The advent of the Fourth Industrial Revolution has resulted in significant changes in the human lifestyle, such as increased life expectancy and a higher prevalence of single living [1]. In particular, the rapid development of service environments that apply smart media ensures that these technologies are involved throughout everyday life. This revolution was founded upon information and communications technology developed in the 20th century. It was realized in the method of combining software with machine along with increased use of Internet and technological breakthroughs in information-processing capacities. Value creation in the context of this technological revolution is rooted in data. Current core technologies include the internet of things (IoT), artificial intelligence (AI), big data, mobile devices, and clouding computers [2]. New service markets are emerging on these platforms [3], and businesses, such as Airbnb, smart farms, and smart beauty, continue to grow.

The aforementioned technologies have been implemented in the beauty industry in various ways. Consequently, the beauty industry is expected to transcend the simple expression of beauty and develop into a customized private service. In such a trend, a representative technology is the IoT-based smart beauty service. This is a novel form of service that uses digital technology, wherein smart devices are combined with existing beauty services, and customized product recommendations are generated using AI [4]. Beauty products and services stimulate consumers’ sensitivity, and their values increase as a result of mechanization and automation. Among five senses of human beings, the sense of smell is known as what people tend to remember for the longest time among five senses. Perfumes for olfactory expression can be used situationally and serve as a second layer of fashion [5]. Moreover, the aroma of the same perfume may differ among individuals when it combines with skin-secreted body odors.

This study aims to provide basic data pertaining to the growth of smart beauty and perfume as a future high value-added market by developing a customized perfume service device that analyzes individual body odors and recommends matching perfumes using IoT-equipped devices.

Theoretical background

Body odor

Human body odor is an unpleasant smell emanated by sweat, bacterial remains, and odorless compounds generated through metabolic actions and secreted from the sebaceous glands. This odor varies among different body parts, including the axillae, feet, and skull [6-9]. The bacteria predominantly responsible for producing body odor are Actinobacteria, Firmicutes, Proteobacteria, and Bacteroidetes [10]. Body odor is particularly abundant in the forehead and upper part of the back, where sebaceous glands are well-developed. In these areas, it is produced primarily by Propionibacterium and Staphylococcus spp. [11].

Ingredients known to induce body odor include acetic, propionic, and isovaleric acids, as well as diacetyl (2,3-butanedione). The biological mechanisms of yeasts and lactic acid bacteria involved in the formation of diacetyl (2,3-butanedione) have been well-studied [12]. While biological mechanism about yeasts and lactic acid bacteria involved in formation of diacetyl (2, 3-butanedione) has been well researched [13]. Another major component of body odor is (E)-3-methyl-2-hexenoic acid (3M2H) [14], which is also the primary cause of the pungent sulfur odor associated with volatile sulfuric compounds containing a chemically related acid, such as 3-hydroxy-3-methylhexanoic acid (HMHA) or 3-methyl-3-sulfanylhexan-1-ol (3M3SH) [15,16]. Deodorants that mask body odors suppress excessive sweat secretion by stress for a certain period of time. However, to effectively eliminate body odor, the activity of bacteria and other responsible components must be minimized.

Perfume

Aroma dominates the sense of smell and accounts for as much as 75% of various causes stimulating human emotion [17]. Perfume allows the aroma’s recipient to think of the perfume user without looking at them. Perfume is derived from the Latin term per fumum, meaning “through smoke.” By emanating a pleasant aroma, perfume creates an attractive image of its wearer to other people [18].

The ingredients used in perfumes may be extracted from flowers, fruits (seeds), resins, branches, leaves, roots, or the whole plant. Vegetable-based aromatic materials, which exhibit volatility, are the primary components of compound aromatic materials. Most animal-based aromatic materials are extracted from the excreta of animal genital glands, including the musk, civet, castoreum, and ambergris, using alcohol. As the extraction of these materials is highly expensive, they are not widely used [19].

Synthetic aromatic materials frequently used for perfumes include isolated aroma chemicals acquired from simple chemical treatments, such as extraction, distillation, and separation; and pure synthetic aromatic materials produced by organic synthetic reactions. Isolated aroma chemicals, also known as extracted aroma chemicals, are obtained by separating specific aromatic ingredients via the physical or chemical manipulation of naturally abundant aromatic materials. Depending on the proportion of the aroma, perfume may be classified as perfume, eau de perfume, eau de toillet, eau de cologne, or shower cologne. As the volatile speed varies according to molecular size, the aroma emission levels may be classified into top, middle, and base notes. In addition, they can be classified by affiliation, as listed in Table 1 [5].

Table 1 . Characteristics of perfume per affiliation.

Affiliation	Characteristic
Aldehyde	A flavor that is highly diffuse, rich in flavor, and luxurious
Aqua	A cool flavor reminiscent of the sea
Chypre	A scent in which the leaves are moist and soothing, quiet, mature, and elegant
Citrus	A refreshing aroma of citrus fruits
Floral	A fragrance used by various flowers to express the scent of flowers
Fruity	Expressing the scent of fruit, a refreshing and attractive flavor
Gourmand	A flavor that reminds of delicious food and turns olfactory stimuli into palate pleasures
Green	A fresh and refreshing aroma reminiscent of nature
Musk	A scent made from the secretion obtained by drying the muskles in the muskpole with a pouch near the genital line of musk deer
Oriental	A fragrance expressing the mysterious and erotic image of the Orient
Spicy	A stimulating aroma reminiscent of spices
Woody	A scent reminiscent of a tree

Because perfume aroma is emitted as a mixture of body odors secreted from the skin, certain aromatic differences may arise depending on the application area. Therefore, appropriate perfumes must be recommended by analyzing the body odors of different individuals.

Internet of things

IoT is a system of mutually connected [20] with wired or wireless communication functions. Most objects within an IoT environment can be connected [21]. These devices must be associated with unique IP addresses, and contain sensors and core technology to collect data from the external environment [22]. As a representative case, an air conditioner receives indoor temperature information collected from a temperature sensor and uses it to adjust the indoor temperature. Various devices in greenhouses, such as heaters, coolers, and sprinklers, process environmental information via algorithms embedded in a management server and help automatically sustain plant growth. IoT technologies are ubiquitous throughout everyday life, and are expected to be used more across multiple fields.

System design

This study developed a system that recommends appropriate perfumes to customers. To this end, an IoT device equipped with a body odor analysis sensor was used to sense the customer’s body odor and transmit it to the management server through a network. The management server can store and manage customer-specific information through a database and subsequently use it to generate personalized perfume product recommendations for customers.

Fig. 1 illustrates the overall structure of the system.

Figure 1. The system structure. IoT, internet of things.

IoT devices

The IoT device that senses the body odors of customers consists of a sensor, power, and communication modules, as shown in Fig. 2. This device is connected to a microcontroller unit (MCU), which controls it. The sensor module includes a sensor to monitor environmental temperature and humidity, an image sensor to photograph customer skin, and a body odor sensor to analyze body smell. The power module operates the IoT device using an external power source, and it is equipped with a lithium-ion battery. The communication module includes wired and wireless equipment for Internet-based communication and the server in a distant place through Wi-Fi. The sensor module transmits all collected data to the management server through the communication module. Both modules are connected to the MCU via the universal asynchronous receiver transmitter method. For this purpose, a protocol was defined between the sensor and communication modules.

Figure 2. Internet of things (IoT) device chart to collect body odor of customer. UART, universal asynchronous receiver transmitter; MCU, microcontroller unit.

Microcontroller unit

All collected body odor, temperature, and humidity information was sent to the MCU through the communication module. The MCU is equipped with a Linux operating system to process and manage the collected information, and an MySQL database to store it. The database for managing customer information is presented in Table 2.

Table 2 . Sensing data table.

Item	No	Customer ID	Time	Temp	Humid	Odor
Data type	Integer	Text	Date time	Integer	Integer	Float

Smart devices

Customer information collected from IoT devices is stored in a database installed on a management server. The following section describes smart devices that display personalized information to customers. In general, a tablet PC equipped with an Android operating system (iOS) is used as a smart device. In the present study, an app providing customized perfume recommendations was developed for smart devices using Android Studio. An example screen to be displayed on a device is shown in Fig. 3, where 1) displays the customer’s skin photograph, 2) displays environmental temperature, 3) displays environmental humidity, and 4) displays a recommended perfume product.

Figure 3. The composition of smart device screen 1) area contains skin photograph of customer; 2) area contains temperature of the place; 3) area does humidity of the place; 4) area shows the information on perfume products suitable to the customer based on the values related with body odor of the customer.

Customized perfume recommendation system

For a customer who visits the store, the IoT device described in chapter 3 collects body odor, environmental temperature, and environmental humidity data. This information is sent to the MCU through the communication module; subsequently, sensing values collected from the sensor module can be identified on a smartphone using the JavaScript object notation (JSON). Table 3 classifies perfume products into three levels based on the aesthetic quality of aroma. Fig. 4 illustrates algorithms to provide customers with customized perfume recommendations based on data relating to body smell-inducing materials.

Figure 4. Algorithm to provide customers with customized perfume products.

Table 3 . Esthetic classification of aroma.

Level	Affiliation
Low	Aqua, Citrus, Floral, Fruity, Green
Middle	Chypre, Gourmand, Woody
High	Aldehyde, Musk, Oriental, Spicy

All data collected by the IoT device are sent to the management server for storage and management. Subsequently, smart devices can be used to obtain appropriate recommendations of perfume products for customers.

In response to the increasing development of smart beauty devices combining IT with the beauty industry, this study developed a device that recommends appropriate perfume products to customers by analyzing body odors using IoT-based technologies. It provides customers with products that can maximize emotional satisfaction by stimulating the sense of human smell.

The characteristics of consumer products have changed with technological advancement. In the past, people primarily consumed the necessities of life. Subsequently, people began to procure new products to keep up with emerging trends. With the Fourth Industrial Revolution, people have begun to use private services that offer personalized products. Body odors are secreted by a wide range of factors, giving everyone a unique body smell. People generally select perfume products that mask their body odors and represent their desired public image.

Under consideration of these factors, the combination of a perfume’s aroma and body odor can generate an unintended smell. Therefore, a device that analyzes information regarding the proportions of the customer’s body odor components, and recommends appropriate products, may increase customer satisfaction. Accordingly, the customized perfume service device developed in this study is expected to play a leading role in the growth of the smart beauty and smart perfume industries as high value-added markets in the future.

This work was supported by a Pai Chai University research grant from 2022.