Providing generic context for mobile games on phones

Providing Generic Context for Mobile Games on Phones (Keynote Talk) Paul Holleis1, Alireza Sahami2, and Albrecht Schmidt2 1 DOCOMO Euro-Labs Landsberger Str. 312 80687 Munich, Germany 2 University of Stuttgart Pfaffenwaldring 5a 70569 Stuttgart, Germany [email protected], {alireza.sahami,albrecht.schmidt}@vis.uni-stuttgart.de Abstract. Mobile phone games are played in context. Although such information has been used in several prototypes, very few context-aware games have made it beyond the research lab. In our research, we investigate how the development of context-aware games needs to be changed such that their commercialization is more feasible and they can be deployed more easily. Based on the findings of the creation and evaluation of a context-based game called ContextSnake, we developed a platform named Gatherer which frees the developer from the burden of collecting, preprocessing, storing, and interpreting raw sensor data. We introduce the novel concept of generic context which enables the use of context in mobile applications without having detailed information about the actual environment in which the system will be deployed. In order to preliminarily validate the platform, a second game called ContextInvaders developed on top of this platform is described. Keywords: Generic context, context-awareness, mobile games, mobile phone. 1 Introduction Mobile phones have become ubiquitous and a widely adapted powerful computing platform with multimedia and network capabilities. They offer various options for creating sophisticated user interfaces and implementing different types of applications such as games. Built-in cameras, microphones, and color displays as well as different types of sensors can provide information about the status of the devices and the context in which it is used. One popular example is using accelerometer sensor data on the phone to control motions in games. In order to make games on mobile phones more valuable, they have to be usable in the users’ everyday situations. The following points distinguish some important differences to traditional games (i.e. board games or digital games on a PC): ─ As mentioned in [13], “killing time is a killer application”. Therefore, the time for playing is often short; a typical scenario is waiting for public transports where people typically spend between 1 and 20 minutes [16]. M. Beigl et al. (Eds.): CONTEXT 2011, LNAI 6967, pp. 5–17, 2011. © Springer-Verlag Berlin Heidelberg 2011 6 P. Holleis, A. Sahami, and A. Schmidt ─ Mobile games are played at different locations in the real world. It is common to start a game in one place and resume playing somewhere else. ─ Mobile phone games are played as ’filler‘, and interruptions occur frequently, e.g., when an incoming call is received. Several projects have looked at mobile gaming, but many of these approaches assume physically very active users (e.g., running around a city and chasing other players [7]) or users that are disconnected from real world activities (e.g., immersed in the game only). The initial experience shows that these are not mainstream mobile gaming scenarios as most people are physically not very active when playing mobile games [16]. In our research, we focus on using contextual information on mobile phones to develop games that work everywhere and motivate people to exploit their context. We evaluate the feasibility of such applications by developing two mobile context-based games and introduce a platform for mobile phones that manages and provides contextual information to simplify developing such context-based applications. The paper makes the following main contributions: ─ Introduction of the concept of generic versus absolute context ─ Overview of the design space and development requirements for generic contextual games ─ Introduction of context-action-tables as a way of describing the semantics of context-aware games ─ A platform for developing contextual mobile games The structure of this paper is as follows: after providing an overview of related work, we describe how context information can be used in mobile games. We provide definitions for the distinction between absolute and generic context. In order to show the feasibility and to examine the requirements for the development of generic context-aware applications, we introduce a game called ContextSnake. Based on the findings during its development and the subsequent evaluation, we describe the design and implementation of a context-aware platform called Gatherer, which organizes context information available on mobile phones and presents it to developers in an easy-to-use way. Finally, we show the applicability of the development platform by describing a second game called ContextInvaders built on top of this platform. 2 Related Work In the process of growing competition and search for novel features for mobile applications, developers have explored the benefits of being able to sense contextual information such as location information in designing applications. The “Backseat Gaming” prototype [5] is a mobile game in which the player has to find virtual objects related to the real world by pointing with the mobile device towards these objects. In [4] “Tycoon”, a market-based distributed resource allocation system is introduced where GSM cell information is used to define the positions of the resources in a trading game. “Feeding Yoshi” is another game in which Yoshis (small creatures) must be fed with fruits growing in different locations, defined by the availability of wireless networks in urban environments [2]. Providing Generic Context for Mobile Games on Phones 7 The variety of information described in early research in context-awareness [19] has not been used to a great extent. Most games and applications making use of contexts other than location have not made their way into products on a large scale. In “Fizzees”, motions are used to allow young children to care for a virtual pet [14]. Since the Tamagotchi1 boom in 1996 proved that people are willing to nurture virtual pets, further games and devices have been developed in this area, e.g., DigiPets2. We expect that by introducing the notion of generic contexts and by providing a platform which includes self-learned generic contexts, we can make the use of context types beyond location easier and more approachable for developers. Besides, the mobility of people makes way for other types of applications that interweave with everyday life [2]. This concept is brought further by the idea of persuasive technology. This includes applications trying to influence users’ behavior in the real world. The two games we introduce in this paper based on generic context follow this line of thought. We refer to the book by Bogost [3] for a thorough treatment of persuasive games. Less game-like examples are the “UbiGreen Transportation Display” [8], a mobile application prototype that semi-automatically senses and reveals information about the transportation behavior to encourage green transportation habits by using mobile phones and the remote control that helps people spend less time watching TV, as well as the exertion interfaces by Mueller et al. [12]. Researchers also have explored some ways to facilitate the development of context-aware applications on mobile devices. In order to reduce the efforts of common tasks and to help developers focus on the application-specific components, various frameworks with different aims for individual mobile platforms (such as Symbian, Windows Mobile, etc.) have been introduced. Recent examples can be found in [10,11,15] which encapsulate certain functionalities and provide services for other applications. However, none of them focuses on capturing and providing generic context to the developers. 3 Absolute vs. Generic Context Although context-aware applications have been the focus of many research projects, up to now, only very few have matured into commercial services. In location-aware and mobile gaming, one common approach is to create specific, absolute contexts, such as absolute positions, that are meaningful within the application. Such games are fixed to a predefined context and are only economically viable if the number of potential users is high (e.g., in a large museum). Creating context-aware games or context-aware applications that work independent of a location, specific time, etc. is a challenge that needs to be addressed. 3.1 Absolute Context Definition: An absolute context is a context that is tied to a specific situation such as a location described in absolute terms. This implies that an absolute context is independent of the history of the user or device. 1 2 http://en.wikipedia.org/w/index.php?title=Tamagotchi (accessed May 2011) http://www.digipet.com/ (accessed May 2011) 8 P. Holleis, A. Sahami, and A. Schmidt Examples for absolute contexts are a location defined by absolute coordinates or the proximity of a device characterized by a specific id. This has traditionally been used in context-aware applications: most existing tourist and museum guides running on mobile devices offer support for a specific setting and location. Either the location of an interesting site has been fixed or the mobile application reacts on the proximity of a specific device attached to an artifact. These applications are difficult to migrate to different scenarios, as it requires at least the redefinition of the absolute context parameters. To overcome this issue, we define generic contexts that do not depend on absolute specifications. 3.2 Generic Context Definition: A generic context is a context that can be described without referencing to a specific or absolute property. Hence, a generic context can specifically be related to the history of the user or device. In the example of a museum guide, there are several levels between absolute and generic context. Let’s say a mobile application reacts when the user is close to a specific location (i.e. position in a specific room) and looks at a specific artifact. When removing the restriction of location, the employed context applies to all copies of that particular artifact. This can be even more generalized by abstracting this to only one type of artifact defined by properties such as date, value, size, usage, etc. Then one instance of a most generic context is the one that does not restrict itself to any specific components at all but relies on the history of the user contexts. Examples for such generic contexts are a new place the device has never been before or the proximity of a set of devices that already have been encountered before (see Fig. 1). Fig. 1. Examples of absolute vs. generic context Generic contexts have the advantage that they can be learned (unsupervised) during usage. A generic context-aware application, once developed, can work in any environment without manual tweaking. Hence, from a production point of view, they can be produced to suit a global mass market without additional efforts in creating specific context descriptions. Providing Generic Context for Mobile Games on Phones 9 4 ContextSnake To investigate the usage of context information in mobile games, we developed a context-aware game called ContextSnake. This development led to the understanding and refinement of requirements for the context toolkit we describe afterwards. In this initial phase we deliberately use a very simple game to explore the impact context can have. We believe that if we can show that even simple games benefit from generic context this is a strong indicator that this may also be true for more complex games. 4.1 Concept and Design We decided to use the straightforward and well-known game Snake in which a user controls a snake moving through a scene and turn it into a context-aware application. This had the advantages that it simplifies the task of introducing users to the application and offers great potential for including contextual parameters. We focused on using simple learning algorithms and statistics to acquire further information about the context (e.g. finding out the favorite places a user plays in with statistics over the acquired cell IDs) and storing contextual parameters persistently. This last requirement is directly connected to fact that people often play mobile games in short breaks and persistently storing and learning of contextual and conventional game parameters enables a longer game experience. For the use of contextual parameters we focused on acquiring the information about battery power (bat), network signal strength (rss), GSM cell ID (cID), and system time (t). In comparison to other parameters (e.g., environmental noise, light condition, Bluetooth devices in vicinity) these parameters are very easy and fast to acquire. Additionally, they are very robust and rely little on the way the user is handling the device. Furthermore, for all sensor values a history is kept and we use functions to abstract the values into higher-level contexts (e.g., frequency of cell changes, estimated activity of the user). 4.2 Context-Action-Table To design such games, we suggest using context-action tables. A context-action table describes the mapping between the sensed parameters and abstracted contexts to the parameter influenced in the game (appearance, logic, functionality). In Table 1, the context-action table for the prototype is shown. In each row a context, expressed by a function on raw context data over time, is described. The values used as input parameter to the function are typically sensor data, values calculated from sensor data, and a history of data. Each column reflects behavior in the game. Marks in the intersections between rows and columns denote a functional dependency. The more dots in a row, the more often this context is used to influence game parameters; the more dots in a column, the more dependent on contexts is this game parameter. Functions then describe the actual mapping, i.e., the semantics of the connection between context and game content. The benefit of using such a form of description is that the relationships can be easily observed and further options for potential new mappings become apparent. In our prototype, these mappings are selected at the design step and reflect the design 10 P. Holleis, A. Sahami, and A. Schmidt choices of the game’s developers. However, it is feasible to have these mappings modeled as functions, too. This makes changes during runtime possible, adding another dimension of options for dynamic and customizable game play. Table 1. Context-Action-Table for the ContextSnake game f0 (rsst) f (rsst,..., rsst-n) f (bat) f (cIDt, ..., cIDt-n) color of the snake size of the field size of the snake the speed of the snake the size of food items growth of the snake per food points earned for a food item time before new food appears   color of the background color of the food f (cID)            Without going into too much detail, some of the mappings between the contextual parameters and the game shown in Table 1 are as follows: ─ ─ ─ ─ ─ Low signal strength increases difficulty (larger snake and smaller field size). Battery-level roughly proportional to snake speed. Quicker changes in signal strength / cell IDs increases size of a food item Visiting an unknown cell ID provides more and more valuable food items Cell ID defines the background color (other colors are calculated to ensure contrast). Overall, the mappings for the prototype have been chosen to be a first attempt to create a compelling and more diverse gaming experience. The rationale of the mappings is mainly due to the way the designers want users to play the game. Even with a restricted set of contextual parameters a large set of relations to the real world can be included. 4.3 Game Implementation ContextSnake is realized as a Python script using the Nokia Series 60 Python SDK. The development was iterative and tests have been conducted throughout the development using the Series-60 Emulator running on the PC (included in the C++ SDK) as well as in the runtime environment on different phones. The implementation of the ContextSnake game is based on the example provided with the SDK (snake.py). A class allows accessing to required system parameters such as GSM information extended the example. Additional functionality was added to store and access recorded information (e.g., contextual parameters) and game status information (e.g., score) persistently in the phone using a database. Changes in the status are also directly written to the external database to ensure that all changes are recorded even if the program is terminated or interrupted. The current implementation was tested and Providing Generic Context for Mobile Games on Phones (a) Playing at home… (b) … while waiting for a train… (c) … while on the bus… 11 (d) … at the favorite café. run with good performance on the Nokia 6630 and Nokia N90 and can be run easily on newer models supporting Python development. 4.4 Evaluation To evaluate the game concept and the prototype, we conducted a study including two parts with ten people, four females and six males, aged between 13 and 32 years. In the first part, we used general questions to find out about mobile phone usage and frequency of playing games on mobile phones. During the second part, participants played the game and explored its features. For this, we observed the participants during play and discussed their experiences. With the questions in the interview the participants could express their agreements on a 5-point Likert-scale. Based on the results, all participants were experienced in using mobile phones. Nine had experience playing games on the phone (one user once per week and eight users about once a month). Situations in which they played games were mainly on public transport or while waiting; only two persons stated that they played mobile phone games at home. Furthermore, all users understood the concept of ContextSnake and in general the idea of contextual gaming. All participants agreed (three users) or strongly agreed (seven users) that the game becomes more interesting when it is context dependent. Nine out of ten agreed or strongly agreed that they would play the game at more locations and different places to get a higher score. Besides, all participants were confident that the game becomes considerably less predictable when contextual parameters are used and can thus remain more interesting in the long term. Fig. 2 shows some examples of the game screen and its use in various everyday situations: playing longer at the same location becomes less rewarding as food gets scarce, see setting (a). Places with little or no signal strength (e.g., at an underground station) render the food larger and easier to get, see setting (b). This can turn the waiting time for a train into a more positive experience. Travelling provides many 12 P. Holleis, A. Sahami, and A. Schmidt location changes and hence a lot of food. The game also motivates users to sometimes explore different routes as otherwise less food will be available, see setting (c). Setting (d) shows a specific café where food never becomes scarce; such a game could be a giveaway of a company that has interest that you come back onto their premises. One important finding is that contextual games are a reason to encourage people exploring their real world environment more (eight participants agreed or strongly agreed). All participants agreed that ContextSnake is more interesting than the conventional snake game. Overall, we received as feedback that using contextual parameters is an intriguing way to make games more interesting. Comments from five participants included that detecting encounters (e.g., Bluetooth, WLAN) could be another exciting contextual parameter. 5 Gaming in Context Generic real-world locations can be linked to a mobile game and provide the stage that reaches from the real world into the virtual environment, creating the game canvas. Consider, for instance, a virtual pet game: in order to eat, the pet has to be in a supermarket, for exercise it has to be in a forest, etc. This pervasive feature adds physicality, as players have to move to certain locations in order for the pet to feel well. Generic context opens several additional opportunities. Continuing the location example, the game can force the players to seek out new locations or, reversely, the game only makes sense in a location where they spend most of their time. These design parameters allow the game designer to induce certain behavior – in the real world – to the player. This has obviously severe implications on the player’s behavior in the real world and hence includes ethical issues, e.g., what is more desirable, to create a game that makes people stay at home or that makes people go and play in places where they meet others? Further to location, investigations in context and context-awareness in previous projects such as [18] show that there is a great set of information that can be taken into account. Examples of real world contexts and a sample way of estimating whether this context applies at the current situation we found to be important include: ─ being in a place that a user has never visited before (unknown GSM base station) ─ being in a place that is minimally populated (seeing only one GSM base station) ─ being in a very quiet environment (based on the noise level from the microphone) ─ being outside in daylight (based on images captured from the phone camera) ─ being in a crowded place (several other Bluetooth devices visible) Although the use of context is gaining importance with an increasing number of pervasive computing applications, it is still a challenge to enable applications to make use of context information. This is especially true for mobile applications, as they are not bound to specific context but rather generic ones. Learning from implementing ContextSnake, we see the strong requirement for a platform on a system level that Providing Generic Context for Mobile Games on Phones 13 provides not only real-world context information but also aggregates, abstracts, and interprets raw data and translates the abstract context data to a more generic one. Specific requirements are: ─ ─ ─ ─ ─ A history on which generic context and many algorithms (e.g. profiling) rely on Supporting generic context and not only absolute context information Extensibility: a modular platform should easily accept new types of sensors etc. Interoperability: an open platform should accept different data formats Independent of any sensor network or infrastructure for enabling contextawareness By using such a platform, developers should be enabled to create games that are dynamic, context-aware, learn about their environment and users’ mobility, and behave differently in different situations so that they are consistent in their reaction to the environment without being fully predictable. 6 Framework: The Gatherer To fill the gaps and fulfill the requirements described in the previous section and to enable mobile applications to access different types of contextual information, a platform called Gatherer was implemented running on top of existing infrastructure in a mobile phone. The Gatherer enables developers to access context information and frees them from developing the whole process chain of collecting raw sensor data, preprocessing, storing, and interpreting it. Developers can treat the Gatherer as a black box, which provides various context data and they need only focus on designing their applications. It not only assists in the creation of context-aware applications but also supports rapid prototyping as it provides various data formats, hence, developers do not need to take care of accessing and processing information. The Gatherer platform is a Python-based application running as a background process (server) on mobile phones and handles the whole process of sensing, preprocessing, storing, interpreting, and formatting context data. This system can be installed and used on mobile phones that include the Python interpreter such as Nokia Symbian phones. In the design stage, it was decided to separate the sub-processes of the platform due to easy maintenance and expansion. Thus, a modular microkernellike design was chosen for the software architecture. This speeds up the development process and new features can be easily added to the system. The system is divided into four layers: ─ Device Abstraction Layer: the actual sensing mechanism, pulling data from available sensors on the phone, preprocessing (aggregating / smoothing) and storing them. Supported sensors include: physical (e.g. accelerometer); infrastructural (GSM cell ids, battery level, nearby Bluetooth enabled devices); informational (incoming / outgoing text messages / calls). ─ Persistency Layer (database): produces the context history by storing sensed context data. In accordance with Dey et al. [6], we use a name-value tuple as the main data type for the context-aware applications. ─ Context Conversion Layer: includes all data access methods and processes all requests from client applications and produces a collection of tuples. 14 P. Holleis, A. Sahami, and A. Schmidt ─ Formatting Layer: formats the data retrieved from the Context Conversion Layer based for specific requests. Currently provided are XML, JSON, and simple CSV. On the other side, context-based applications (clients) which want to use the Gatherer platform simply establish a connection to a specific port on the localhost of the mobile phone and send requests. In the requests, the required context information and data format are specified. The clients can then extract the information from the responses. 7 ContextInvaders In order to explore how the usage of context information in combination with traditional games can change the compelling and fun part of a game experience, another game called ContextInvaders was developed. In addition, it served as an example to evaluate the Gatherer platform. ContextInvaders is based on the Space Invaders game3 – one of the first arcade shooting games released in 1978. In contrast to the original game, it uses context information to modify the game’s parameters. The mission of the player is to save the earth from an alien invasion by shooting their spaceships before they reach the earth using a laser cannon. As the aliens also shoot the player, one can take cover behind defense bunkers placed between the player and the aliens. Table 2 shows the context-action table for the ContextInvaders game. As the Gatherer platform handles context acquisition, representation, interpretation, and provides more generic data, the table can now be written with more generic and understandable descriptions. The table explains what kind of information provided by the Gatherer is used within the game logic: ─ Average movement (by accelerometer) modifies the player’s speed ─ Current GSM cell ID and duration spent in a location influences game parameters (e.g. number of lives, number of spaceships, …) ─ Current time modifies background and spaceship’s color (simulating day / night) ─ Battery level information changes the amount of the player’s shots 7.1 Preliminarily User Study The ContextInvaders game was evaluated with ten participants, five males and females (age range 19-24), who all had a gaming background on mobile phones. The game and the Gatherer framework were installed on a Nokia N95. The users played both the classical Space Invaders and ContextInvaders for three days in a row and provided feedback through a questionnaire. Although all users understood the concept and meaning of context-aware games, not all were able to find out how the sensed context information changed the game. The main reason might have been the lack of context parameters visualization in the game as described in the previous prototype. Interestingly, just three of 3 http://en.wikipedia.org/wiki/Space_Invaders (accessed May 2011) Providing Generic Context for Mobile Games on Phones 15 the participants stated that this encouraged them to play the game at different locations in order to achieve more scores. This can be partially attributed to the fact that mobile games are mostly played when there is some time to kill between events of interest [13]. The question whether the context enhanced mode was more challenging to play than the normal mode divided the group into half. This is probably the result of the usage of context parameters, which makes the game harder or easier based on the phone’s status. Another important factor can be the different gaming experiences the participants had. Some of the features are only available and more feasible when using the game for a longer period of time and thus experiencing more of the defined generic situations. Table 2. Context-action table for the ContextInvaders game. LP (directly proportional) means an increase in context parameter value leads to an increase in the game’s parameters; IP (inversely proportional) means an increase in context value leads to a decrease in the game’s parameters. Battery level Location familiarity Enter new location number of aliens spaceships IP LP number of player’s lives LP IP player movement speed  color of the background number of player’s shots Time of day color of spaceships Crowdedness LP number of bunkers number of points User activity LP IP  IP LP 8 Discussion and Conclusion Playing in context is a reality as users already play games with mobile devices in a variety of everyday situations. So far, most games are still ignorant of the actual physical environment. We have assessed how mobile gaming can benefit from the mobile usage model and investigated what opportunities arise in the domain of contextual gaming. Introducing the concept of generic context, we showed that interesting contextual elements can be added without the effort of specifically designing applications and games for a specific environment. To validate the concepts, we implemented a context-aware version of the popular Snake game available on most mobile phones. Beyond using the current context, we saw that using statistics over context histories as parameters for the game is also promising as a parameter. For visualizing the mapping between context information and the parameters in a contextual game, we introduced the context-action table. Our evaluation of the game based on interviews and the users playing the game provides evidence that making games context-aware will make them more interesting – and this applies even to very simple games. The game can receive a more realistic touch and become much more intriguing over time when physically moving around has a direct impact on the game. 16 P. Holleis, A. Sahami, and A. Schmidt From implementing and evaluating the relatively simple game, we learned and discussed requirements for creating a platform on the system level for the development of such applications. Especially when using parameters that rely on context-histories (e.g., how familiar is this place / has the user ever been here before?), a common contextual gaming platform is beneficial for game developers. The Gatherer platform was introduced to give developers access to context information by taking care of the whole process chain of collecting raw sensor data, preprocessing, storing, interpreting, and providing more generic data. Based on its flexible core and system structure, this platform can be easily extended and run on various platforms. A second game, ContextInvaders was then developed on top of the Gatherer platform. It is a context-aware variant of the well-known Space Invaders game. The game’s logic uses different types of context information provided by the Gatherer to modify the parameters. One main advantage of using generic contexts is that the developers do not need to know the actual location and context of deployment; hence, applications built on this concept are easier to deploy and maintain. Using generic contextual parameters, game designers can create games that trigger people to interact with their environment and with others in novel and playful ways, enabling easier designs for Homo Ludens [9]. Games can be created to push the players towards a certain behavior, e.g. encourage people to explore their environments or to go to certain places. 9 Future Work In this paper we focused on using generic context in games and showing the potential in this domain. Our experience so far suggests that the notion of generic context may also be useful for context-aware applications in general. To our knowledge, many context-aware applications do not make their way from the laboratory to the real world, as defining the absolute contexts required on a national or global scale is not practical. In future work we plan to assess how the concept of generic contexts can help to provide more context-aware locations in realistic environments. A further dimension of future work evaluates how interaction with games and applications changes when the application logic is linked to generic contexts. We assume that when people learn that some of their actions have impact on the game, it may lead to a change of their behavior to maximize the benefit, without even reflecting on their behavior change, e.g., opting for the bus instead of the subway as the game is more exciting on the bus route. We plan to investigate if this can be used to take the concept of implicit interaction [17] to a new level where people are subconsciously influenced. One promising domain is the creation of persuasive games. By providing incentives in the game that are linked to actions in the physical world we image many possibilities. Such an approach may have interesting implications in designing healthcare and wellness application but can be frightening at the same time if we think of possibilities in advertising. Providing Generic Context for Mobile Games on Phones 17 References 1. Bell, D., Hooker, B., Raby, F.: FLIRT: Social Services for the Urban Context. In: Proc. HCI International 2003 (2003) 2. Bell, M., Chalmers, M., Barkhuus, L., Hall, M., Sherwood, S., Tennent, P., Brown, B., Rowland, D., Benford, S.: Interweaving Mobile Games with Everyday Life. In: Proc. CHI 2006, pp. 417–426 (2006) 3. Bogost, I.: Persuasive Games: The Expressive Power of Videogames. The MIT Press, Cambridge (2007) 4. Broll, G., Benford, S.: Seamful Design for Location-Based Mobile Games. In: Kishino, F., Kitamura, Y., Kato, H., Nagata, N. (eds.) ICEC 2005. LNCS, vol. 3711, pp. 155–166. Springer, Heidelberg (2005) 5. Brunnberg, L., Juhlin, O.: Movement and Spatiality in a Gaming Situation - Boosting Mobile Computer Games with the Highway Experience. In: Proc. INTERACT 2003 (2003) 6. Dey, A.K., Abowd, G.D., Salber, D.: A Conceptual Framework and a Toolkit for Supporting the Rapid Prototyping of Context-aware Applications. Human Computer Interaction Journal 16(2), 97–166 (2001) 7. Flintham, M., Benford, S., Anastasi, R., Hemmings, T., Crabtree, A., Greenhalgh, C., Tandavanitj, N., Adams, M., Row, F.J.: Where On-line Meets on the Streets: Experiences with Mobile Mixed Reality Games. In: Proc. CHI 2003, pp. 569–576 (2003) 8. Froehlich, J., Dillahunt, T., Klasnja, P., Mankoff, J., Consolvo, S., Harrison, B., Landay, J.A.: UbiGreen: Investigating a Mobile Tool for Tracking and Supporting Green Transportation Habits. In: Proc. CHI 2009, pp. 1043–1052 (2009) 9. Gaver, W.: Designing for Homo Ludens. I3 Magazine 12 (June 2002) 10. Kukkonen, J., Lagerspetz, E., Nurmi, P., Andersson, M.: BeTelGeuse: A Platform for Gathering and Processing Situational Data. IEEE Pervasive Computing 8(2), 49–56 (2009) 11. Linner, D., Kirsch, F., Radusch, I., Steglich, S.A.: Context-aware Multimedia Provisioning for Pervasive Games. In: Proc. ISM 2005, pp. 60–68 (2005) 12. Mueller, F., Agamanolis, S., Vetere, F., Gibbs, M.R.: A Framework for Exertion Interactions over a Distance. In: Proc. SIGGRAPH 2009, pp. 143–150 (2009) 13. Nielsen, J.A.: WAP Field Study Findings (December 10, 2000), http://www.useit.com/alertbox/20001210.html 14. Page, A., Cooper, A.: Fizzees (Physical Electronic Energisers), http://archive.futurelab.org.uk/projects/fizzees 15. Raento, M., Oulasvirta, A., Petit, R., Toivonen, H.: Context-Phone: a Prototyping Platform for Context-Aware Mobile Applications. IEEE Pervasive Computing 4(2), 51–59 (2005) 16. Rukzio, E., Schmidt, A., Hussmann, H.: Physical Posters as Gateways to Context-aware Services for Mobile Devices. In: Proc. WMCSA 2004 (2004) 17. Schmidt, A.: Implicit Human Computer Interaction Through Context. Springer Journal on Personal and Ubiquitous Computing 4(2/3), 191–199 (2000) 18. Schmidt, A., Aidoo, K.A., Takaluoma, A., Tuomela, U., Van Laerhoven, K., Van de Velde, W.: Advanced Interaction in Context. In: Gellersen, H.-W. (ed.) HUC 1999. LNCS, vol. 1707, p. 89. Springer, Heidelberg (1999) 19. Schmidt, A., Beigl, M., Gellersen, H.-W.: There is More to Context than Location. Computers & Graphics 23(6), 893–901 (1999) 20. Zeimpekis, V., Giaglis, G.M., Lekakos, G.: A Taxonomy of Indoor and Outdoor Positioning Techniques for Mobile Location Services. SIGecom Exch., 19–27 (2003)