I. INTRODUCTION
Smartwatches have been on the market since 1998 - “the Ruputer” - and since IBM launched the first commercial smartwatch in 2000, this technology has become a promising trend, especially from 2014 with the release of Google’s Android Wear OS, the smartwatch industry has been booming with new and exciting additions, which has seen a further increase with the Apple Watch release in 2015. The market was predicted to increase to 200 million by 2019, further proving the interest available in this, mostly, untouched market.
The main idea of a smartwatch is to complement a smartphone and emphasize its features while improving the ease of use. The issue is that everything that can be done on a smartwatch can be done directly on a smartphone, with the exception of the health monitor thanks to its proprietary sensors. These features create new possibilities for smartwatches to be used in many industries, especially in remote applications, and as an outcome, the topic of smartwatches is a major discussion in both the industry and academia.
Therefore, it would be intriguing to see the application of a smartwatch that uses a multitude of sensors to check in real-time the surrounding environment while also monitoring the health of the wearer, collecting data that can be processed and supplied through a voice assistant – that is also used for immediate commands and user interactivity.
II. MARKET ANALYSIS
Figure 1: Smartwatch shipments Q12018-Q42020
With the growing market following yearly releases from big companies such as Google and Apple, market acceptance has become the main argument for engineers and designers. As shown in Figure 1, in 2020, even with the COVID-19 pandemic affecting all sectors of the market, the smartwatch shipments rose 1.5%, with the Apple Watch Series 6 selling 12.9 million units, meanwhile Huawei had a market grow year-to-year of 26% thanks to all the online shares and the Chinese market. However, throughout the years, there has been a significant market shift towards more expensive watches, with premium brands such as Apple, Samsung and Huawei following suit. During 2020, the ASP (average selling price) increased by 8% in the $300-plus segments, with the low-cost segment shrinking by 7%.[0] Analysts believe that because of such interest in high quality premium devices, in the following years the ASP will further increase, giving manufacturers the possibilities to include more sensors and features, together with high quality materials.
Different researches argue about the significance of the user experience in the general use of smartwatches. Some researches claim that engineers are too focused on optimizing the hardware and software, neglecting hedonic qualities such as watch design and fun. [1]
Figure 2: Characteristics influencing attitude towards smartwatches
Based on data generated in Taiwan with 212 participants, it was found out that ease of use was not of high importance, as shown in Figure 2, nor was the hardware or software, but rather the perceived enjoyment of the device, and the luxury aspect of it, which is quite paradoxical since a smartwatch focuses more on interactivity and use, instead of the classical approach to materials and design of luxury timepieces. These characteristics are a significant factor in changing the attitude of customers towards smartwatches, which is the main driving force behind them. [2]
Figure 3: Smartwatches as IT Innovation and Luxury Product
From another point of view, seen in Figure 3, smartwatches can be seen as an IT product or a fashion product. From the findings of 562 Korean interviewees, it is of importance to note that the design of a smartwatch as a fashion product significantly increases the intention to use one, this is further strengthened by the individual’s preference for distinctiveness. A limited effect of vanity on self-expressiveness implies that the smartwatch is not yet deemed a luxury commodity.
In order to further develop this study, the present research focuses on two perspectives: the smartwatch as technology and IT innovation, and the smartwatch as a fashion luxury product. For the initial view, perceived interactivity and usefulness in day-to-day activities were significant factors in what people found as innovative. As for the latter, it seemed most people preferred a smartwatch design similar to a regular timepiece with high quality materials and a unique design, matching that of a Rolex or Patek Philippe. On a different note, another issue of high importance is the fact that a smartwatch is simply an extension/remote of the smartphone, therefore, deeming it redundant. [3]
The data shown in Figure 4 was collected and analyzed in cooperation with Macromill Embrain Co., an international survey firm, with HQ in Korea, in possess of the largest survey panels available on the market in the Asian region. The questionnaire took place in Korea because of its large development of technology and the population’s embrace of it, generating a better scenario for analyzing the interest in such devices. The subjects were selected using stratified sampling based on all age groups from 20 to 60 years old, as well as from different educational backgrounds and genders. From these numbers, 50.50% were male and 49.50% were female [2]. It was observed that the female respondents had a higher interest in the design of the watch, and that the education level of both males and females, does not influence their beliefs. Furthermore, people with interest or present in a technological environment do not have any concerns on the aspect of utilization, as they perceive smartwatches as easy to use. [2]
Figure 4: Interest in smartwatches based on age/gender/education in Korea
III. FEATURES AVAILABLE ON THE MARKET
On all smartwatches present on the market, there is already a suite of sensors such as a GPS, gyroscope or heart-beat sensor. The features already available based on these technologies include heart rate monitoring, oxygen levels present in blood cells and fitness capabilities; meanwhile, on the other side of the coin, accelerometers and gyroscopes, are used for user interface attributes, such as being able to change between applications or notification with just a flick of the wrist.
Figure 5: Accelerometer signal generated
The most common and used feature on any smartwatch is the fitness feature that works by using a GPS and an accelerometer sensor. These sensors generate data presented in Figure 5, and are used with mathematical algorithms can filter the different motions and assign them to activities, described in Figure 6. Some of these activities include clapping, walking, running, jumping or even fall detection. [32][34][37]
Figure 6: Activities model for accelerometer/gyroscope data
So based on these functions, we can define that there is a readily available platform which can be used for motion-based behavioral biometrics. That means that by performing advanced research on the signals created, a larger division with more activities can be created. These could consist of sleeping routines, morning routines, daily hydration patterns and eating movements. The main issue with such a highly developed system, presented in figure 7, is that a specific controller for signal filtering is necessary, this would require a lot of research and development resources that companies may not want to spend, [6] [7].
Figure 7: Smartwatch system diagram
Furthermore, the same sensors can be used for posture tracking, a smart way of fixing one of the most common issues during our times because of all the office work. This can be done by using the onboard IMU together with standard models that are associated with different posture positions. Moreover, as can be seen from Apple’s developments, the same system can be used to develop a fall system. Falling is a big health risk, especially for individual elderly people, such a feature will enable the elderly to live with a lower risk by having a mechanism that can alert emergency services in case of emergency. But, in order to achieve this, a high accuracy is required to prevent negative situations that may occur. [7]
For the health monitoring however, there are two types of sensors that are usually used in such a device, a heart-rate sensor and an SpO2 sensors which is able to measure the oxygen level present in the blood by sending and receiving a light beam through the finger’s skin and blood vessels. [8]
All of these features are key selling points of smartwatches and in the future, through further research and optimization, will become a must for any user, and a complex computer, se
IV. EXTRA CAPABILITIES
Beyond the actual market, multiple studies have been carried out on additional algorithms used in tandem with an array of sensors, for advanced interactivity, as well as health and environmental monitoring, which add extra functionally compared to a smartphone, rendering the smartwatch a standalone ultra-modern device with specific uses, which complements the main point of this literature review.
For example, with the use of optical sensors, irregular pulses can be detected, which could be used to identify atrial fibrillation.
With a number of 419 297 respondents, over a period of 8 months, 2161 users were notified of having an irregular pulse with this system. From this number, 450 decided to have a professional ECG performed, which showed atrial fibrillation for as much as 34% of the people aged 29 to 39, meanwhile for those with an age over 65, 35% had heart issues [11]. By using such a system together with a medical ECG, the risk of heart attacks can be highly reduced, to further solidify the point, 57% of all the participants contacted health care providers to further check their health. Such a feature could highly increase the interest and sales numbers of a smartwatch. [11][35][38]
Furthermore, epidermal sensors which could be mounted on the back of the watch in order to form a perfect contact with the skin, are able to provide information about cardiovascular health, electrophysiology and dermatology, in an unobtrusive manner, without the need of a doctor appointment and high costs. However, to enable a clinical application, such a device would need to be worn for a continuous long period of time in order to have results of high accuracy. [12]
On top of this, a research by Michele Magno, titled “Autonomous smartwatch with flexible sensors for accurate and continuous mapping of skin temperature”, talks about the creation of a modern system which uses low-power epidermal sensors together with a controller, to map the skin’s temperature, fortifying the medical capabilities of a standalone smartwatch. [13]
Figure 8: FOQUS Application Diagram
Outside of this topic, during a conference in October 2016, an app, called FOQUS, seen in Figure 8, has been presented running on a smartwatch, that is able to aid adults with mental disabilities such as ADHD and attention deficiency. In order to achieve this, the system uses a flexible implementation of the Pomodoro time management technique, a tool used for guided meditation and positive thinking. The initial testing indicate that such a tool is viable long term, and that it can address mental health and stress related conditions.
V. MACHINE LEARNING
As explained earlier, algorithms can be used to detect mundane activities, if however, these algorithms are combined with machine learning, this functionality can be further advanced, by having pattern detection and analysis. By generating and analyzing daily pattern with a neural network, changes in physical and mental behavior can be observed, this can be further compared to models already analyzed in order to understand the behavior of the user, and in case of an emergency, an accurate report can be generated or the emergency services can be alerted. Such a technique would highly help individuals suffering from depression or that have suicidal tendencies. In order to identify such patterns, the data is based on non-overlapping 10-second intervals, which is then analyzed and mapped to an expected activity. Research has demonstrated that using personal recognition models generated specifically for the user by the AI, and then comparing the newly analyzed data with what is expected, vastly outperform impersonal models and provides way more accurate results. [12][31][39]
VI. UNCONVENTIONAL SENSORS
From an experimental note, there are many sensors available on market that have not been used and nor researched in the field of smartwatches.
Figure 9: Temperature user feel for Comfort Equation
For instance, in physics there is a formula called “The Comfort Equation”, such an equation is used to determine the comfort grade of a room or environment, based on temperature, humidity, pressure, air quality, bulb colour etc. as can be observed in Figure 9. By adding such simple and sensors to a smartwatch, yet another feature can be added, a small wrist-worn low power device that can detect, calculate and compare with existing models, to determine the exact grade of comfort of the environment around the wearer. Additionally, the pressure sensors can be used to detect the air pressure which can have a negative impact on the elderly and therefore, it can determine is a person aged 60+ should continue being present in that environment. [15]
Moreover, the atmospheric composition, as well as pollution level, can be determined by using sensors as the MQ-135. These sensors are called “electrochemical sensors” and are based on a chemical reaction between gases in the air and the electrode in a liquid inside a sensor. By utilizing such a sensor, within a time frame of around half a minute, it can determine the percentage of NO2, SO2, O3, NO and CO present in the air, and from here by using the machine learning and algorithms already discussed, the user can be alerted in the values found are over the threshold limit.
Figure 11: MEMS Magnetic Sensor Data Chart
Magnetic fields are generated by the Earth, thunderstorm or even electricity, a short-term exposure to very high levels of electromagnetic field can be harmful to human health, meanwhile low-level electromagnetic field do not pose any threats and are not harmful [16]. But, by using a MEMS ultra-sensitive magnetic field sensor in a smartwatch, Figure 11, the extra capability of detecting effects of the Lorenz force is added, which in turn can calculate the level of the fields around the wearer and if it poses a threat to the user. [17] [18]
On the topic of user safety, a flame sensor can be added in the bezel or case of the watch, so that in case of a fire emergency, the watch is able to call the emergency providers without any user interaction, especially in such a situation where stress and panic levels can be high. Furthermore, a shock sensor is able to detect large shock waves, such as a glass window being broken, or a gunshot, which can alert the user of the danger present and even perform an emergency action. [26]
From an interactivity point of view, as of now, the main interaction happens through a touchscreen display, but in order to overcome these limitations, analog infrared sensors could be mounted along the bezel of the watch in order to create an invisible area of control, from there the user can move their hand, and through gestures the software can be controlled. Such technology already exists and is used on devices such as the Pixel 4 from Google. By combining this 3D Area of effect with wrist movements, a cheap, low power touch free interface can be created, overcoming one of the biggest problems with smartwatches, being their small display screen, which is particularly hard to use by seniors. [20][30]
VII. VOICE ASSISTANT
Figure 12: Voice Assistants Characteristics
Artificial Intelligence technologies are emerging in fast pace, and they can be adopted even in the field of a smartwatches for hand-free interactivity [21]. Currently, voice assistants are embedded in a variety of products such as smartphones, smart speakers and even more specific appliances such as smart microwaves or fridges, the main leader in this field being Samsung and Amazon. But this technology is also present in smartwatches, and since the two main operating systems on the market use Apple’s Watch OS or Google’s Wear OS, usually the way voice assistance is implemented is with the use of Siri or Google Assistant, systems that are also present on a smartphone, and that require a 24/7 internet connection in order to function. Furthermore, the information that such an assistance can generate is usually statistical or factual information can be found with a search engine manually by a user, and the information is never directly related to the wearer/user itself [22].
The main idea is to create a model with the answers expected and then based on this model, generate audio responses with a powerful neural network, that does not require to be present on the device itself, nor does it require to be remotely accessed through the internet of things. Furthermore, the responses can be divided into mini-phrases that can be attached together with others, to form proper phrases. As an example, the response “The temperature in this environment is 27 degrees”, can be divided into the root “The temperature in this environment is”, plus the attachment “is x degrees”, where x is the value calculated from the temperature sensor on board which then selects the correct rest of the phrase. These responses could just be saved as high-quality audio files on the smartwatch’s memory and then played in the order required based on the sensor data and question from the user.
Due to such an approach to the voice assistant system, for further interest and novelty, rather than using a neural network that performs a text to speech process, such as IBM Watson or CereProc’s service, a voice actor could be used to record the assistant’s voice, which would then remove the “robot-like feel” of the responses, an example would be the way J.A.R.V.I.S., an artificial intelligence assistant, was created in the Marvel movies, by having an actual actor perform the voice lines.
Humanizing voice assistants, described in Figure 12, is a major topic of debate nowadays because of the high presence of assistants on the market, and the people’s scarce interest in using them.[28] Researchers found out that humanizing the AI voice, in order to create a more “to life” voice, would highly increase the interest in using such a technology. Moreover, this could also empower medical services for mentally-challenged people or children, where a human, natural voice, can lead to satisfaction and a higher willingness to continue using the device [22]. By using a pre-recorded human voice, this issue is completely eliminated. Of course, such a technique comes with its own disadvantages, which in this case is the ability to perform a natural unprecedented conversation, while using topics and data from previous questions, something that Google’s Assistance is able to achieve. [27]
On the other hand, if a neural network assistant is used, other than just using the assistant as a means of interactivity for the device, the voice patters and intonations used by the user could be analyzed by the network, in order to sense elements of depression, sadness, stress and anger, therefore, from a medical point of view, this approach would have multiple uses, although it would require an always present internet connection. Training the neural network can be achieved by querying individuals about the mundane experiences, activities, feelings and thoughts in a day, as well as analyzing the change in speech based on the aforementioned categories. [23][29]
VIII. DESIGN
As explained throughout the first section of this document, the design of the device is the biggest selling point, requiring high quality materials, uniqueness and a luxury look and feel. However, such a device could be used in the fields of medicine, military and space, therefore, a different design would be required based on the field since the capabilities of a smartwatch would be requested everywhere.
Thanks to the potential of smartwatches to measure valuable physiological signals on the move, there is a high interest in the community for elderly care, required in assisting nursing, and because of this, the design process has been heavily changed to include necessary features in the devices aimed at the elderly. Such features include an easy-to-use watch, with low weight, neutral materials in order to not cause allergic reactions, as well as a silicon strap that is easy to remove and does not create much pressure on the wrist. [33]
For the military, a rugged design is required, strong materials such as steel, ceramic and titanium alloys can be used to reinforce the casing of the smartwatch in order to have it capable of withstanding rough handling and environments. It would also require a large battery combined with many low-power sensors since the probability of recharging the device is way lower compared to the average use.
Figure 13: Concept design for smartwatches space usage
For the space industry a very minimalistic approach is required, seen in Figure 13, with the sensors present on a wrist band outside the space suit, as shown in the image, and the assistance implemented in the suit of the astronaut for ease of use, having environmental data presented with just a voice command. The wrist band must have the components divided into sections, in order for the wrist band to be removable but easy to link with the rest of the system present in the suit. Furthermore, the device must work in extreme weather and temperature, while being dust and water proof.
For a user centric device, luxury and uniqueness is of utmost importance, together with high quality materials creating a premium device comparable to classic timepieces. Figure 14 presents an angular two-tone design of the case and bezel, with sapphire front glass and a ceramic back, complemented by a stainless-steel band with design elements taken from a Maglia Milanese. Such premium materials and design are sure to attract interest from customers in search of a thrill. The problem with this approach is that because of the small dimensions, similar to that of a mechanical watch, many sensors cannot be added, and most of the functionality has to be based on the algorithms done through software. There is however, a current development in battery technology that will enable engineers to include the battery in the strap itself, therefore, increasing the space available in the case for different technologies [25].
Figure 14: Concept design for smartwatches as customer luxury product
IX. CONCLUSION
This paper presented an overview of the market and features available in the field of smartwatches. The next stage consists in developing new technologies that can be applied to this field, as such devices can prove a gold mine when it comes to health aspects, as well as being financially profitable for the companies engaging in their production.
Based on all research, we can see that the smartwatch industry is getting quite the traction and continues to increase tremendously in popularity. Therefore, new research on the available technology should be performed to understand the capabilities present and the new technology required to push the boundaries of what a smartwatch can do. Many research papers already exist on the topic, testing different methodologies for achieving diverse tasks that can highly change the attitude towards smartwatches and their usage.
The previously mentioned “perfect” smartwatch, can be built as soon as there is a request for such devices on the market, and in order to drive interest, creative visions must be undertaken to demonstrate what can be done; a catalyst, just the way Tesla was in the field of electric vehicles, which has shown other manufactures what could be.