This study explores the design and development of serious games using augmented reality (AR) technologies integrated with IoT sensors. The aim is to raise student awareness of air quality in their study environments. The study employed qualitative data gathered from three iterations of game implementations with thematic analysis. Results were classified into four key dimensions: IoT integration and data management, game design and mechanics, user experience and engagement, and educational impact and outcomes. Findings indicate that real-time environmental data can reinforce IAQ awareness and prompt healthier decision-making within a game context, although sustaining user engagement and overcoming technical barriers remain challenges. Moreover, incorporating refined game design elements such as improved feedback loops may enhance both educational impact and motivational appeal.
Indoor air quality (IAQ) in educational institutions has garnered increasing attention due to its
direct influence on student performance and well-being. Classrooms’ indoor environmental
conditions have been recognized to affect academic achievement among students and teachers in
higher education, highlighting the importance of adequate ventilation, temperature control, and air
filtration measures [
People in both developing and developed countries spend an estimated 80–90% of their time
indoors. Although HVAC systems aim to maintain indoor environmental quality (IEQ), they are
not always effective, and health organizations’ guidelines for IEQ are often only voluntary. Various
pollutants from building materials and occupants contribute to poor IAQ, which can lead to
numerous negative health outcomes. For instance, insufficient air circulation can facilitate the
spread of infectious diseases [
Studies on the IEQ of university classrooms revealed that poor indoor air quality, which is
characterized by high levels of CO₂, inadequate ventilation, and other environmental factors, has a
negative impact on students' concentration, productivity, and overall health. The students are at
greater risk because of the long hours they spend in these environments. As a result, improving the
IAQ in classrooms is critical for improving student wellbeing and academic performance [
The Internet of Things (IoT) has been developed and implemented over the last decade for
numerous use cases ranging from industry use and many ways it has become a cornerstone of our
digital society [
Serious games are digital or mixed-reality experiences that engage players through story and
gameplay to inform or influence [
This research explores the potential of IoT-enabled augmented reality (AR) serious games to raise student awareness of IAQ and promote healthier behaviors. By integrating real-time IoT sensor data with AR, we developed the Legend of Alumnus, an interactive game that merges virtual and physical environments to educate students on the impact of IAQ on their health and behavior. The primary objective is to design and assess the effectiveness of such games in fostering IAQ awareness and encouraging behavioral change by motivating students to navigate their campus toward locations with better air quality. This integration creates an educational experience that extends beyond the digital screen, reinforcing the importance of IAQ in everyday decisions.
• •
RQ1: How can real-time IAQ sensor data be effectively integrated within an AR game to enhance students’ awareness of indoor air quality? RQ2: Which game design features and user experience factors most significantly influence student engagement and healthy behavior reinforcement in an IoT-enabled AR game aimed at IAQ awareness?
The scientific community recognizes the potential of serious games in healthcare, particularly
when integrated with IoT-enabled monitoring devices to promote healthy behaviors like physical
exercise. Personalization within these games enhances their effectiveness, with engagement levels
reflecting the success of the healthy behaviors encouraged by the IoT-enabled serious games
(IoTeSGs) [
[
[
This study adopts an exploratory case study design as outlined by Yin [
This section presents the Legend of Alumnus game, its three iterations, their game design, why certain design aspects or principles were selected, and how these were implemented in conjunction with the IAQ sensor data. The game utilizes both AR game and IoTeSG design paradigms. The purpose of the game is to explore new ways to utilize IoT sensors so they can purposefully affect the gameplay mechanics. The project spanned from 2022 to 2024, over three iterations, with separate student groups from the Research & Development project course in Information Processing Science at the University of Oulu. Throughout the process, the researcher and product owner guided the design and development to ensure that the outcomes were meaningful for research into IoT sensor and AR usage in games.
The initial concept of the game, presented to the student groups, involved creating a game based on university mythology, incorporating location-based gameplay and AR, while leveraging the smart campus IAQ sensor network that had been implemented throughout the campus. Although the concept and certain project requirements were part of the assignment, the student groups were given the freedom to design the game and its mechanics as they saw fit.
The game development followed an iterative design process across three distinct versions. The iterations are summarized in Table 1, which provides an overview of each version’s platform, genre, game loop, and the integration of IoT sensor data and location-based mechanics. Figure 1 presents the user interfaces of each iteration. Each iteration focused on improving specific aspects of the game, such as IoT sensor data integration, location-based mechanics, and gameplay progression.
The initial prototype of Legend of Alumnus was developed as a location-based AR action roleplaying game (RPG) for Android. Players navigated the campus physically, encountering ghosts spawning near IAQ sensor locations. Due to integration challenges, historical IAQ data was used, with ghost spawn rates linked to local IAQ levels. The game featured a 3D campus map, and ghost encounters led to AR battles where players shot ghosts to reduce their health until capture. An inventory provided capture aids, and a ghost database allowed players to collect and track different ghost types. In this first iteration, testing was conducted exclusively within the development team and the project’s customer company. The primary focus was on ensuring the game’s basic functionality and testing the initial integration of historical air quality data with the game mechanics. At this stage, the emphasis was placed on identifying technical challenges, resolving gameplay issues, and improving the integration of IAQ data. No formal usability tests were conducted during this phase, as the game was still in an early prototype stage.
The second prototype took a different approach in its game design. While the first prototype concentrated on AR-style gameplay with location-based mechanics, the second iteration presented a classic RPG design and gameplay, where smart campus sensors influenced the strength of enemies appearing on the game map. There was no player location integration, and the player navigated the pixel-style university map using a keyboard and mouse on PC. In this iteration, the game loop consisted of the player navigating the game world, following a storyline, and collecting items, such as keys, to progress. When encountering ghosts, a turn-based battle scene would start, where the player had options to attack, defend, heal, or wait. Additionally, the game included a subsidiary loop in which the player earned rewards and experience points (XP) to level up.
The second iteration of the game introduced a structured evaluation, applying a heuristic
evaluation framework based on [
The third iteration took a more serious turn; while the game design of the two first iterations concentrated on creating a fun and engaging experience of playing a game revolving around the campus, in third iteration, the IAQ on campus and how it affects the health of students was in center of attention. The core design aimed to encourage students to avoid areas with poor air quality, particularly where CO2 levels exceeded a certain threshold.
The game genre remained location-based AR game, but it incorporated strong design elements from contemporary idle games. The main game loop involved the player navigating the campus and selecting study locations based on IAQ information provided by the game. Players were tasked with placing a virtual pet-style spirit companion in areas with good air quality. To nurture their spirit companion, the player had to periodically relocate the spirit to maintain its health in good IAQ zones. If left in a bad IAQ area for too long, the spirit would become sick, reducing the player’s ability to earn Intellectual Points (XP).
While the player was idle, they could occasionally feed the spirit with resources collected from specific locations on campus. In addition, players were periodically awarded with spirit evolutions, adding another layer of gameplay complexity through a secondary loop.
In this final iteration usability testing was conducted using an Android test APK, involving both
internal team members (e.g., four student developers) and external volunteers (four test users) with
diverse backgrounds in technology and gaming. Following best practices in usability research by
[
At the end of each iteration, developer experiences were systematically collected and documented through various project artifacts and reports. These included game design documents, expertise reports, project plans, steering group meeting minutes, mid- and final reports, project portfolios, seminar papers, PowerPoint presentations, Unity game project scripts, communiques between the project team and the customer, game evaluation reports, and the game builds for both Android and Windows platforms. This comprehensive set of documentation captured the development process, challenges encountered, decisions made, and the outcomes of each iteration. Alongside user feedback from the testing phases, these materials served as the primary data for analysis.
The data collected from these artifacts were analyzed using Braun and Clarke’s thematic analysis
method [
The first phase was a thorough examination of all the relevant documentation and artifacts of the three iterations of the project. Straight from the data, inductive reasoning was used to extract evidence patterns and insights.
The evidence was then categorized under key aspects of game development. To ensure that the coding was limited to evidence that contributed to unique insights, this stage required filtering out common knowledge components to concentrate on material that could yield new understandings. Abductive reasoning was used.
Then, similar bits of data were grouped into coherent categories to uncover first-order themes. Inductive reasoning was used to combine cohesive themes from related evidence. After that, second-order themes and aggregate dimensions were created by further abstracting and synthesizing first-order themes for structured and comprehensive analysis of the qualitative data. In this stage, inductive reasoning was still used.
In the final step, the underlying core causes and effects of each aggregate dimension were investigated. This step involved connecting the aggregate dimensions with reasoning and evidence from literature. The goal was to confirm the findings by connecting the aggregate dimensions to existing theoretical frameworks, utilizing abductive reasoning to ensure robustness.
This chapter presents the key results of the thematic analysis, which are organized into four primary aggregate dimensions: IoT integration and data management, game design and mechanics, user experience and engagement, and educational impact and outcomes. Figure 2 presents the key thematic dimensions identified in the analysis.
Methodological approach
Effective integration and management of IoT sensor data played a crucial role in the game's development and functionality. The reliability and accessibility of data were central to ensuring that the game could dynamically respond to real-world conditions, and several critical factors were identified in the process.
Data accessibility and reliability emerged as one of the key components for successful implementation. The use of persistent flags within the MQTT broker was instrumental in maintaining data reliability, particularly when sensor updates were infrequent. These flags allowed the system to preserve the most recent data, ensuring that even when the sensors weren’t continuously sending updates, the game could still access reliable, up-to-date information. The inclusion of these persistent flags mitigated potential data losses and allowed for smoother game performance
Another key finding in this area involved the structure of the data itself. Initially, all devices were publishing sensor data under a single topic within the MQTT broker, which created significant bottlenecks during game initialization. The lack of a structured, atomized approach meant that the system had to process a large volume of irrelevant data, causing delays in loading the game levels. This inefficiency was addressed by restructuring the MQTT broker to publish data in separate topics, allowing the game to initialize much faster. Furthermore, atomizing the data structure, combined with persistent flags, proved to be highly effective in enhancing data management.
Despite these improvements, real-time data access remained a challenging aspect of the development process. The MQTT broker lacked the capability to specifically target the most relevant sensors, which presented difficulties when the game needed to fetch real-time data for dynamic game mechanics. This issue was particularly noticeable during the third iteration, where the game struggled to fetch only the latest sensor readings. Consequently, the project team implemented a solution that relied on historical data retrieved through the smart campus REST API to expedite the initialization process. While this workaround helped resolve initialization delays, it also highlighted the limitations of the MQTT broker's real-time data handling capabilities.
The use of Unity game objects to fetch real-time data directly from the MQTT broker further optimized the gameplay experience. The Unity game objects were specifically designed to handle the real-time sensor data that influenced various aspects of the game, such as updating enemy power levels and affecting quest outcomes. By integrating live environmental data into the game mechanics, the game could mirror real-world conditions in real time, creating a more immersive and responsive experience for the players. However, the inability to consistently retrieve only the latest sensor readings meant that in some cases, game initialization was slower than desired, particularly when dealing with a large number of sensors.
The design and mechanics of the game evolved through iterations, with a consistent focus on integrating real-world data into the virtual environment, aiming to enhance player engagement, and refining gameplay elements based on sensor inputs. Each aspect of the game’s design was shaped by the need to mirror real-world conditions, allowing players to interact with both their physical and virtual surroundings.
One of the defining features of the game was its use of real-world air quality data to influence in-game mechanics. Throughout the iterations, players were required to physically move within the game’s environment to seek out areas with better air quality, which would provide in-game incentives. This was particularly evident in the third iteration, where players had to place virtual pets in locations with favorable air quality to ensure their pets thrived. The University of Oulu campus was recreated virtually, and the game’s mechanics were designed to reflect live sensor data from this real-world location.
This approach was not just about providing players with information but about integrating it seamlessly into the gameplay. The player's real-world position was mirrored within the virtual world, ensuring that decisions made in the game were influenced by real-time conditions. For example, players could not place their virtual pets in locations that were far from their actual physical position, further emphasizing the link between the virtual and physical worlds. Sensor data thus became a critical gameplay element, informing where pets could be placed and how they interacted with the environment.
Another key feature of the game was the use of dynamic game elements, driven by real-time air quality data. The game used the classification of air quality, particularly CO2 levels, to generate passive points for the player’s virtual pets. Each sensor in the real world corresponded to a specific area in the virtual game world, and the CO2 levels from these sensors determined how the environment within that area would behave. For instance, areas with poor air quality would result in fewer rewards for the player, while areas with better air quality would provide additional points and benefits. This dynamic interaction between real-world environmental data and in-game mechanics added complexity to the gameplay, as players were encouraged to make strategic decisions based on real-time data.
In addition to influencing passive point generation, environmental data also shaped the behavior and characteristics of in-game enemies. The number, size, and power levels of enemies in the game were directly linked to the environmental data captured by specific sensors. In the first and second iterations, the strength of enemies and their spawning rates were tied to air quality levels, with tougher enemies appearing in areas with poorer environmental conditions. This added a layer of challenge to the game, as players had to factor in real-world environmental factors when planning their movements and strategies.
The iterative design process allowed for continuous refinement of the game’s user interface (UI) and interaction elements. The development team adopted a multi-stage prototyping approach, starting with low-fidelity sketches and wireframes to test basic functionalities. These early designs provided a foundation for high-fidelity prototypes that were later developed in tools such as Figma, which enabled the team to fine-tune the aesthetic and functional aspects of the game.
A key focus during this iterative process was ensuring that the game’s UI could adapt flexibly to meet both aesthetic and functional demands. By working through multiple stages of prototyping, the team was able to refine the user experience, improving both the visual appeal and the usability of the game. The placeholders used in early iterations allowed for the testing of features and scaling requirements, ensuring that the final design could handle the complex interactions between real-world data and virtual game elements.
Another important aspect of the iteration process was the use of generative AI to fill gaps in art and asset creation. The team faced challenges in creating game assets that fit the selected themes due to limited resources and skills in art and design. To overcome this, they employed generative AI, which allowed them to produce better quality assets. While the lack of existing protocols or guidelines for using generative AI in game development was an initial barrier, the approach proved to be successful for the project outcomes in the second iteration where generative AI was used.
User experience and engagement in the game were evaluated through testing and feedback in the third iteration. Key insights revealed issues with player motivation, navigation, interaction, immersion, and accessibility. A primary challenge was the low engagement with the game’s environmental goals. The virtual pet system, involving pet care and point rewards, lacked strong incentives to keep players interested. The pets’ health and points linked to air quality were not compelling enough, leading to reduced player motivation to engage with environmental aspects like air quality monitoring or real-world actions based on in-game data.
The usability testing also highlighted several issues with navigation and interaction within the game. Although the app's basic navigation was generally perceived as simple, users expressed a need for clearer and more intuitive navigational aids. For example, participants found it difficult to distinguish player pins and virtual pet locations on the map, which affected their ability to interact smoothly with the game world. This confusion indicated that the UI needed further refinement, particularly in terms of its icons and map elements.
Furthermore, some users struggled to complete tasks related to collecting points or interacting with specific game features, pointing to gaps in the clarity of the game’s instructions. The need for more explicit, in-game guidance was evident, as players often found themselves uncertain about the steps required to progress or complete certain tasks. These findings suggested that better UI enhancements and clearer instructions were crucial to improving the overall player experience.
The game’s design prioritized immersion and accessibility, using real-world data integration to boost player engagement. Real-time interactions between the virtual and physical worlds created a strong immersive effect, helping players feel connected to both. This feature, which linked realtime data to in-game outcomes, was well-received, supporting the game’s educational and environmental goals. However, some technical issues slightly impacted the overall immersion.
One significant issue was the game's use of GPS for indoor location tracking, which resulted in inaccurate measurements. The initial implementation of the location service API in the Unity engine defaulted to GPS, which is generally ill-suited for indoor environments due to non-line-ofsight (NLOS) conditions. These conditions led to inaccurate tracking, further compounded by the Unity API’s insufficient documentation on indoor location tracking. Since the gameplay environment was entirely indoors, these challenges had a considerable impact on the user experience.
To address these issues, the development team implemented network-based localization techniques, which significantly improved the accuracy of indoor tracking. This solution helped bring the location tracking to an acceptable level, making the game more responsive and enhancing the immersive experience for the players. In addition, the game’s use of standard gestures, such as pinching and zooming, contributed to a more intuitive gameplay experience, ensuring that players could interact with the virtual environment in a familiar and accessible manner.
The game's design was aiming to encourage healthier behaviors and provide informal learning opportunities related to indoor air quality.
One of the main goals of the game was to promote behavioral change by motivating players to spend more time in environments with better air quality, thereby encouraging healthier habits. Players were required to be physically present on campus to progress in the game (in first and third iterations), tying real-world movement and location to game rewards. This requirement was seen as motivating to stay in campus for longer periods and encouraging to stay in areas with better IAQ.
The use of virtual pets in the third iteration served as a key mechanism for reinforcing these behaviors. Players were encouraged to place their virtual pets in locations with good IAQ, as poor air quality would negatively affect the pets’ health, causing them to devolve. This negative reinforcement mirrored real-world consequences and pushed players to seek out healthier environments, thus integrating the educational goal of raising awareness about the importance of IAQ. The design also included notifications as reminders to take breaks or move to better air quality locations.
The effectiveness of these behavioral cues was enhanced by the lack of a need for continuous active tracking. Players did not need to monitor the game constantly; rather, periodic interactions with their virtual pets acted as brief study breaks and moments to reflect on the indoor environment, subtly encouraging behavioral change without overwhelming the player.
In addition to promoting behavioral change, the game fostered informal learning about indoor air quality through its use of real-time data and interactive UI elements. The game’s interface displayed IAQ information in a way that was directly linked to the player's environment, allowing them to see how air quality affected their virtual pets and overall game progression. User feedback from the questionnaire, however, revealed mixed understanding of the app's purpose and mechanics, suggesting a gap in clear communication. Moreover, motivation to engage with the game's environmental goals was generally low, underscoring a need for better incentives and improved feedback loop.
The inclusion of notifications contributed to the informal learning process. Reminders to check on virtual pets after closing the app served as gentle prompts, encouraging players to reflect on air quality and its importance. These notifications were designed to trigger with a delay, ensuring that players remained engaged with the game’s educational goals even during periods of inactivity. The periodic check-ins with virtual pets acted as a tool for reinforcing the knowledge that better air quality leads to more positive in-game outcomes, indirectly teaching players the value of maintaining healthier environments in real life. However, malfunctioning CO₂-level alerts and other notifications, which are crucial for enhancing gameplay engagement and the educational use of real-time data, were significant issues, with participants expressing a desire for a fully functional notification system to better support both gameplay and educational objectives.
The study contributes to the field of IoT-enabled serious AR game development in four key aspects: IoT integration and data management, game design and mechanics, user experience and engagement, and educational impacts and outcomes.
The IoT integration and data management aspects include accessibility and reliability, data
structure, and real-time data access. Data accessibility and reliability were critical, with persistent
flags in MQTT broker publish events enhancing reliability when updates were infrequent. Efficient
data management required an atomized data structure and separate MQTT server topics for device
data, enabling faster game level initialization. [
The game design and mechanics include three core aspects: real-world integration, dynamic
game elements, and prototyping with iteration. Real-world integration involves three key elements:
sensor data reflecting real-world conditions to inform and impact the virtual environment, live data
integration to create a mirrored, hybrid interaction space, and game rules based on player
proximity, allowing interaction with real-world locations. Integrating real-world elements into AR
objects creates a unique experience in AR gaming, though seamlessly blending virtual and physical
content remains challenging [
Three main components were identified in user experience and engagement: motivation and
retention, navigation and interaction, and immersion and accessibility. Findings showed that key
factors for motivation and retention included faster game responsiveness and well-designed
incentives, aligning with literature linking engagement to progression and diverse rewards [
The educational impact of IoT-enabled serious AR games was observed in two areas: behavioral
change and informal learning. Real-time data interactions through notifications and UI elements
supported informal learning about indoor environmental conditions, aligning with research
showing that AR-based avatars enhance educational outcomes by combining learning with
entertainment [
However, usability tests of the third iteration identified technical and pedagogical challenges. Participants found navigation manageable but requested clearer map icons and instructions. Technical issues, such as slow responsiveness, hindered immersion and sensor data integration, reducing both engagement and IAQ awareness. Additionally, the game emphasized recognizing poor air quality over providing actionable steps, highlighting the need for more behavior-oriented design elements.
This study explored the design and development of IoT-enabled serious games using AR
technologies to raise student awareness of IAQ and promote healthy behaviors. Utilizing data from
three iterative implementations and applying Braun and Clarke's thematic analysis [
One significant limitation of this project was the initial lack of clarity in objectives. The original idea was simply to create a game that incorporated real-life data into its game mechanics, without a well-defined long-term vision or rigorous methodology. As the project progressed, the concept was gradually refined through iterative development, but the process lacked a structured approach. Consequently, the collection and categorization of research data were inconsistent. Future research must create explicit theoretical and practical foundations for the usage of IoT sensors in games. Integrating established game design principles with IoT sensor technology will result in a more structured and focused approach to IoT-enabled serious game design and development. This will ensure that future initiatives accomplish their aims and provide vital insights on how to effectively integrate IoT technology into educational and environmental awareness games. The practical application of generative AI throughout various stages of the game development process represents a promising area for future research. This technology empowers developers with limited resources or skills to design impactful games more efficiently and creatively.
The research and development during the second iteration were supported by The Finnish Information Processing Association (TIVIA) and Tietotekniikan tutkimussäätiö in 2022, along with the Metapilot Factory project funded by the Joint Task Force (JTF) for the main author in 2024. The second author received funding from the Infotech Emerging Project under the University of Oulu.
We extend special thanks to the student groups. The first group included Joni Kilpeläinen, Juuso Laivamaa, Bashir Kasozi, and Miko Korhonen; the second group, Renata Moilanen, Janne Mourujärvi, Heikki Kallankari, Tommi Lämsä, and Amine Zeggaf; and the third group, Emma Kemppainen, Jere Pesälä, Iiso Kramsu, and Juho Seinijoki. Their dedication in creating the designs and technical prototypes, along with their background research, was crucial to the project’s success.
We extend our sincere gratitude to the Smart Campus organization at the University of Oulu for their support with sensor data integration. Finally, we thank Ikune Labs for providing the research case.