust Based Context‐Aware Secure Network. In: Y. Lee, et al. eds. Aging Friendly Technology for Health and Independence. Seoul: Springer, pp. 17‐23.

Acapela group, 2015. Acapela group. [Online] Available at: http://www.acapela‐group.com/ [Accessed April 2015].

Agueci, G., 2015. MyriaNed as Smart Building Platform, Eindhoven: Van Mierlo Ingenieursbureau, Intelligente elektronica voor meet‐ en regeltechniek.

Ajzen, I. & Fishbein, M., 1980. Understanding attitudes and predicting social behavior, Englewood Cliffs: Prentice Hall.

Alam, R. M., Reaz, M. B. & Ali, M. A., 2012. A Review of Smart Homes—Past, Present, and Future. Systems, Man, and Cybernetics, Part C: Applications and Reviews, 42(6), pp. 1190‐ 1203.

Alavifar, A., Karimimalayer, M. & Anuar, M. K., 2012. Structural equation modeling VS multiple regression. Engineering Science and Technology: An International Journal (ESTIJ), 2(2), pp. 326‐329.

Aldrich, F. K., 2003. Smart Homes: Past, Present and Future. In: R. Harper, ed. Inside the Smart Home. London: Springer‐Verlag London Limited, pp. 17‐39.

Allameh, E. et al., 2012. The role of Smart Home in Smart Real Estate. Journal of European Real Estate Research, 5(2), pp. 156‐170.

Allameh, E. et al., 2013. Smart Homes from vision to reality: eliciting users' preferences of Smart Homes by a virtual experimental method. Tokyo, Japan, ICCBEI, pp. 297‐305.

Allameh, E. et al., 2014. Modeling users' Work Activities in a Smart Home. Procedia Environmental Sciences, Volume 22, pp. 78‐88.

Allameh, E., Jozam, M. H., de Vries, B. & Timmermans, H., 2014. Smart Home Design. [Online] Available at: http://www.tue.nl/onderzoek/strategic‐area‐health/onderzoek/smart‐ environment/smart‐home‐design/

Almende, 2010. CCF2 ‐ Independent Living. [Online] Available at: http://www.almende.com/radu‐ serban?p_p_auth=rFVdbXg7&p_p_id=101&p_p_lifecycle=0&p_p_state=maximized&p_p_ mode=view&_101_struts_action=%2Fasset_publisher%2Fview_content&_101_assetEntry Id=16098&_101_type=content&_101_urlTitle=ccf2‐independent‐living&redirect

Almende, 2011. ALwEN milestones. [Online] Available at: http://www.almende.com/radu‐ serban?p_p_auth=rFVdbXg7&p_p_id=101&p_p_lifecycle=0&p_p_state=maximized&p_p_ mode=view&_101_struts_action=%2Fasset_publisher%2Fview_content&_101_assetEntry Id=14142&_101_type=content&_101_urlTitle=alwen‐milestones&redirect=http%3

89 Ansari, S. & Chandak, M. B., 2014. Design and Implementation of Human Computer Interaction Using Brain Waves Electric Signals. Kuala Lumpur, IOSR Journal of Computer Science, pp. 24‐28.

Arcelus, A., Jones, M. H., Knoefel, F. & Goubran, R., 2007. Integration of Smart Home Technologies in a Health Monitoring System for the Elderly. Niagara Falls, IEEE, pp. 820‐ 825.

Atzori, L., Iera, A. & Morabito, G., 2010. The Internet of Things: A survey. Computer Networks, Volume 54, pp. 2787‐2805.

Augusto, J. C., 2010. Past, Present and Future of Ambient Intelligence and Smart Environments. In: Agents and Artificial Intelligence. Berlin Heidelberg: Springer, pp. 3‐15.

Augusto, J. C. & Nugent, C. D., 2006. Smart Homes Can Be Smarter. In: J. C. Augusto & C. D. Nugent, eds. Designing Smart Homes: The Role of Artificial Intelligence. Berlin: Springer, pp. 1‐15.

Baarda, D. B. & De Goede, M. P., 2006. Basisboek Methoden en technieken: Handleiding voor het opzetten en uitvoeren van kwantitatief onderzoek. Vierde druk ed. Groningen/ Houten: Wolters‐Noordhoff.

Bagozzi, R. P., 2007. The legacy of the Technology Acceptance Model and a proposal for a paradigm shift. Journal of the Association for Information Systems, 8(4), pp. 244‐254.

Bal, M., Shen, W. & Hao, Q. X. H., 2011. Collaborative Smart Home Technologies for Senior Independent Living: A Review. Lausanne, IEEE, pp. 481‐488.

Balta‐Ozkan, N., Boteler, B. & Amerighi, O., 2014. European smart home market development: Public views on technical and economic aspects across the United Kingdom, Germany and Italy. Energy Research & Social Science, Volume 3, pp. 65‐77.

Balta‐Ozkan, N., Davidson, R., Bicket, M. & Whitmarsh, L., 2013. Social barriers to the adoption of Smart Homes. Energy Policy, Volume 63, pp. 363‐374.

Balta‐Ozkan, N., Davidson, R., Bicket, M. & Whitmarsh, L., 2013. The development of smart homes market in the UK. Energy, Volume 60, pp. 361‐372.

Belley, C., Gaboury, S., Bouchard, B. & Bouzouane, A., 2014. An efficient and inexpensive method for activity recognition within a smart home based on load signatures of appliances. Pervasive and Mobile Computing, Volume 12, pp. 58‐78.

Bergvall‐Kåreborn, B., Ihlström Eriksson, C., Ståhlbröst, A. & Svensson, J., 2009. A Milieu for Innovation ‐ Defining Living Labs. New York, ISPIM.

Bernheim Brush, A. J. et al., 2011. Home automation in the wild: challenges and opportunities. New York, NY, CHI '11, pp. 2115‐2124.

Berrios, J. P., 2008. WSN – Sensor Network Platform. [Online] Available at: https://jeanpaulberrios.wordpress.com/2008/09/08/wsn‐plataform/

90 Bierhoff, I. et al., 2007. Smart home environment. In: P. R. Roe, ed. Towards an inclusive future; Impact and wider potential of information and communication technologies. Brussels: COST, pp. 110‐156.

Bouchon‐Meunier, B. et al., 2015. SHALL ‐ The Smart Home Ambient Living Lab, Paris: European Commission.

Busnel, P. & Giroux, S., 2010. Security, Privacy, and Dependability in Smart Homes: A Pattern Catalog Approach. In: Y. Lee, et al. eds. Aging Friendly Technology for Health and Independence. Seoul: Springer, pp. 24‐31.

Caine, K. E., Rogers, W. A. & Fisk, A. D., 2005. Privacy perceptions of an aware home with visual sensing devices. Orlando, FL, SAGE Journals, pp. 1856‐1858.

Capitanelli, A., Papetti, A., Peruzzini, M. & Germani, M., 2014. A Smart Home Information Management Model for Device Interoperability Simulation. Procedia CIRP, Volume 21, pp. 64‐69.

Caprani, N. et al., 2012. Technology Use in Everyday Life: Implications for Designing for Older Users. Galway, Ireland, iHCL, pp. 16‐21.

Carr, J. J. & Brown, J. M., 1998. Introduction to Biomedical Equipment Technology, Third Edition, Upper Saddle River, NJ: Prentice Hall.

CCV, 2011. Hoofdstuk 5: Welk onderzoeksmodel kies je?. In: Effecten van toezicht en handhaving meten: Een handreiking. Utrecht: Centrum voor Criminaliteitspreventie en Veiligheid, pp. 45‐57.

Chan, M., Campo, E., Estève, D. & Fourniols, J., 2009. Smart homes ‐ Current features and future perspectives. Maturitas, Volume 64, pp. 90‐97.

Chan, M., Estève, D., Escriba, C. & Campo, E., 2008. A review of smart homes—Present state and future challenges. Computer Methods and Programs in Biomedicini, 91(1), pp. 55‐81.

Chess, 2014. Wireless by Nordic. [Online] Available at: http://chess.nl/wireless‐by‐nordic/ [Accessed 30 June 2015].

Chin, W. W., 1998. The Partial Least Squares Approach for Structural Equation Modeling.. In: G. A. Marcoulides, ed. Modern Methods dor Business Research. Mahwah, NJ: Lawrence Erlbaum Associates, pp. 295‐336.

Christ, R. D. & Wernli, R. L., 2013. The ROV Manual, Second Edition: A User Guide for Remotely Operated Vehicles. Oxford: Elsevier.

Cohen, J., Cohen, P., Aiken, L. S. & West, S. H., 2003. Applied multiple regression/correlation analysis for behavioral sciences. Hillsdale, N.J: L. Erlbaum Associates.

91 Corriea, A. R., 2013. How developers are trying to solve motion sickness in video games. [Online] Available at: http://www.polygon.com/2013/10/26/4862474/video‐games‐and‐motion‐ sickness‐dying‐light‐techland‐fps [Accessed 4 August 2015].

Courtney, K. L., 2006. Privacy and senior adoption of assistive technology in residential care, Columbia, MO: University of Missouri.

Dargie, W. W. & Poellabauer, C., 2010. Fundamentals of Wireless Sensor Networks: Theory and Practice. Chichester, UK: John Wiley & Sons Ltd..

Davidoff, S., Lee, M. K., Zimmerman, J. & Dey, A., 2006. Socially‐Aware Requirements for a Smart Home. Cambridge, UK, CiteSeer, pp. 41‐44.

Davis, F. D., 1986. PHD Thesis: A technology acceptance model for empirically testing new end‐user information system: theory and results, Massachusetts: Massachusetts Instutute of Technology, Sloan School of Management.

Davis, F. D., 1989. Perceived usefullness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), pp. 319‐340.

Davis, F. D., 1993. User acceptance of information technology: system characteristics, user perceptions and behavioral impacts. Machine Studies, Volume 38, pp. 475‐487.

Davis, F. D., Bagozzi, R. P. & Warshaw, P. R., 1989. User Acceptance of Computer Technology: A comparison of two Theoretical Models. Management Science, Volume 35, pp. 982‐1003. de Ruiter, E., 2013. De eerste stappen met MyriaNed. Netwerktechnologie, Issue 11, pp. 18‐19.

De Sanctis, M. et al., 2012. Interoperability solutions between smartphones and Wireless Sensor Networks. Rome, ESTEL, pp. 1‐6.

De Silva, C. W., 2007. Sensors and Actuators: Control System Instrumentation. Boca Raton, FL: CRC Press: Taylor & Francis Group.

De Silva, L. C., Morikawa, C. & Petra, I. M., 2012. State of the art of smart homes. Engineering Applications of Artificial Intelligence, Volume 25, pp. 1313‐1321.

DEMANES, 2013. Design, Monitoring and Operation of Adaptive Networked Embedded, s.l.: The DEMANES Consortium.

Demiris, G., 2009. Independence and Shared Decision Making: The Role of Smart Home Technology in Empowering Older Adults. Minneapolis, IEEE, pp. 6432‐6436.

Dengler, S., Awad, A. & Dressler, F., 2007. Sensor/Actuator Networks in Smart Homes for Supporting Elderly and Handicapped People. Niagara Falls, Ont., IEEE, pp. 863‐868.

DeVellis, R. F., 2012. Scale Development: Theory and Applications. Third edition ed. Thousand Oaks, CA: SAGE Publications.

92 DevLab, 2011. Sensoren houden oogje op zelfstandig wonende ouderen, Deventer: Etotaal.

DevLab, 2013. Maakt MyriaNed de gebouwcomputer overbodig? ‐ Wat is MyriaNed?. [Online] Available at: http://gebouwautomatisering.fhi.nl/images/Activiteiten/2013WK20%20‐ %2014mei%20Myrianed/MyriaNed.pdf

DiCarlo, A. S., 2010. Smart Homes (Home Automation), Glen Cove, NY: Livable New York Resource Manual.

Ding, D., Cooper, R. A., Pasquina, P. F. & Fici‐Pasquina, L., 2011. Sensor technology for smart homes. Maturitas, Volume 69, pp. 131‐136.

DUB, 2013. Meisjes kiezen voor Leiden en Utrecht, jongens voor Eindhoven. [Online] Available at: http://www.dub.uu.nl/plussen‐en‐minnen/2013/06/10/meisjes‐kiezen‐voor‐ leiden‐en‐utrecht‐jongens‐voor‐eindhoven.html [Accessed 15 July 2015].

D'Ulizia, A., Ferri, F., Grifoni, P. & Guzzo, T., 2010. Smart Homes to Support Elderly People: Innovative Technologies and Social Impacts. In: A. Coronato & G. De Pietro, eds. Pervasive and Smart Technologies for Healthcare: Ubiquitous Methodologies and Tools. Hershey, PA: IGI Global, pp. 25‐39.

Dutch Cochrane Centre (DCC), 2014. Beoordeling systematische review van randomised controlled trials, Utrecht: Dutch Cochrane Centre.

Ehrenhard, M., Kijl, B. & Nieuwenhuis, L., 2014. Marktverkenning: Barrières in marktadoptie van domotica, Enschede: University of Twente.

Ehrenhard, M., Kijl, B. & Nieuwenhuis, L., 2014. Market adoption barriers of multi‐ stakeholder technology: Smart homes for the aging population. Technological Forecasting & Social Change, Volume 89, pp. 306‐315.

EmenEm, 2013. Perspectief voor draadloze sensornetwerken in gebouwen en thuiszorg ‐ MyriaNed. Rotterdam, FHI ‐ Federatie van technologiebranches.

Engineersonline, 2013. Experimenteren met draadloze sensornetwerken. [Online] Available at: http://www.engineersonline.nl/nieuws/id21770‐experimenteren‐met‐ draadloze‐sensornetwerken.html

EnOcean, 2008. EnOcean wireless sensor module. [Online] Available at: http://www.adhocelectronics.net/download/EnOcean/STM100_Brochure.pdf

Fabian, B. & Feldhaus, T., 2014. Privacy‐preserving data infrastructure for smart home appliances based on the Octopus DHT. Computers in Industry, Volume 65, pp. 1147‐1160.

Fahn, C., Wu, M. & Liu, W., 2013. On the Use of Augmented Reality Technology for Creating Interactive Computer Games. In: R. Shumaker, ed. Virtual, Augmented and Mixed Reality: Systems and Applications, Part II. Las Vegas: Springer; HCI International 2013, pp. 353‐363.

93 Fang, H. et al., 2014. Human activity recognition based on feature selection in smart home using back‐propagation algorithm. ISA Transactions, 53(5), pp. 1629‐1638.

Fischione, C., 2009. IEEE 8IEEE 802.15.4, ZigBee, WirelessHART, ISA SP‐100, ROLL: Principles of Wireless Sensor Networks, Stockholm: Royal Institue of Technology ‐ KTH.

Fishbein, M. & Ajzen, I., 1975. Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research.. Readin, MA: Addison‐Wesley.

Fisher Box, J., 1987. Guinness, Gosset, Fisher, and Small Samples. Statistical Science, February, 2(1), pp. 45‐52.

Flynn, D. S., 1990. Traditional Sensors. In: D. R. Omstead, ed. Computer Control of Fermentation Processes. Boca Raton, FL: CRC Press, pp. 1‐15.

Frank, R., 2013. Understanding Smart Sensors. Norwood, MA: Artech House.

Freudenthal, A. & Mook, H., 2003. The evaluation of an innovative intelligent thermostat interface: universal usability and age differences. Cognition, Technology & Work, 5(1), pp. 55‐66.

Gaddam, A., Kaur, K., Mukhopadhyay, S. & Sen Gupta, G., 2009. Selection and Optimization of Wireless Sensors in a Smart Digital Home for the Elderly. Christchurch, IEEE, pp. 1382‐1386.

Gann, D., Barlow, J. & Venables, T., 1999. Digital Futures: Making Homes Smarter. York/Coventry: JRF/CIH.

Gardner, C. & Amoroso, D. L., 2004. Development of an Instrument to Measure the Acceptance of Internet Technology by Consumers. International Conference on System Sciences, Volume 37.

Gartner, 2014. Gartner Says Annual Smartphone Sales Surpassed Sales of Feature Phones for the First Time in 2013. [Online] Available at: http://www.gartner.com/newsroom/id/2665715

Gartner, 2015. Gartner Says Smartphone Sales Surpassed One Billion Units in 2014. [Online] Available at: http://www.gartner.com/newsroom/id/2996817

Gaul, S. & Ziefle, M., 2009. Smart Home Technologies: Insights into Generation‐Specific Acceptance Motives. Lecture Notes in Computer Science, Volume 5889, pp. 312‐332.

Gill, K., Yang, S., Yao, F. & Lu, X., 2009. A ZigBee‐Based Home Automation System. Transactions on Consumer Electronics, 55(2), pp. 422 ‐ 430.

Gomez, C. & Paradells, J., 2010. Wireless Homes Automation Networks: A Survey of Architectures and Technologies. Communications Magazine, 48(6), pp. 92‐101.

Gopalakrishnan Nair, N., Morrow, P. J. & Parr, G. P., 2011. Design Considerations for a Self‐Managed Wireless Sensor Cloud for Emergency Response Scenario. Liverpool, PGNet.

Gratton, D. A., 2013. The Handbook of Personal Area Networking Technologies and Protocols. New York: Cambridge University Press.

94 Grønland, M., 2010. User Acceptance of Information Technology: An Empiral Study of It's Learning, Trondheim: Norwegian University of Science and Technology.

Haase, J., 2013. Wireless Network Standards for Building Automation. In: C. Grimm, P. Neumann & S. Mahlknecht, eds. Embedded Systems for Smart Appliances and Energy Management. New York: Springer, pp. 53‐67.

Hair, J. F., Black, W. C., Babin, B. J. & Anderson, R. E., 2009. Multivariate Data Analysis: A Global Perspective. 7th ed. ed. Upper Saddle River, NJ: Prentice Hall.

Hamilton, C. & Sampath, V., 2011. Performance of ZigBee PRO Mesh Networks with Moving Nodes. PennScience Journal, pp. 29‐33.

Heck, W., 2009. Smart energy meter will not be compulsory. [Online] Available at: http://vorige.nrc.nl/international/article2207260.ece/Smart_energy_meter_will_not_be_ compulsory [Accessed 9 March 2015].

Heidari, M. et al., 2014. Smart‐BIM virtual prototype implementation. Automation in Construction, April, Volume 39, pp. 134‐144.

Heijmans, 2015. Heijmans levert koopwoning standaard met huisbesturing. [Online] Available at: http://heijmans.nl/nl/nieuws/heijmans‐levert‐eengezinskoopwoningen‐ standaard‐met‐huisbesturing/ [Accessed 3 August 2015].

Herman, P., 2014. Connected smart lighting, Eindhoven: Metatronics: For smarter electronic solutions.

HOMEMAKERSonline, 2012. How to save electricity in your home. [Online] Available at: http://www.homemakersonline.co.za/features/31/how‐to‐save‐electricity‐ in‐your‐home [Accessed 20 July 2015].

Honeywell, 2014. Smart homes of the future ‐ an industry view, Berkshire, UK: Honeywell International Inc..

Hudson, P., 2000. History by Numbers: An Introduction to Quatitative Approaches. London: Arnold.

Irish, J., 2005. Lecture on Instrumentation Specifications , St. John's, Newfoundland and Labrador: Fisheries and Marine Institute of Memorial University of Newfoundland.

Iyer, A., Kulkarni, S. S., Mhatre, V. & Rosenberg, C. P., 2008. A Taxonomy‐based Approach to Design of Large‐scale Sensor Networks. In: Y. Li, M. T. Thai & W. Wu, eds. Wireless Sensor Networks and Applications. New York, NY: Springer, pp. 3‐33.

Jeng, T., 2009. Toward a Ubiquitous Smart Space Design Framework. Journal of Information Science and Engineering, Volume 25, pp. 675‐686.

Kamel, S., 2004. Technology Acceptance Model, Cairo: The American University in Cairo.

95 Kaur, H. & Sharma, S., 2013. A Comparative Study of Wireless Technologies: Zigbee, Bluetooth LE, Enocean, Wavenis, Insteon and UWB. Hamirpur, India, RTCCE, pp. 273‐276.

Kavitha, R., Nasira, G. M. & Nachamai, N., 2012. Smart Home systems using wireless sensor network ‐ A comparative analysis. International Journal of Computer Engineering & Technology, 3(3), pp. 94‐103.

Kenny, T., 2005. Basic Sensor Technology. In: J. S. Wilson, ed. Sensor Technology Handbook. Oxford, UK: Elsevier, pp. 1‐21.

Khan, R. A. et al., 2012. Wireless Sensor Networks: A Solution for Smart Transportation. Journal of Emerging Trends in Computing and Information Sciences, 3(4), pp. 566‐571.

Kim, S., Jeong, Y. & Park, S., 2013. RFID‐based indoor location tracking to ensure the safety of elderly in smart home environments. In: S. Kim, Y. Jeong & S. Park, eds. Personal and Ubiquitous Computing. London: Springer, pp. 1699‐1707.

Kumar, V., 2013. Fundamentals of Pervasive Information Management Systems. Ney Jersey: John Wiley & Sons.

Kusiak, A., 2007. Innovation: The Living Laboratory Perspective. Computer‐Aided Design & Applications, 4(6), pp. 863‐876.

Kyriacou, P. A., 2009. Biomedical Sensors: Temperature Sensor Technology. In: D. P. Jones, ed. Biomedical Sensors. New York, NY: Momentum Press, pp. 1‐39.

Laerd Statistics, 2015. Linear Regression Analysis using SPSS Statistics. [Online] Available at: https://statistics.laerd.com/spss‐tutorials/linear‐regression‐using‐spss‐ statistics.php [Accessed 17 July 2015].

Lai, C., Huang, H., Liaw, S. & Huang, W., 2009. A study of user's acceptance in three‐ dimensional virtual reality applied in medical education. Bulletin of Educational Psychology, 40(3), pp. 341‐362.

LaViola, J. J., 2000. A discussion of cybersickness in virtual environments. ACM SIGCHI Bulletin, 32(1), pp. 47‐56.

Layton, J., 2009. 5 Energy‐hungry Appliances. [Online] Available at: http://science.howstuffworks.com/environmental/green‐ tech/sustainable/5‐energy‐hungry‐appliances.htm [Accessed 20 July 2015].

Lee, S. et al., 2013. Occupancy Prediction Algorithms for Thermostat Control Systems Using Mobile Devices. IEEE Transactions on Smart Grid, 4(3), pp. 1332‐1340.

Lertlakkhanakul, J., Choi, J. W. & Kim, M. Y., 2008. Building data model and simulation platform for spatial interaction management in Smart Home. Automation in Construction, pp. 948‐957.

Lewis, F. L., 2005. Wireless Sensor Networks. In: D. J. Cook & S. K. Das, eds. Smart Environments: Technology, Protocols and Applications. New York: John Wiley, pp. 13‐47.

96 Likert, R., 1932. A technique for the measurement of attitudes. Archives of Psychology, 22(140), pp. 1‐55.

Lönn, J. & Olsson, J., 2005. ZigBee for wireless networking, Linköping: Linköping University.

Lovecraft, H. P., 1973. Supernatural Horror in Literature. New York: Dover Publications.

Lu, J. et al., 2010. The smart thermostat: using occupancy sensors to save energy in homes. Zurich, Switzerland, ACM, pp. 221‐224.

Magnusson, L. & Hanson, E. J., 2003. Ethical issues arising from a research, technology and development project to support frail older people and their family carers at home. Health and Social Care in the Community, 11(5), pp. 431‐439.

Mahalik, N. P., 2007. Introduction. In: N. P. Mahalik, ed. Sensor Networks and Configuration: Fundamentals, Standards, Platforms, and Applications. Berlin: Springer, pp. 1‐18.

Ma, M., Jain, L. C. & Anderson, P., 2014. Future Trends of Virtual, Augmented Reality, and Games for Health. In: M. Ma, L. C. Jain & P. Anderson, eds. Virtual, Augmented Reality and Serious Games for Healthcare. Berlin Heidelberg: Springer, pp. 1‐6.

Martin, G., 2007. Wireless sensor solutions for home & building automation ‐ the successful standard uses energy harvesting, Cottonwood Heights, UT: EnOcean.

Mathur, A. & Newe, T., 2014. Comparison and overview of Wireless sensor network systems for Medical Applications. Liverpool, ICST.

Mayer, P., Volland, D., Thiesse, F. & Fleisch, E., 2011. User Acceptance of 'Smart Products': An Empirical Investigation. Wirtschaftinformatik Proceedings 2011, p. Paper 9.

McGrath, M. J. & Scanaill, C. N., 2014. Sensor Technologies: Healthcare, Wellness and Environmental Applications. New York, NY: Apress Media LLC.

McMahan, R. P., Bowman, D. A., Zielinski, D. J. & Brady, R. B., 2012. Evaluating Display Fidelity and Interaction Fidelity in a Virtual Reality Game. Visualization and Computer Graphics, 18(4), pp. 626‐633.

Ministerie van Economische Zaken, 2014. Besluit van 16 oktober 2014, houdende wijziging van het Besluit op afstand uitleesbare meetinrichtingen ten behoeve van de grootschalige uitrol van de slimme meter. Staatsblad van het Koninkrijk der Nederlanden, 28 Oktober.Volume 392.

Missaoui, R., Joumaa, H., Ploix, S. & Bacha, S., 2014. Managing energy Smart Homes according to energy prices: Analysis of a Building Energy Management System. Energy and Buildings, Volume 71, pp. 155‐167.

Molinari, F., 2011. Best practices Database for Living Labs: ‐ Overview of the Living Lab approach and Living Lab Best Practice Database Specification, Genoa: Alcotra.

Monecke, A. & Leisch, F., 2012. semPLS: Structural Equation Modeling Using. Journal of Statistical Software, 48(3), pp. 1‐32.

97 Nabi, M., Geilen, M. & Basten, T., 2012. Demonstrating on‐demand listening and data forwarding in wireless body area networks. Seoul, IEEE, pp. 82‐84.

National University of Singapore, 2015. Errors of Observation, Singapore: National University of Singapore.

NI, 2015. Analog Sample Quality: Accuracy, Sensitivity, Precision, and Noise, Austin, TX: National Instruments.

Nixon, M., 2012. A Comparison of WirelessHART™ and ISA100.11a, Round Rock, TX: Emerson Process Management.

NOS, 2014. Wat Chriet Titulaer goed voorspelde. [Online] Available at: http://nos.nl/artikel/592589‐wat‐chriet‐titulaer‐goed‐voorspelde.html

Nugent, C. D., McClean, S. I., Cleland, I. & Burns, W., 2014. Sensor Technology for a Safe and Smart Living Environment for the Aged and Infirm at Home. In: M. S. Hashmi, ed. Comprehensive Materials Processing. Shannon, County Clare: Elsevier Ltd., pp. 459‐472.

Nunnally, J., 1978. Psychometric Theory. 2nd ed. ed. New York: McGraw‐Hill.

Oculus VR, 2015. Rift: Next‐generation virtual reality. [Online] Available at: https://www.oculus.com/en‐us/rift/ [Accessed 5 August 2015].

Okeyo, G., Chen, L. & Wang, H., 2014. Combining ontological and temporal formalisms for composite activity modelling and recognition in Smart Homes. Future Generation Computer Systems, Volume 39, pp. 29‐43.

ONE‐NET, 2012. ONE‐NET Wireless Control for Everyone. [Online] Available at: www.one‐net.info/images/files/epri.ppt

Orengo, G., Saggio, G., Bocchetti, S. & Giannini, F., 2010. Advanced characterization of piezoresistive sensors for human body movement tracking. Paris, IEEE, pp. 1181‐1184.

Ott, R., Thalmann, D. & Vexo, F., 2007. Haptic Feedback in Mixed‐Reality Environment. The Visual Computer: International Journal of Computer Graphics, 23(9), pp. 843‐849.

Pallot, M., 2009. Methodologies for Engaging Users into Research & Innovation: The Living Lab way as an Open Innovation Ecosystem. Leiden, Netherlands, ICE.

Papageorgiou, P., 2003. Wireless Sensor Networks: An Overview, College Park, MD: University of Maryland.

Patel, R., Pariyani, S. & Ukani, V., 2011. Energy and Throughput Analysis of Hierarchical Routing Protocol (LEACH) for Wireless Sensor Network. International Journal of Computer Applications, 20(4), pp. 32‐36.

Peine, A., 2009. Understanding the dynamics of technological configurations: A conceptual framework and the case of Smart Homes. Technological Forecasting & Social Change, Volume 76, pp. 396‐409.

Ploennigs, J., Ryssel, U. & Kabitzsch, K., 2010. Performance Analysis of the EnOcean Wireless Sensor Network Protocol. Bilbao, ETFA, pp. 1‐9.

98 Pountourakis, I. E. & Baziana, P. A., 2007. A Collision‐free with Propagation Latency WDMA Protocol Analysis. Optical FIber Technology, 13(2), pp. 160‐169.

Rathnayaka, A. J., Potdar, V. M. & Kuruppu, S. J., 2011. Evaluation of Wireless Home Automation Technologies. Daejeon, Korea, IEEE DEST, pp. 76‐81.

Rathnayaka, A. J., Potdar, V. M. & Kuruppu, S. J., 2012. Evaluation of Wireless Home Automation Technologies for Smart Mining Camps in Remote Western Australia. Smart Innovation, Systems and Technologies, Volume 12, pp. 109‐118.

Robles, R. J. & Kim, T., 2010. Applications, Systems and Methods in Smart Home Technology: A Review. International Journal of Advanced Science and Technology, Volume 15, pp. 37‐48.

Rogers, E. M., 1983. Diffusions of Innovations. New York: Free Press.

Roosegaarde, D., 2015. Interview with Dutch artist and innovator Daan Roosegaarde [Interview in TV show, presented by Ivo Niehe] (16 June 2015).

Rowling, J. K., 2000. Harry Potter and the Goblet of Fire. London: Bloomsbury (UK).

Saidi, E., 2011. Technology‐Push or Market‐Pull? A model for Managing the Innovation Process in Malawian Firms. International Journal of Innovation in the Digital Economy, 2(4), pp. 33‐44.

Sanaee, A., 2014. Smart Home Technology and Architecture. ICP Engineering and Technology, 1(1), pp. 18‐22.

Sanders, E. B., 2002. From User‐Centered to Participatory Design Approaches. In: J. Frascara, ed. Design and the Social Sciences: Making Connections. London and New York: Taylor & Francis Inc, pp. 1‐8.

Schumacher, A. J. & Feurstein, B. K., 2007. Living Labs ‐ a new multi‐stakeholder approach to user integration. In: R. J. Gonçalves, J. P. Müller, K. Mertins & M. Zelm, eds. Enterprise Interoperability II: New Challenges and Approaches. Madeira: Springer‐Verlag, pp. 281‐ 285.

Schumacher, J., 2011. Alcotra Innovation project: Living Labs ‐ Definition, Harmonization Cube Indicators & Good Practices, Genoa: Alcotra.

Schuts, M. et al., 2009. Modeling Clock Synchronization in the Chess gMAC WSN protocol, Nijmegen: Radboud Universiteit Nijmegen.

SCME, 2009. Introduction to Transducers, Sensors and Actuators, Albuquerque, NM: Southwest Center for Microsystems Education (SCME).

Sharma, D., Verma, S. & Sharma, K., 2013. Network Topologies in Wireless Sensor Networks: A Review. The International Journal of Electronics & Communication Technology, 4(3), pp. 93‐97.

Sharma, S., Kumar, D. & Kishore, K., 2013. Wireless Sensor Networks‐ A Review on Topologies and Node Architecture. International Journal of Computer Sciences and Engineering, 1(2), pp. 19‐25.

99 Shelby, Z. & Bormann, C., 2009. 6LoWPAN: The wireless embedded Internet. Chichester, UK: John Wiley & Sons Ltd.

Simoneau, P., 2006. The OSI Model: Understanding the Seven Layers of Computer Networks, Cary, NC: Global Knowledge.

Sivasankari, A., Sudarvizhi, S. & Sarala, L., 2014. A comparative study of wireless technologies based on home automation bluetooth low energy, ZigBeem, INSTEON and EnOcean. International Journal of Computer Science and Information Technology Research, 2(3), pp. 255‐259.

Smart Homes, 2015. Smart Homes: De Slimste Woning. [Online] Available at: http://www.smart‐homes.nl/Ervaren/De‐Slimste‐Woning.aspx [Accessed 8 July 2015].

Sponselee, A., 2013. PhD: Acceptance and Effectiveness of Telecare Services from the End‐ User Perspective, Eindhoven: Eindhoven University of Technology.

Sponselee, A. A., Schouten, B. A. & Bouwhuis, D. G., 2008. Effective Use of Smart Home Technology to Increase Well‐being, Eindhoven: Fontys University of Applied Sciences.

Sponselee, A., Schouten, B., Bouwhuis, D. & Willems, C., 2008. Smart Home Technology for the Elderly: Perceptions of Multidisciplinary Stakeholders. In: M. Mühlhäuser, A. Ferscha & E. Aitenbichler, eds. Constructing Ambient Intelligence ‐ AmI 2007 Workshops. Berlin ‐ Heidelberg: Springer‐Verlag, pp. 314‐326.

StatisticSolutions, 2015. The Multiple Linear Regression Analysis in SPSS. [Online] Available at: http://www.statisticssolutions.com/the‐multiple‐linear‐regression‐analysis‐ in‐spss/ [Accessed 17 July 2015].

Stefanescu, D. M., 2011. Handbook of Force Transducers: Principles and Components. Berlin ‐ Heidelberg: Springer‐Verlag.

Stichting de Beauvoir, 2012. Stichting de Beauvoir: Bevordert de participatie van vrouwen in de wetenschap. [Online] Available at: http://www.stichtingdebeauvoir.nl/monitor/man‐vrouw‐verdeling‐aan‐de‐ nederlandse‐universiteiten/ [Accessed 15 July 2015].

Surendran, P., 2012. Technology Acceptance Model: A Survey of Literature. International Journal of Business and Social Research (IJBSR), 2(4), pp. 175‐178.

Titulaer, C., 1985. Een voorspelling van de toekomst, Goirle: s.n.

Townsend, C. & Arms, S., 2005. Wireless Sensor Networks: Principles and Applications. In: J. S. Wilson, ed. Sensor Technology Handbook. Oxford, UK: Elsevier, pp. 575‐591.

TU/e, 2006. Graag meer vrouwen op de TU/e: Nederland heeft erg weinig vrouwelijke techniekhoogleraren. Cursor, 21 September.49(3).

100 van Berlo, A., 2015. Of is het toch aan de woningbouwers. [Online] Available at: https://www.linkedin.com/grp/post/2750916‐ 6022074803668729860?trk=groups‐post‐b‐all‐cmnts [Accessed 3 August 2015]. van der Horn, G. & Huijsing, J., 1998. Integrated Smart Sensors: Design and Calibration. Dordrecht: Kluwer Academic Publisher. van Dijken, B., 2010. Design of a wireless sensor nodes application for elderly people supporting their indepence, Delft: Delft University of Technology. van Gijsel, L., Peters, B. & Korsmit, Z., 2012. Industrial Demo Abstract: MyriaNed: A biology inspired self‐organizing, gossiping Wireless Sensor Network. Povo, Italy, EWSN, pp. 93‐95. van Kraaij, B., 2015. MyriaNed at Van Mierlo Ingenieursbureau, Intelligente elektronica voor meet‐ en regeltechniek [Interview] (22 January 2015). van Mierlo, M. & van Lieshout, M., 2015. Interview at van Mierlo Ingenieursbureau, Intelligente elektronica voor meet‐ en regeltechniek [Interview] (8 April 2015). van Snick, A. & Veys, C., 2014. The market consultations ‐ Summary, Brussels: Smart@Fire. van Straten, R., 2012. ''Draadloos door de jaren heen'', Tilburg: Betronic Solutions. van Vliet, G., 1973. Hoofdstuk VII: Experimentele en controle‐groep of experimentele en controle‐conditie?. In: Het psychologisch experiment. Nijmegen: Berkhout Testmateriaal B.V., pp. 39‐44.

Varmah, K. R., 2010. Control Systems. New Delhi: Tata McGraw Hill Education Private Limited.

Vastenburg, M. H., Keyson, D. V. & de Ridder, H., 2007. Measuring User Experiences of Prototypical Autonous Products in a Simulated Home Environment. Lecture Notes in Computer Science, Volume 4551, pp. 998‐1007.

Venkatesh, V., 2015. Technology Acceptance Models. [Online] Available at: http://www.vvenkatesh.com/it/organizations/Theoretical_Models.asp#Con=structdefs

Venkatesh, V. & Bala, H., 2008. Technology Acceptance Model 3 and a Research Agenda on Interventions. Decision Sciences, 39(2), pp. 273‐315.

Venkatesh, V. & Davis, F. D., 2000. A theoretical extension of the technology acceptance model: Fout longitudinal field studies. Management Science, 46(2), pp. 186‐204.

Venkatesh, V. & Morris, M., 2000. "Why Don't Men Ever Stop to Ask for Directions? Gender, Social Influence, and Their Role in Technology Acceptance and Usage Behavior.". MIS Quarterly, 24(1), pp. 115‐139.

Venkatesh, V., Morris, M. G., Davis, G. B. & Davis, F. D., 2003. User acceptance of information technology: Toward a unified view. MIS Quaterly, 27(3), pp. 425‐478.

101 Wang, P. & Liu, Q., 2011. Biomedical Sensors and Measurement: Advanced topics in science and technology in China. Berlin: Springer.

Whitehouse, K. et al., 2012. Towards Occupancy‐Driven Heating and Cooling. Design & Test of Computers, 29(4), pp. 17‐25.

WorldViz, 2015. Downloads. [Online] Available at: http://www.worldviz.com/support/download [Accessed 20 July 2015].

WorldViz, 2015. Projection Walls. [Online] Available at: http://www.worldviz.com/systems/projection‐wall [Accessed 21 July 2015].

Yong, T., Hua, M. & Mei, G., 2007. Empirical study on impact of logistics operations capability on supply chain performance.. Shanghai, WiCOM.

Zhang, J., Shan, Y. & Huang, K., 2015. ISEE Smart Home (ISH): Smart video analysis for home security. Neurocomputing, Volume 149, pp. 752‐766.

Zijderveld, J., 2012. Slim draadloos network: MyriaNed voor vele toepassingen. Netwerk technologie, Volume 11, p. 32.

Zyda, M., 2005. From visual simulation to virtual reality to games. Computer, 38(9), pp. 25‐32.

102