Modeling and Analysis of Intrabody Communication for Biometric Identity in Wireless Body Area Networks

Abstract

Intrabody communication (IBC) establishes a wireless connection between devices in a Wireless Body Area Network (WBAN) by utilizing the human body as a transmission medium. The characteristics of the IBC channel are significantly influenced by the geometric and biological features of the human body and tissues. This paper analyzes a dataset with experimental real subjects’ data on signal loss in a galvanic IBC channel, models IBC identification using the K-Nearest Neighbors (KNN) algorithm, and proposes a novel IBC WBAN architecture incorporating an identification function. The analysis of the dataset revealed that the IBC channel gain exhibits a wide range of variations depending on individual human body characteristics such as height, weight, body mass index, and body composition. Consequently, biometric identification can be leveraged within the IBC WBAN paradigm. Through modeling IBC identification on cleaned and labeled data, we demonstrated an identification accuracy of 99.9% based on the results of our modeling. The proposed IBC WBAN architecture with an integrated identification function is anticipated to enhance the application scope and accelerate the development of IBC WBANs.

1. Introduction and Motivation

Wireless Body Area Networks (WBANs) represent a class of wireless networks that facilitate data exchange between mobile devices located on or within the human body as part of the Narrowband Internet of Things (NB-IoT) group of devices. In recent years, WBANs have gained widespread adoption due to the diminutive sizes and weights of their constituent devices. WBANs typically comprise transceivers and sensors that communicate wirelessly. Presently, WBANs predominantly employ radio frequency (RF) communication technologies such as ZigBee, Bluetooth, and 6LoWPAN. However, these RF transceivers suffer from drawbacks, including high power consumption and mutual interference [1,2,3], thereby limiting their applicability. Intrabody communication (IBC) technology has emerged to mitigate these issues and reduce power consumption [4,5].

Intrabody communication leverages the human body as a conductive medium for wireless communication between WBAN devices. IBC systems find applications in diverse domains, such as healthcare, sports, military, and security. Researchers at the MIT Media Laboratory have introduced a wearable monitor for temperature measurement and hypothermia detection in soldiers [6]. Additionally, smart textiles equipped with acceleration sensors and leveraging IBC have been developed for applications in sports and medical rehabilitation [7,8,9,10,11]. IBC has also been explored for enhancing the reliability and security of access control systems [12,13,14,15]. Furthermore, inter-segment communication enables the establishment of connections between IBC-enabled WBAN devices located on different individuals [16].

The development of Industry 4.0 also drives the emergence of new requirements for organizing interactions between different participants in the manufacturing process: autonomous robotic systems and collaborative robots (cobots), humans and the production system. Researchers note that evolving production tasks and the adoption of technologies such as the Internet of Things (IoT) and artificial intelligence (AI) are reshaping the role of humans in manufacturing. One prominent approach is the Operator 4.0 concept, first introduced in [17] as a “human-centric” vision for the successful adoption and implementation of Industry 4.0 technologies in smart factories. A recent analysis of the evolution of this approach in [18] demonstrated that the nature of work, the social and organizational work environment, and related individual factors are key categories influencing the psychosocial state of Operator 4.0.

The task of identifying individual workers at a production site within a manufacturing facility enables the consideration of human-specific characteristics for fine-tuning equipment or specialized robotic systems (for instance, industrial exoskeleton) that collaborate with humans on production lines.

Research has shown that IBC can address some of the prevalent challenges associated with traditional communication methods used in industrial settings, such as high energy consumption and susceptibility to interference. The energy efficiency of IBC aligns well with the goals of Industry 4.0, where the emphasis is on creating sustainable and low-energy solutions. For instance, IBC-enabled wearable devices can facilitate real-time health monitoring among factory workers, ensuring both safety and productivity while simultaneously reducing operational costs linked with energy consumption [5,12]

This necessity highlights the importance of exploring solutions such as Intrabody communication for biometric identification. Intrabody communication facilitates wireless connections within wireless body area networks (WBANs) through the human body acting as a transmission medium. However, the efficacy of IBC is significantly affected by the intricate geometric and biological variances among individuals, which contributes to substantial signal loss and variations in channel gain. These challenges hinder the reliability of communication between devices in WBANs, thereby limiting the potential applications of IBC technology.

Numerical biological tissue modeling plays a pivotal role in the development and optimization of intrabody communication devices. Modeling is widely used to determine the characteristics and parameters of body area networks. IBC modeling is a complex task because of the propagation of the signal through tissues with varying dielectric properties. Modeling plays a crucial role in IBC design, enabling the optimization of wearable devices to achieve maximum reliability and data transfer rates while minimizing power consumption. The goal of IBC modeling is to assess the functional dependencies between the parameters of a real system during operation, such as the relationship between signal quality and the properties of external and internal tissues, the distance between transceivers, and the size and shape of the tissue channel [19]. However, experimental studies are necessary to validate these simulation results. The use of physical tissue models or phantoms allows repeatable measurements under controlled conditions, significantly simplifying the verification process. In [20], the authors measured the characteristics of the IBC channel in five subjects and two phantoms to investigate the feasibility of using these characteristics for biometric identification. This study confirmed the possibility of biometric identification with an accuracy of 98.5% using machine learning algorithms for data analysis.

This paper models the feasibility and effectiveness of using IBC channel characteristics for biometric identity, which further can be used in industrial applications based on the last known dataset containing experimentally measured signal loss data in IBC channels [21]. The K-Nearest Neighbors (KNN) algorithm was used to assess the feasibility of identifying and recognizing individual subjects in the dataset based on features extracted from their IBC channel characteristics.

The main contributions of our study are the following:

• Our research presents a study that integrates biometric identification techniques into the IBC system, utilizing channel gain influenced by individual physical characteristics (such as height, weight, and body composition) to enhance user identification and authentication from an experimental dataset containing data about real subjects. By using the KNN algorithm, we achieved an accuracy of 99.9%, demonstrating the potential for IBC to serve as reliable biometric identifiers.
• We proposed a novel IBC WBAN architecture that incorporates biometric identification functionalities, enhancing the performance and security of existing systems. This architectural advancement provides a framework for the development of future IBC-enabled devices and applications focused on personalized user experiences.
• By exploring the synergy between IBC technologies and biometric identification, our study addresses significant challenges in the realms of security and energy efficiency relevant to Industry 4.0. Our findings suggest pathways for the deployment of energy-efficient, secure WBANs within industrial settings, highlighting opportunities for health monitoring and workforce management.
• This analysis lays the groundwork for future studies assessing the applicability of IBC in diverse contexts.

This paper is organized as follows. Section 2 covers related work in IBC. Section 3 provides a brief overview of types, identification, and measurement of signal loss in IBC channels. Section 4 is devoted to the analysis of the dataset and description of the influence of individual human body characteristics on IBC channel gain. Section 5 presents a biometric identity modeling based on IBC channel gain data from the dataset using the KNN algorithm. Section 6 is devoted to the architecture for an IBC system with biometric identity. Section 7 is focused on discussion of the results and future studies. Finally, Section 8 concludes the paper.

2. Related Work

The concept of leveraging the human body as a communication medium was pioneered by Zimmerman [4], who demonstrated the feasibility of capacitive coupling for low-power data transmission through the body. Early work focused on basic signal transmission, but recent studies have explored IBC’s potential for biometric identification [22]. In [23], several parameters of IBC were measured using body channel response (BCR) as a unique identifier, noting that variations in body composition (e.g., fat-to-muscle ratio, hydration levels) induce distinct signal attenuation patterns. But practical implementations were limited by simplistic feature extraction methods.

Early studies like [22] modeled IBC intra-palm propagation signals as a function of frequency (10–100 MHz) and electrode placement, identifying optimal operating frequencies for minimal loss. Recent research has linked channel gain dynamics to biometric identity. Advancements in machine learning have enabled more sophisticated biometric frameworks. In [20], machine learning algorithms were introduced to identify users based on IBC channel characteristics, achieving 98,5% accuracy using the Naive Bayes algorithm. K-nearest neighbors yielded the second-best results, with identification accuracy of 97.8%. It provides good results with the simplest model and least computational power budget. Their dataset, however, relied on controlled laboratory settings, limiting generalizability. Research in the field of Intrabody Communication is often focused on using data obtained from artificial phantoms [20,24,25,26,27] rather than actual human subjects. This limits researchers’ ability to account for individual biometric characteristics and geometric variations that can significantly influence the performance of IBC channels. The use of phantoms may not fully capture the real-world conditions of communication, highlighting the need for more comprehensive studies based on data from real users to ensure the reliability and accuracy of IBC technologies. Despite progress, these studies largely overlooked the impact of dynamic channel conditions (e.g., motion artifacts, environmental interference) on biometric reliability.

Comparative studies have also explored modulation schemes for IBC-based biometrics. For example, in [28], the combination of Ultra-Wide-Band IBC showed that offers superior resilience to motion artifacts but at the cost of higher power consumption. These findings underscore the trade-offs between biometric accuracy and energy efficiency in WBAN design. The absence of standardized IBC datasets has hindered reproducibility in biometric research.

WBAN architectures for IBC biometrics must reconcile security, latency, and energy constraints. The effectiveness and increased demand of WBAN forced the development of communication standard IEEE 802.15.6 [29] to bring low-power sensor devices and protocols under one umbrella for seamless communication and to expand the range of applications. In [30], it was proposed a tiered WBAN framework with centralized hubs for data aggregation, but their design prioritized throughput over security. However, their work did not address biometric-specific challenges like template protection.

The integration of Intrabody Communication in Wireless Body Area Networks is gaining traction, particularly in the context of Industry 4.0, where the convergence of physical and digital systems is creating opportunities for innovative solutions across various domains. IBC utilizes the human body as an effective transmission medium, enabling seamless and secure communication between wearable devices utilized in industrial environments. The existing literature on IBC often neglects a comprehensive examination of how individual biometric characteristics—such as height, weight, body mass index (BMI), and body composition—impact IBC channel characteristics. Additionally, while numerous studies have explored the advantages of IBC over traditional radio frequency (RF) communication, very few have proposed robust methodologies for integrating biometric identification into IBC frameworks. This gap in the research highlights the need for a systematic approach to utilize IBC for reliable user identification within WBANs.

3. Materials and Methods

This section begins by analyzing two types of IBC and then discusses research related to measuring IBC parameters and their use for user identification.

3.1. IBC

Currently, two IBC techniques are known: capacitive and galvanic, which were analyzed in [31]. Capacitive coupling does not depend on external wires, but the quality of transmission is influenced by the environment. For capacitive coupling, a ground connection between the input and output circuits is necessary. The signal is transmitted between the body channel of the transceivers via a current loop, which consists of a transmitter electrode, body channel, receiver electrode, and a capacitive return path through the external ground.

Galvanic coupling operates without a connection to the external ground. High-frequency electromagnetic waves are generated at the input. They propagate throughout the body and reach the receiver electrodes. This method does not require external wires, and the transmission quality does not depend on the environment. Two transmitting electrodes generate signals, and the body is used as a waveguide for signal propagation. Then, the signal is received by two receiver electrodes.

The main difference between these two techniques is that the IBC channel characteristics in capacitive coupling are highly dependent on the environment around the body, whereas in galvanic coupling, they depend on the physical characteristics of the human body.

IBC data transmission offers the following advantages:

• Security: IBC provides a secure and private communication network, offering natural security and interference-resistant communication [32]. IBC requires a significantly lower operating frequency compared to RF systems. This means that signals are limited to human proximity, so data reading requires body contact [33].
• Energy efficiency: research has shown that IBC consumes an order of magnitude less energy, at data rates up to 10 Mbps, making it an attractive communication technique for WBAN applications [24,34].
• Frequency reuse: IBC forms a near-field communication network inside and around the human body. Consequently, it allows for the reuse of the same frequency band by other WBAN users.

3.2. IBC Identification

In [20], a study was conducted to measure the signal transmitted characteristics through galvanic coupling. The signal includes information about the transmission channel (channel gain profile). This information is primarily determined by the channel gain/attenuation profile, which develops due to the unique body biological and geometric characteristics [25]. Conducted measurement used a miniVNA Pro vector network analyzer on five subjects and two phantoms. The phantoms represent a five-layer model constructed using materials that accurately mimic the tissues dielectric properties (dielectric permittivity and conductivity), as shown in [26]. Over 150 measurements were conducted at distances between the TX and RX of 10, 15, and 20 cm. The possibility of using the channel gain profile in the galvanic link frequency range to obtain unique characteristics defining the biometric identity of each individual was investigated.

To identify a subject, the received signal is compared to previously recorded identifiers (unique to that subject) using machine learning algorithms. In [20], the following machine learning algorithms were used: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), J48, Random Forest, and Naive Bayes classifier. These algorithms were used to train and test models capable of identifying individuals using features extracted from IBC channel characteristics obtained from the measurements. After training the model on IBC channel features, the model was tested on a test set of instances, where the model must accurately and reliably identify the subject.

In [20], a biometric classifier is used to determine the user measurement sample. IBC channel features were selected in order to not make the model more complicated, but to maintain accuracy and performance. The IBC channel gain was used as a feature. An identification accuracy ranging from 92% to 100% was observed when measurements were conducted at a distance of 10 cm. However, when using data for larger distances (15 and 20 cm), accuracy decreased as signal attenuation increased with increasing distance. The study concluded that KNN provides high accuracy and resilience against noise, which can make it particularly suitable for real-time applications. The simplicity of KNN also contributes to reduced computational complexity compared to more involved algorithms like Support Vector Machine or Random Forest, which may require more extensive tuning and computational resources. Overall, while several algorithms were tested, KNN’s optimal balance of accuracy and computational efficiency positions it as a highly suitable choice for biometric identity verification.

Therefore, the KNN algorithm was selected for modeling in our research.

3.3. Signal Loss Measurement in IBC Channel

In [35], the authors presented a dataset on signal loss in galvanic-coupled IBC channels [21], containing over 17,000 measurements. The data were measured on 30 experimental participants (subjects) in a frequency range from 50 kHz to 20 MHz. The signal was transmitted from a signal generator through the human body using galvanic coupling to a receiver. Stable skin–electrode contact was achieved by applying ECG gel to the skin under the electrode contact plates. One oscilloscope measured the signal on the transmitter side, and the second on the receiver side. The IBC channel gain was calculated based on the measurement data.

The dataset contains the following data: participant identifier; experiment number; participant’s height and weight, body mass index (BMI), percentage of body fat, age, and gender (male or female); transmitter point number; receiver point number; distance between transmitter and receiver; average fat content at the transmission point; average fat content at the reception point; sum of fat levels at the transmission and reception points; measurement frequency; receiver gain using a 50 Ω load resistance; and receiver gain using a 1 mΩ load resistance. Currently, study [35] is the most extensive experiment on measuring IBC characteristics; therefore, in our study, we used that dataset.

4. Data Analysis and IBC Channel Gain Characteristics

For data analysis, we used the dataset “all_measurements_by_freq_filtered.csv” from [21], as it contains key results for a single frequency in each row, and noise was cleaned from it. The dataset consists of data in the frequency range of 50 kHz to 20 MHz. Figure 1 shows the dependence of IBC channel gain on the signal transmission frequency. Based on the analysis, we identified the most suitable frequency for signal transmission at 12.5 MHz.

We analyzed only data measured at a frequency of 12.5 MHz from the first experiment. All measurements were carried out under the same environmental conditions and at the same time, and they also contain the largest number of measurements compared to other experiments. The frequency of 12.5 MHz has the lowest signal attenuation (Figure 1); therefore, that frequency is the most suitable for signal transmission. We analyzed 616 measurements in our study.

In Figure 2, the dependence of the receiver gain on the individual subject body parameters is presented. The gain of the receiver for a 50 Ω load is in the range −38.8 to −46.0 dB, and for a 1 mΩ load −31.2 to −37.4 dB. Based on the dataset analysis, we concluded that the IBC channel gain depends on the subject’s individual characteristics. As shown in Figure 3, the receiver gain depends on the subject’s height and is in the range −29.0 to −46.0 dB for a 50 Ω load and −31.8 to −37.4 dB for a 1 mΩ load. Figure 4 shows the dependence of the receiver gain on the subject’s weight. The gain of the receiver for a 50 Ω load is in the range −23.6 to −46.0 dB, and for a 1 mΩ load −31.2 to −37.4 dB. As shown in Figure 5, the receiver gain depends on the subject’s body mass index and is in the range −21.2 to −44.9 dB for a 50 Ω load and −32.5 to −35.8 dB for a 1 mΩ load. Figure 6 shows the dependence of the receiver gain on the subject’s body fat composition. The gain of the receiver for a 50 Ω load is in the range −19.6 to −45.2 dB, and for a 1 mΩ load −31.8 to −36.9 dB. As shown in Figure 7, the receiver gain depends on the sum of the fat levels at the transmit and receive points and ranges from −25.6 to −39.6 dB for a 50 Ω load and −30.9 to −36.5 dB for a 1 mΩ load.

As shown in Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7, the receiver gain varies over a wide range depending on the subject’s body characteristics; therefore, the IBC channel gain is influenced by height, weight, body mass index, and body fat composition. These indicators form a channel gain profile that can be used to identify IBC system users. In case of any changes in the human body, those changes are recorded in the database, and further, the accumulated changes are analyzed by the system in order to predict possible changes in the future. One of the major findings is that it is possible to arrange biometric identity using the individual human body characteristics in IBC.

5. Modeling of IBC Identification

This section describes the modeling of IBC identification conducted in the Jupyter Notebook software 7.0.8 [36] using the NumPy library [37] and the KNN algorithm. Jupyter Notebook is an interactive browser-based environment that includes a convenient computer navigator and consists of fields that are filled with text information or Python 3.11.7 code. Jupyter Notebook is used for data analysis. NumPy is an open-source tool designed for array computations. KNN is a classification algorithm based on the assumption that objects located close to each other in the feature space have similar values of the target variable or belong to the same class.

There are three main metrics for calculating the distance between objects: Euclidean distance, Manhattan distance, or other distance metric.

After removing unnecessary data from “all_measurements_by_freq_filtered dataset” from [21], we obtained modeling

Table 1. Modeling table (out of 616 rows).

Subject_id	Distance	Rx_gain_50	Rx_gain_1M
1256	20.6	−47.771266	−39.482652
1256	16.3	−44.810831	−36.444843
1256	13.6	−44.475018	−35.952138
1256	16.3	−48.714588	−40.176981
1256	14.9	−44.007403	−35.684326
1256	12.9	−42.942264	−34.625295
1256	8.6	−37.340486	−29.146415
1256	8	−38.88304	−30.088628
1256	6.7	−38.236727	−29.62718
1256	10.25	−43.56398	−34.519372
1256	5.5	−37.712044	−29.149732
1256	18.6	−44.371963	−35.712716
1256	13	−42.146326	−33.774958
1256	10.3	−41.808655	−33.768054
1256	4.15	−47.139468	−39.070489
1256	11.6	−41.295583	−33.238293
1686	16.3	−43.848785	−35.944426
1686	13.6	−43.064913	−35.35695
1686	16.3	−45.95929	−37.969667
…	…	…	…

. The table only left the data of the first experiment and a frequency of 12.5 MHz. The table contains data, from 616 measurements made on 30 subjects, and several columns:

• subject_id—unique subject identifiers.
• distance—distance between the transmission and reception points.
• rx_gain_50—receiver gain using a 50 Ω load resistance.
• rx_gain_1M—receiver gain using a 1 mΩ load resistance.

This table was loaded into NumPy (Figure 8) and divided into a measurement matrix and a subject vector. Figure 9 shows the measurement matrix, which has a size of 616 rows and three columns: distance, rx_gain_50, and rx_gain_1M. Figure 10 shows the subject vector containing the subject identifiers from the subject_id column.

Ten values were randomly selected from the measurement matrix and subject vector for modeling purposes. Selected values from the measurement matrix, selected values from the subject vector, and selected values from the overall matrix were loaded into NumPy.

The Euclidean distance was calculated between each vector from the matrix of selected values and each vector from the measurement matrix using the KNN algorithm. The KNN algorithm is defined as:

$$L\left(p,q\right)=\sqrt{{\sum }_{i=1}^n{\left(q_i-p_i\right)}^2},$$

(1)

where q_i is a feature of the measurement matrix, p_i is a feature of the random value matrix, and n is the number of features in the dataset. Listing 1 shows a function that returns a distance matrix from each vector of the random value matrix to each vector of the measurement matrix, performing a single loop over the random value matrix. In Listing 1, the measurement matrix is denoted as M, and the random value matrix is denoted as M_random. The distances calculated using the KNN algorithm were recorded in a distance matrix.

Listing 1. Python code to calculate distance matrix.

def dist_one_loop(M_random):           dist = np.zeros ( (M_random.shape[0],M.shape[0]) )           for i, m_random in enumerate(M_random):                    m_len = M – m_random                    dist[i, :] = np.sqrt( (m_len ** 2).sum(axis=1) )           return dist dist = dist_one_loop(M_random) print(dist)

Then, we formed an array of class labels for a given sample of vectors. The subject vector was used to obtain the labels.

Ten modeling attempts were conducted, with 10 random values selected in each. As a modeling result, 100 out of 100 values were correctly identified; therefore, the achieved identification accuracy was 99.9%. The modeling was carried out on cleaned and labeled data, so this is an ideal case. In a practical case, the identification accuracy can be as low as 98.5%, as indicated in [20].

It is also necessary to mention that, based on regression analysis, the channel amplification is influenced by the characteristics of the human body when using a load resistance of 50 Ω (rx_gain_50) (Figure 11). In contrast, when a load resistance of 1 m Ω is employed, the characteristics of the human body have a negligible effect on channel amplification (Figure 12).

6. IBC WBAN Architecture with Identification Functionality

Biometric identity is a security method based on the technology of recognizing a person by their unique biological characteristics. A person’s biometric data are uploaded to a database and serve as a pattern. During biometric identity, the user’s biometric data are compared to the pattern stored in the database. If the data match, then biometric identity is considered successful.

We proposed energy-efficient IBC WBAN architecture in [38]. In our current study, we propose to expand the IBC WBAN functionality with user biometric identity. Figure 13 shows the proposed IBC system architecture with biometric identity.

The system operates as follows:

1.

Implantable Devices and Signal Propagation

Miniaturized implants embedded within the body generate low-power electrical signals using galvanic coupling, where the human body acts as a conductive medium. Signal propagation is modulated by the individual’s unique biological parameters, including body composition, geometric dimensions [25], and tissue dielectric properties. These characteristics impart distinct channel gain and attenuation profiles to the transmitted signals.

2.

IBC Transceiver

The IBC transceiver employs separate coupling to establish dual communication pathways: one for signal transmission through the body and another for environmental coupling. This design minimizes cross-interference between internal and external communication channels. Simultaneously, it generates coupling to stabilize signal transmission across dynamic physiological conditions (e.g., movement, hydration changes).

3.

Mobile Transceiver and Biometric Extraction

A wearable or handheld mobile transceiver interface with the IBC transceiver to relay data to external networks. Integrated within this unit is a mobile identification module, which extracts real-time IBC channel characteristics, such as:

○

Receiver gain.

○

Frequency-dependent attenuation profiles.

○

Transmitter–receiver separation distance. (The inclusion of distance as a feature mitigates false identification caused by overlapping gain values at varying transmission distances.)

4.

Biometric Identification Workflow

The extracted channel characteristics are compared against pre-registered biometric templates stored in a secure database. A machine learning classifier (e.g., K-Nearest Neighbors) analyzes the feature space to authenticate users. Matching thresholds are dynamically adjusted to accommodate minor physiological variations (e.g., weight fluctuations), ensuring robustness without compromising security.

5.

Security and Scalability

The architecture leverages the inherent physical boundaries of the human body to restrict signal propagation, reducing eavesdropping risks. Biometric templates are encrypted and stored in a decentralized database, enabling scalable deployment across large populations. The system supports real-time updates to biometric profiles, ensuring adaptability to long-term physiological changes.

Technical advantages of the proposed approach consist of the following:

• Energy Efficiency: galvanic coupling reduces power consumption by one-to-two orders of magnitude compared to RF-based systems.
• High Accuracy: biometric identification achieves 99.9% accuracy in controlled environments, driven by unique channel gain profiles.
• Interference Resilience: separate coupling pathways isolate intra-body communication from external electromagnetic noise.

This architecture pioneers the use of IBC channel properties as biometric identifiers, enabling secure, personalized applications in healthcare monitoring, industrial wearables, and IoT access control.

7. Discussion

In the context of WBANs, which facilitate communication between devices located on or within the human body, the challenges associated with traditional radio frequency (RF) communication methods—such as high power consumption and susceptibility to interference—can be effectively mitigated through the use of Intrabody Communication [1,2,3]. IBC utilizes the human body as a conductive medium, allowing for more reliable data transmission while significantly reducing power requirements. This characteristic aligns well with the goals of NB-IoT, which seeks to enable long-range communication with low energy consumption. Moreover, the incorporation of biometric identity into IBC-enabled WBANs enhances both security and personalization [26,39,40,41]. By leveraging the unique biological and geometric characteristics of individuals, such as height, weight, and body composition, the IBC channel gain can serve as a biometric identifier. This capability not only improves the accuracy of user identification but also provides new opportunities for industrial applications in various fields, including healthcare and security.

The results of our research indicate that the identification accuracy achieved through the KNN algorithm, based on IBC channel characteristics, reached 99.9% as a modeling result. Usage of the KNN algorithm has advantages like simplicity, no training phase, and effective for small datasets, as well as limitations like computationally intensive for large datasets and sensitivity to irrelevant features. Further application of the K-Dimensional Tree (also known as KD Tree) might influence the accuracy and processing power requirements. It is necessary to mention the dataset that was used was the most advanced for the time modeling was provided. Diversification and further development of the experimental dataset are suggestions for future research, as these limit the study, where we have only 30 subjects being recorded and their dataset being analyzed. The high level of accuracy underscores the potential for biometric identification to be seamlessly integrated into IBC systems, thereby enhancing the overall functionality of WBANs. As the technology continues to evolve, the combination of NB-IoT, IBC, LTE-M, and biometric identity could revolutionize the way we approach health monitoring and personal security, making these systems more efficient, reliable, and user-centric. In the future, it is planned to investigate how body movements will affect the channel gain profile and, therefore, the functionality of a real biometric identification system.

In conclusion, the synergy between NB-IoT and WBANs (with the use of LTE-M), coupled with the innovative use of biometric identity, presents a transformative opportunity in the field of wearable technology. This integration not only addresses existing challenges in communication and power consumption but also enhances the security and personalization of health-monitoring systems, paving the way for broader adoption and application in various domains. As research and development in this area progress, we anticipate significant advancements that will further expand the capabilities and applications of IBC-enabled WBANs.

8. Concluding Remarks

In this paper, we analyzed an existing experimental dataset on galvanic-coupled IBC signal loss. Based on the data analysis, we concluded that the IBC channel gain varies significantly depending on the individual body characteristics; hence, the IBC channel gain is influenced by height, weight, body mass index, and body composition in terms of fat. It is possible to use biometric identity based on the individual body characteristics for IBC systems and WBANs.

Furthermore, IBC biometric identity modeling was conducted to test the KNN algorithm applicability for IBC identification and to determine the algorithm identification accuracy. Identification accuracy of 99.9% was achieved as the modeling result on cleaned and labeled data.

This research proposed an IBC WBAN architecture with a user biometric identity to expand its functionality. The possibility of using IBC channel characteristics as a biometric identifier will expand the application areas of IBC systems and accelerate their development.