2.1. Location-based games

By combining the augmented reality capabilities, exercise games (exergames), and location-based multiplayer features, Pokémon Go exploded onto smartphones in the summer of 2016. Many users have reported that their daily activity had increased since starting to play games such as Pokémon Go (Althoff et al., Reference Althoff, White and Horvitz2016), which shows that exergames can have a great benefit to the user’s health. In the short term, the benefits that exergames bring to users’ health are undeniable. However, for a user to maintain these health benefits for the long term, consideration of how to keep the user engaged has to be considered (Althoff et al., Reference Althoff, White and Horvitz2016). Pokémon Go is still widely played by many users, with 147 million still playing as of May 2018 and over 1 billion downloads as of May 2019, which shows that this game has successfully kept users engaged for the long term (Wang and Skjervold, Reference Wang and Skjervold2021).

While Pokémon Go is still one of the most popular location-based games, there has been a rise in the number of location-based exercise games to promote physical activity. Some of these include “World of Workout” which is a mobile exergame that detects users’ steps to play a conventional role-playing the game, completing quests by walking a specific number of steps (Doran et al., Reference Doran, Pickford, Austin, Walker and Barnes2010). Similarly, in PiNiZoRo, players had to find enemies by walking and then complete puzzle games to defeat them (Stanley et al., Reference Stanley, Livingston, Bandurka, Kapiszka and Mandryk2010) and Zombies, Run! encourages players to run in different directions in order to avoid virtual zombies (Clare, Reference Clare2014).

All of these games promote physical activity, such as walking or running, using only a smartphone app; however, they all rely on GPS, not taking advantage of advances in IoT and AI to enable new game options and increase engagement. For example, current location-based games are often focused on the game logic and the reward mechanism rather than the connection between the player and the environment, and the space around them.

Location-based games that rely purely on GPS also include more technical issues due to the limitations and inherent flaws in GPS. In one study, GPS service was problematic, particularly on the smartwatch, and required the linked smartphone to be nearby to solve the problems they were facing (Vukovic et al., Reference Vukovic, Car, Fertalj, Penezic, Miklausic, Ivsac, Pavlin-Bernardic and Mandic2016). When users were asked about their experience using “Pokémon Go,” GPS and networking issues were among the list of technical problems reported (Paavilainen et al., Reference Paavilainen, Korhonen, Alha, Stenros, Koskinen and Mäyrä2017). Since most location-based games rely heavily on GPS for location tracking, they do not link players to their local environment or key points of interest, which means the same game can be played in any place without giving the current area or venue any consideration. This demonstrates the need for utilizing new emerging technologies that can incorporate AI and IoT for proximity and context awareness, as there are currently no stand-alone wearable location-based games.

2.2. Wearable technology

Smartwatches are a wearable technology that has become increasingly popular in recent years. Although some of these wearables are capable of running standalone, most commercial wearable devices ultimately have to rely on a mobile phone and are intended to be companion devices to a mobile phone to help better monitor health, deliver notifications, and bring some basic applications to the wrist. These advances create more opportunities in several areas, including the area of gaming (Seneviratne et al., Reference Seneviratne, Hu, Nguyen, Lan, Khalifa, Thilakarathna, Hassan and Seneviratne2017). Advances in the communication technologies present on wearable devices have improved too, with most modern wearables being equipped with several different communication methods such as Wi-Fi, Cellular, and Bluetooth Low Energy (BLE) (Sun et al., Reference Sun, Zhang, Hu and Qian2018).

In terms of wearable location-based games, there is a range of gaming options available on smartwatches. Previously, six different categories for smartwatch games were drawn out, including location-based games where physical activity, GPS, and sensors are used for game progression (Asadi, Reference Asadi2020). This shows that there is the capability and demand for location-based games on wearables, considering their increasing popularity on smartphones.

The year following the launch of Pokémon Go, Niantic introduced an accessory called the Pokémon Go Plus. The Pokémon Go Plus is an optional wearable in addition to the Pokémon game that allows players to complete key game actions without the player needing to look at the mobile phone (Sablatura and Karabiyik, Reference Sablatura and Karabiyik2017). This demonstrates the benefits of a wearable approach to location-based gaming. However, the Pokémon Go Plus wearable did not embed a screen and therefore offered very little interaction or gamification. Alternatively, a location-based smartwatch application tracked people with complex communication needs, although this did not include gamification to motivate players and had to rely on a smartphone as backup as it was not capable of running independently (Vukovic et al., Reference Vukovic, Car, Fertalj, Penezic, Miklausic, Ivsac, Pavlin-Bernardic and Mandic2016).

Smartwatches provide many opportunities for healthcare monitoring with many able to monitor physical activity such as steps along with physiological data such as heart rate. However, there has been little consideration of gamifying a smartwatch experience for physical activity monitoring, with many existing health monitoring technologies requiring additional, often expensive hardware such as virtual reality headsets.

Much work has been completed on location-based and ubiquitous games since the emergence of smart mobile phones. However, our proposed system is significantly different in terms of the utilization of wearable technology to facilitate the game interaction in open space using sensors and IoT features for health monitoring. Smartwatches bring many benefits for remote health monitoring, and the addition of a gamified experience to motivate users to be more physically active presents unique opportunities.

3.1. Gameplay

The aim of the game is to connect users with their local environment while also promoting physical activity through the use of gamification. The game offers a proximity-based treasure hunt experience to the user combined with real-time step indicators. The goal of the game is to walk around the area to try and find all of the hidden treasure hunt locations. This is aided by on-screen representations of the location to visit along with visual indications of how far the player is from the location, as shown in Figure 2. As users walk closer to one of the treasure hunt locations, the background color of the screen will change from red to amber to green, allowing users to quickly glance at the watch to understand their relative proximity to the next location. There is no specific order in which users must find the treasure hunt locations, and not all locations need to be found in a single gaming session, ensuring the game can be played repeatedly. Furthermore, the treasure hunt locations can be simply changed by moving the Bluetooth beacon, allowing the game to be regularly updated. We envision the game being utilized in large areas such as local parks and in unfamiliar locations such as a university campus to gamify the experience of showing new students around the campus.

Figure 2. The developed wearable game showing different locations to visit along with score and step count and the final game completion screen.

3.2. Physical activity monitoring

A core part of the system is the cloud platform for the storage of the game results and physical activity metrics. The smartwatch records the user movements and steps, storing all of the data locally, while at the same time, indicating the results live on the watch screen.

Once the user reaches one of the treasure hunt locations, all recorded data (step count, number of locations visited, and timestamp) are automatically uploaded to the real-time database for remote monitoring. The use of metrics allows for care professionals to remotely assess different parameters of users’ physical health over time, enabling trends to be established.

3.3. Proximity detection

A vital aspect of the application is the ability to wirelessly scan for nearby Bluetooth beacons and calculate their approximate distance (Figure 3), enabling the app to understand when a treasure hunt location has been found. The developed smartwatch application uses the following RSSI equations to calculate the distance the player is from the nearby BLE beacons.

Figure 3. BLE smartwatch system schematic.

Suppose that the distance estimation is based on M samples of $ {RSSI}_{k,i} $ , which represents the ith RSSI sample measured by the kth the receiver node. For getting a good performance, the median value of $ {RSSI}_{k,i} $ is used to obtain the distance estimate:

(3.1) $$ {d}_k=10\left(\frac{A_k-{RSSI}_K}{10^{\ast }{n}_k}\right), $$

where $ {RSSI}_k $ is the median RSSI value measured by the kth and the receiver node is given by

(3.2) $$ {RSSI}_k= MEDIAN\left({RSSI}_{k,i},i=1,\dots, M\right). $$

Once a location is detected, the approximate distance can be measured. The distance of propagation path loss show the channel fading characteristic follows a lognormal distribution. Thus, the instant RSSI distance measurement generally uses the logarithmic distance path-loss model and the propagation model that reveals the corresponding relationship between distance and RSSI can be expressed as Equation (3.3) (Faragher and Harle, Reference Faragher and Harle2015):

(3.3) $$ RSSI=-10 nlog\left(\frac{D}{D_0}\right)+A+{X}_{\sigma }, $$

Where RSSI is a dependent variable of the received signal strength indication, D is the estimated distance between the transmitter and the receiver, and n is a path-loss parameter related to the specific wireless transmission environment. The more obstacles there are, the larger n will be. A is the RSSI with distance $ {D}_0 $ from the transmitter, which is a constant value. $ \sigma $ is a parameter representing the path loss exponent while $ {X}_{\sigma } $ is a Gaussian-distribution random variable with mean 0 and variance $ {\sigma}^2 $ . For the convenience of calculation, $ {D}_0 $ usually takes a constant value. Since $ {X}_{\sigma } $ has a mean of 0, the distance-loss model can be obtained expressed as Equation (3.4):

(3.4) $$ RSSI=-10{n}_k\mathit{\log}\left({d}_k\right)+{A}_k, $$

where $ {d}_k $ is the distance from the unknown transmitter node to the kth the receiver node, and $ {A}_k $ and $ {n}_K $ are the model parameters of the kth receiver node. $ {A}_k $ is the measured RSSI when the received node is a fixed distance away from the transmitting node. The $ {n}_k $ parameters are relevant with the wireless transmission environment which can be obtained through the optimization of many experimental measurements. $ {A}_k $ depends on the transmitting power of Bluetooth. Ideally, $ {A}_k $ should be determined by specifying one of the Bluetooth signals.

Accurately calculating the relationship between RSSI and distances using the logarithmic distance loss model due to complex environments is extremely difficult for researchers (Subhan et al., Reference Subhan, Khan, Saleem, Ahmed, Imran, Asghar and Bangash2022). Therefore, there are several methods for modeling the RSSI with the most accurate distances based on any application system. However, several applications do not require a high accuracy localization, but need area-based localization. Similarly, this article does not require the actual position of the tracked object but rather ensures that the smartwatches are within a certain range of the POI.

4.1. Methodology

A total of seven participants were recruited to complete the game finding six different locations using the same smartwatch and the same BLE beacons. Participants repeated the game in six different conditions: indoors with limited interference, indoors with high pedestrian traffic, indoors with high BLE traffic, outdoors with limited interference, outdoors with high pedestrian traffic, and outdoors with high BLE traffic. While playing the game in different environments, the distance the wearable detected the BLE beacon was measured to explore the impact pedestrian and building interference has on the gameplay.

4.2. Results

The developed smartwatch application demonstrates an exergame that is capable of running entirely independently on a smartwatch, not requiring a connected smartphone like many commercial smartwatches. This section outlines the key performance indicators when testing the watch playing the game.

4.2.1. Bluetooth

Accurate Bluetooth performance is vital for the developed gaming platform to find the POI and enable the virtual treasure hunt. Table 1 shows the average Bluetooth performance results of the indoor gaming sessions with limited interference measured at 1 m and 4 m distance from the POI after 5 s. The RSSI equations, described previously in Section 3.3, were used to estimate distances based on the measured signal strengths during testing.

Table 1. Bluetooth RSSI and distance at 1 m and 4 m intervals across five indoor locations with limited interference, using the developed wearable gaming platform

The results demonstrate that Bluetooth measurement is not reliable for accurate distance measurement even with little interference as the accuracy widely varies even though all locations are indoors with similar surroundings. However, it is possible to measure approximate distances such as 1–2 m and 3–4 m. There is significantly more variance at 1 m compared with 4 m showing small distances are more difficult to reliably measure. For the developed game on the smartwatch, precise measurement is not required; instead, it is only required to ensure the user has visited the location (~4 m), which can be reliably achieved using BLE.

Further experiments were conducted to explore the impact on different environmental conditions on the RSSI signal. The experiments explored the impact of indoor and external conditions without artificial interference, high BLE traffic where numerous BLE devices were placed between the wearable and beacon, and finally pedestrian traffic where people continuously walked between the wearable and the beacon. Figure 4 shows the difference in performance for BLE and pedestrian interference both indoors and outdoors. The test was conducted similar to the initial test with the RSSI measured at 1 m and 4 m distances at the six different conditions. The results show that outdoor and indoor environments have a significant impact on the RSSI value at both 1 m and 4 m distances. At a 1 m distance, the received signal strength was stronger in the outdoor conditions, whereas at 4 m, the received signal strength was stronger in the indoor conditions. This may be because there is more potential for interference outdoors such as weather and obstacles when measuring over larger distances. Surprisingly, pedestrian or BLE traffic had very little impact on the signal strength, with the largest variation being outdoor BLE interference. The largest variation was indoor pedestrian traffic which increased the RSSI value by 18.2% with all other interferences having less than 10% impact on distance accuracy. The BLE and pedestrian traffic had an average accuracy impact of 1.6% at 1 m and 4.7% at 4 m showing that while interference increases as the distance increases, it has little impact on overall accuracy.

Figure 4. RSSI (dBm) values across six different conditions at 1 m and 4 m distances.

5.1. Display

The small 1.3″ display on the Lilygo T-watch presented challenges in displaying detailed game graphics or extensive tutorial instructions. To provide an intuitive user experience within the screen space, simple color changes and icon representations were used to indicate proximity and locations. However, additional tutorial content may be necessary for first-time users to understand this non-textual guidance. Future iterations could investigate minimalist graphical tutorials optimized for tiny wearable displays.

5.2. Offloading tasks

By introducing a novel stand-alone wearable game, we are demonstrating the growth and possibilities of the wearable technology. For example, one area of growth was the little processing power available to wearable devices because of their small footprint. To help resolve this limitation, methods of how to offload heavy computational tasks to either the paired mobile or to a server could be used, but this can increase latency and therefore the overall time it takes to achieve the task. We successfully implemented the game solely on the wearable device, negating any need for additional hardware, making the game more accessible to those without smartphones, such as children, while also making the game easier to view and use in real-world environments. However, we did encounter limitations when scanning for numerous BLE beacons simultaneously, resulting in limiting the total number of locations.

5.3. Physical device constraints

Wearables are limited due to their size and what they can offer. For example, all of the mainstream smartwatches do not embed a camera, as placing a camera on a small form factor is challenging based on size constraints and difficulties placing it in an optimal position to give the best results. The exclusion of additional sensors, such as a camera, limits potential game options. However, by making the game passive, whereby users do not need to actively interact with the screen or additional sensors it makes the game simpler to use in real-world environments.

5.4. Wearable power

The smartwatch battery life of 1 hour limited continuous gameplay time. A benefit of not having consistent companion communications and GPS communication in this game means that the battery is capable of lasting longer, which would be required considering the consistent BLE scanning required for the game. The constant BLE scanning and frequent display updates for game graphics were the primary battery drain factors. Optimizing the scanning duty cycle and reducing unnecessary interface updates could extend the playable time per charge. However, even with current battery life, gameplay sessions exceeded most standard daily exercise recommendations of around 30 min (National Health Service, 2018). Longer-term deployments would need to consider typical smartwatch charging patterns to ensure availability throughout the day.

Some wearables do not have an interactive screen and they therefore have much more battery life in comparison to other smartwatches that do have screens. One of the most power-consuming items on wearables and smartphones is powering a display. Research shows that battery life of wearables is significantly greater in those without interactive screens (Liang et al., Reference Liang, Xian, Liu, Fu, Zhang, Tang and Lei2018). These screens are much smaller than their mobile phone counterparts, but so is the battery powering these devices. However, the screen is necessary aspect of the developed exergame to inform users of their proximity to the locations while also offering additional context such as step count.

5.5. BLE proximity detection

Frequent BLE scanning for multiple beacons caused some instability in the proximity detection accuracy. In testing, measured distances varied from 60 to 120% of the actual distance at a 1-m separation between the wearable and BLE beacons. This variability arose from interference between concurrent advertising from multiple beacons. To mitigate this, the beacons were placed over 20 m apart, and median RSSI values were used to calculate proximity. While this allowed general positioning, more robust statistical filtering techniques could improve stable accuracy under dense beacon deployments.

There are many factors that can affect Bluetooth’s received signal strength, including (i) emission power, (ii) emitting device antenna path, (iii) fight path, (iv) receiving device antenna path, and (v) receiver sensitivity. The first two are tied to the emitter and the last two are tied to the receiver. The middle one is tied to what happens between the two. However, research has found that the greatest signal attenuation and variation was caused by pedestrian traffic blocking the line of sight between transmitter and receiver, which is something to consider when placing the Bluetooth beacons for the treasure hunt (Kwok et al., Reference Kwok, Wong, Griffiths, Wong, Kam, Chin, Xiong and Mok2020). High temperature and strong winds also caused minor discrepancies to the signals, whereas trees and nearby vehicle traffic did not have any negative effects on the signals (Kwok et al., Reference Kwok, Wong, Griffiths, Wong, Kam, Chin, Xiong and Mok2020). This highlights the importance of the placement of the BLE beacons for the treasure hunt locations. However, as beacon detection is only required within a large area (~4 m²) for the exergame rather than accurate measurement, this is not a major limitation.