An Efficient Communication Protocol for Real-Time Body Sensor Data Acquisition and Feedback in Interactive Wearable Systems

Abstract

We introduce a novel wired communication approach for interactive wearable systems, employing a single signal wire and innovative group addressing protocol to reduce overhead. While wireless solutions dominate body sensor networks, wired approaches offer advantages for interactive applications that require low latency, high reliability, and communication with high-density nodes; yet they have been less explored in the context of wearable systems. Many commercial products use wired connections without disclosing technical details, limiting broader adoption. To address this gap, we present and test a new group addressing protocol implemented using Universal Asynchronous Receiver–Transmitter (UART) hardware, disclosing frame diagrams and node architectures. We developed a prototype interactive jacket with nine sensor/actuator nodes connected via three wires for power supply and data transmission to a wireless gateway. Mathematical analysis showed an overhead reduction of approximately 50% compared to traditional individual addressing. Our solution is the most wire-efficient among wired interactive wearable systems reviewed in the literature, using only one signal wire; other methods require at least two wires and often have overlapping topologies. Performance experimental evaluation revealed a total feedback delay of 2.27 ms and a maximum data frame rate of 435.4 Hz, comparable to the best-performing products and leaving room for twice the performance calculated theoretically. These results indicate that the proposed approach is suitable for interactive wearable systems, both for real-time applications and high-resolution data acquisition.

1. Introduction

Interactive wearable systems that integrate sensors with real-time feedback are revolutionizing industries such as occupational health, rehabilitation, sports, and entertainment. These systems have proven useful for vital sign monitoring [1,2,3,4,5], physical activity tracking [6,7,8], sports performance analysis [9,10,11,12], virtual reality [13,14], and more. As interest grows in applications requiring not only sensing but also real-time feedback, for example, to optimize training, alert users to potential hazards, and prevent injuries, the demands on communication networks increase. While these advancements offer significant benefits to individuals, organizations, and society, existing communication approaches for the body sensor networks (BSNs) struggle to support the requirements of advanced real-time interactive applications, particularly those involving networks with large numbers of nodes distributed across the body.

Despite the reliance on wireless communication technologies in BSNs, which is the foundation of interactive wearable systems, wireless solutions pose challenges in the dense and complex topologies required for advanced applications. Issues related to cost, complexity, and interference become significant barriers [15]. Wired communication offers a promising solution to these problems, especially when combined with wireless options to maintain wearability and freedom of movement. However, wired networks for wearable systems remain largely unexplored in the current literature. Communication protocols used in high-performance products (see Table 1) are not disclosed for broader adoption, and those used in research prototypes struggle to achieve comparable performance. Furthermore, there is a lack of specialized approaches developed for wearable applications that consider specific needs such as efficient communication with reduced overhead and wire usage.

The aim of this study is to advance communication solutions for interactive wearable systems by introducing a high-performance, efficient, and reliable communication method to connect numerous sensor and actuator nodes distributed across the body. Specifically, we present an innovative group-based communication protocol that reduces overhead and can be adapted for various well-established semi-duplex communication interfaces. Although the protocol can be applied to several low-power MCU interfaces (e.g., Serial Peripheral Interface (SPI), inter-integrated circuit protocol (I2C), and UART), we chose a UART semi-duplex configuration with a single data wire to minimize cabling—an approach, to our knowledge, not previously demonstrated in multi-sensor wearable systems. Using mathematical analysis, we evaluated the overhead reduction ratio, data and action delays, goodput, and maximum frame rate of the proposed approach.

In a laboratory setup, we integrated nine sensor/actuator nodes into an interactive jacket prototype, connected via a three-wire bus that provides both power supply and bidirectional communication in a bus topology. Such an approach, not previously demonstrated in wearable multisensor systems, allowed us to measure total data and feedback delays, as well as the maximum reliable frame rate. The results showed that the proposed approach is suitable for interactive wearable systems, supporting both real-time interactive applications and high-resolution data acquisition with reduced wiring compared to state-of-the-art multi-sensor wearable systems. Moreover, it opens new possibilities for developing high-performance, reliable, and scalable interactive wearable systems across a broad range of applications.

The rest of this article is structured as follows: Section 2 provides a literature review and background on communication technologies used in BSNs. Section 3 details the concept of the proposed communication protocol for wired on-body communication between groups of sensor and actuator nodes in interactive wearable systems, along with considerations for its technical implementation. Section 4 details the adaptation of the proposed wired communication and group addressing concepts to develop an interactive motion-tracking jacket with vibrating feedback, including its hardware implementation and experimental evaluation. Finally, Section 5 concludes the paper with a discussion of the findings and their implications.

2. Literature Review

Extensive and diverse research on wireless communication technologies for wearable devices and BSNs has been conducted, as highlighted in the survey by [15]. Among these efforts, Wireless Body Area Networks were standardized in the IEEE 802.15.6 standard [16], released in 2012 to address a wide range of requirements for various wearable sensor applications. However, on 30 March 2023, the standard was declared Inactive-Reserved due to it not undergoing the required revision process within the 10-year period. This status change reflects the evolving landscape of wireless technologies, with the wearable device industry increasingly adopting more versatile standards like Bluetooth, Bluetooth Low Energy (BLE), Zigbee, and Wi-Fi, which offer greater integration with mobile devices and broader ecosystem support.

While Bluetooth and Wi-Fi are widely adopted, they still face issues with network congestion and interference in high-density environments [17], limiting their effectiveness for critical applications. Also, several other challenges remain unresolved for body-worn devices and BSNs, including network security for low-cost wearable nodes and efficient power management. These issues continue to be critical in enabling the full potential of BSNs across various fields such as health monitoring, assisted living, sports, and military activities [18,19].

While research has focused extensively on wireless communication technologies for wearable devices and BSNs, wired communication has received far less attention due to concerns about mobility and user comfort. However, recent advances in flexible sensor and conductor materials as well as fabrication techniques for wearable electronics could make wired solutions more viable [20,21,22,23], particularly for interactive applications that require low-latency, high-reliability communication with a large number of wearable nodes, such as detailed shape reconstruction [24] or motion capture for aquatic exercises [25], where wireless systems often struggle to meet performance demands.

In research and wearable system prototyping, the I2C protocol is predominantly used for creating wired networks due to its modularity and strong support in integrated circuits. For example, ref. [25] describes a suit for aquatic exercises that uses the I2C protocol to communicate with inertial sensor nodes via wired connections. I2C was chosen over SPI because it requires less wiring and is already supported by the integrated circuits of the sensors, reducing the need for additional microcontroller unit (MCU) on the sensor boards. However, the need to overcome the limitation of usable sensor I2C addresses led to the use of a multibus topology, where the gateway is multiplexed between multiple I2C buses. This resulted in a poorly scalable solution that does not utilize wiring efficiently. Moreover, the performance of the system is significantly lower compared to commercial products, with only five sensor nodes sampled at a rate of 20 Hz.

Similarly, another study employing a multiplexed I2C architecture developed the iFeel suit for musculoskeletal disorder prevention [26]. While this study achieved better results—sampling 10 IMU sensor nodes at a rate of 80 Hz—it remains inefficient in terms of wire usage and performs considerably poorly than commercial devices with claimed sample rates up to 400 Hz (Table 1). To address wiring complexity, ref. [27] proposed and utilized a new cascaded I2C approach, which, in contrast to the star-based networking approach, reduces wiring complexity. The experimental setup for evaluating strength exercises used five nodes and sampled them at a rate of 100 Hz, which is the highest among systems in the reviewed literature using I2C, however, still considerably less than commercial devices. Also, the number of sensors on a single I2C bus for this solution was limited to eight, and to mitigate the effects of crosstalk and the high capacitance of the I2C bus, the data lines were distributed over multiple 6-wire cables, each accommodating more than one I2C connection. This approach resulted in increased use of wires and reduced material efficiency.

For applications requiring a larger number of sensor nodes, such as detailed shape sensing, the authors in [28] proposed a modified SPI daisy chain configuration for efficient real-time sensor data acquisition using two signal wires. The authors estimate that this approach is suitable for centralized data aggregation from inertial sensor arrays with over 200 nodes, operating at a rate of 50 Hz. In [24], this approach was practically tested in the development of a shape-sensing fabric that connected up to 63 nodes and sampled at a 10 Hz rate. However, the protocol, utilizing only two signal wires, is simplex; to enable two-way communication, an extra wire would be required.

Furthermore, their approach is restricted to a line topology, which limits the physical routing of wires to connect all sensor nodes, resulting in redundant wire usage. This issue can be mitigated by adapting the protocol for a multi-line topology, as demonstrated in [7] for a full-body motion tracking suit. However, as shown in the prototype pictures in [7], even in this configuration, the overlapping of wired connection paths was not entirely eliminated.

In Table 1, we summarize the sensors, BSN architectures, feedback mechanisms, latencies, and data update rates offered by multi-sensor wearable prototypes and commercial products that represent the state of the art in the field today.

Table 1

An overview of sensors, network architectures, feedback mechanisms, latencies and data update rates for state-of-the-art multi-sensor wearable products and prototypes.

Wearable System	Type	Sensors	BSN Architecture *	Feedback	Sensor Latency	Claimed Update/Sampling Rates, Hz
Hexoskin Astroskin [2]	Product	BP, IMU, TEMP, SpOx, ECG, BR	Mixed (n/a)	Visual	–	256
Sensor Suit [7]	Prototype	15 × IMU	Mixed (m-line)	Visual	–	320
Sensor fabric [24]	Prototype	63 × IMU	Mixed (line)	Visual	–	166
Aqua suit [25]	Prototype	5 IMU	Mixed (m-bus)	Visual	–	20
iFeel Suit [26]	Prototype	10 IMU	Mixed (m-bus)	Visual	–	80
AiQS suit [29]	Product	31 ×	Mixed (n/a)	Visual	–	200
Cometa Systems TrackX [30]	Product	36 × IMU, EMG	Wireless	Visual	–	400
Cometa Systems WaveTrack [31]	Product	IMU, EMG	Wireless	Visual	–	280
E-TeCS [32]	Prototype	34 × TEMP, IMU	Mixed (m-bus)	Visual	–	-
Nandi X Yoga pants [33]	Product	IMU	Mixed (n/a)	Visual, Audio	–	-
Nansense [34]	Product	50 × IMU	Mixed (n/a)	Visual	30 ms	240
Noraxon ultium motion [35]	Product	16 × IMU	Wireless	Visual	–	400
OWO Game Haptic Suit [36]	Product	–	Mixed (n/a)	Haptic	–	-
Perception Neuron 3 [37]	Product	17 × IMU	Wireless	Visual	–	60
Perception Neuron Studio [38]	Product	17 × IMU	Wireless	Visual	–	240
Rokoko Smart Suit Pro II [39]	Product	19 × IMU	Mixed (m-line/bus)	Visual	15 ms	200
Shadow Motion [40]	Product	17 × IMU	Mixed (m-line/bus)	Visual	20 ms	400
STT Systems iSen [41]	Product	16 × IMU	Wireless	Visual	–	400
Teslasuit [42]	Product	14 × IMU, PPG	Mixed (n/a)	Vibration, EMS, TENS, FES, visual	–	100
Wireless motion capture [43]	Prototype	10 × IMU	Wireless	Visual	–	59
Xsens Awinda [44]	Product	20 × IMU	Wireless	Visual	30 ms	60
Xsens Link [45]	Product	17 × IMU	Mixed (m-line/bus)	Visual	20 ms	240

* Mixed communication involves wired connections between nodes located on the body and wireless connections between on-body and off-body nodes. In parentheses, the topology is specified based on the available information: ‘n/a’ indicates no information, ‘m-bus’ denotes a multi-bus topology, and ‘m-line’ denotes a multi-line topology.

Summarizing the sensors used in the systems presented in Table 1, we identify several types employed for various forms of body-related information acquisition. Inertial Measurement Units (IMUs) are the most prevalent, extensively utilized in rehabilitation [8,46,47], sports [10,11], and virtual reality [14,48]. The number of nodes typically ranges from 10 for rough detailing, 15–20 being the most common, and up to 63 for detailed shape sensing.

Ten commercial products listed in Table 1, such as Xsense Link Shadow Motion, Nansensr, Rokoko Smart Suit, and others, utilize mixed communication architectures that integrate wired connections within wearable suits to collect sensor data at a central gateway, which is then transmitted via Bluetooth or Wi-Fi. However, these high-performance products rely on proprietary wired protocols, which severely limit scalability and accessibility for integration into other systems or broader applications.

In this context, our research explores the potential of wired communication for BSNs, addressing the critical gaps in scalable, high-demand wearable applications with minimal wiring and overhead. This paper builds on the aforementioned insights and presents an approach to developing interactive wearable systems that are optimized for real-time data acquisition and providing feedback in BSNs with a large number of sensors with minimal wiring using a novel group-based addressing protocol.

3. The Proposed Approach

3.1. Overall Architecture

Considering the requirements essential for wearable systems with numerous nodes distributed across the body (see Table 2), such as in motion capture or body shape reconstruction, the overall system architecture depicted in Figure 1 was selected.

3.2. Proposed Protocol

For the wired network illustrated in Figure 1, we propose a novel communication protocol that would enable sensor node data aggregation and actuator feedback intensity transmission to the nodes with reduced overhead compared to known alternatives. The key feature of the proposed protocol is its introduction of group communication for sensor data acquisition, replacing communication with individual nodes. The principles of both individual and group communication approaches are illustrated in Figure 2a and Figure 2b, respectively. Using individual addressing, sensors transmit their corresponding data after receiving their specific addressing blocks (A1–A3). In contrast, group addressing utilizes a single addressing block for the entire group (AG), and sensors transmit their data in a pre-negotiated order.

To prevent communication issues in case a sensor fails to respond in the expected way, a communication timeout is defined. After this timeout elapses, all network nodes assume that the group frame has ended, allowing for the bus to become available for other communication frames.

Beyond predefined communication order and timeouts, the proposed principle does not restrict how sensors are grouped, whether by order, size, or relationships. The choice of grouping depends on the specific application.

With slight modifications, this approach can be adapted to various connectivity interfaces commonly integrated in low-power microcontroller units (MCUs), such as SPI, I2C, and UART. Given the importance of minimizing wiring to enable seamless integration of imperceptible wired sensor networks into clothing, we believe UART is the most suitable choice as it does not require a dedicated wire for clock signals. Therefore, in the scope of this article, we further discuss the proposed communication protocol only in the context of a UART data transmission interface.

Although UART is considered a universal hardware block for asynchronous communication, generally, the configuration options for the UART physical layer in MCUs are relatively limited. The general format of a UART communication block is shown in Figure 3. It begins with a single start bit (ST); followed by 5 to 8 data bits, an optional address bit (AD) used for the multiprocessor protocol, an optional parity bit (PB); and ends with 1 to 2 stop bits (SP).

For scenarios requiring communication between more than two nodes, several multiprocessor communication modes are available for UART to implement a data link layer and higher. Common approaches include using idle frames, address bit, or a sync word to differentiate between the data packets. All of these can be adopted to implement the proposed group addressing protocol.

To indicate how many times this approach can potentially reduce the overhead for each addressing scenario, we consider a single group of N nodes, where each node’s data are b bytes-long, and define the overhead reduction coefficient $\mu$ corresponding to the ratio of the time spent on overhead with individual addressing to the time spent on overhead with group addressing (indicated by an asterisk):

(1)$\mu ={\displaystyle \frac{t_{\mathrm{frame}}-t_{\mathrm{data}}}{t_{\mathrm{frame}}^{\ast }-t_{\mathrm{data}}^{\ast }}}.$

Considering UART communication blocks with 8 data bits, no parity, and one stop bit, the overhead reduction achieved by utilizing the proposed communication approach for sensor data acquisition in each addressing scheme can be expressed as follows:

(2)${\mu }_{\mathrm{idle}}={\displaystyle \frac{20·N+2·N·b}{20+2·N·b}},$

(3)${\mu }_{\mathrm{addr}}={\displaystyle \frac{11·N+3·N·b}{11+3·N·b}},$

Figure 4 illustrates how overhead reduction is influenced by the number of data bytes (b) produced by each sensor and the number of sensors (N). It is evident that the potential for overhead reduction increases with a higher number of nodes and a smaller number of data bytes. For example, using the proposed protocol and UART idle frames, it is possible to reduce communication overhead by a factor of more than 1.8 when each sensor transmits 10 bytes of data and there are at least 10 sensors. However, if the amount of data per sensor increases to 50 bytes with the same number of nodes, the overhead ratio decreases to less than 1.2. Note that in these cases, no metadata accompanying the addressing block is considered. Therefore, the calculated overhead reduction represents the minimal possible improvement. In more practical implementations—such as those described in Section 4.2, where metadata for sampled sensors is included—the overhead reduction would be even greater.

There are several types of latencies important for real-time interactive systems. However, most of them characterize not only the communication protocol but the entire system. In the scope of our study, we considered two types of latencies:

1. Sensor Data Latency is measured from the moment data are sampled until they are available for processing. Referring to the frame diagram in Figure 2b, a convenient choice for synchronizing sensor sampling is immediately after the group addressing block (AG). In this configuration, the sensor data latency for a particular sensor depends on the baud rate S and the amount of data sent by the sensors preceding it. Consequently, the highest sensor data latency occurs for the last sensor.

2. Action Latency is measured from the moment feedback data become available until they are received by the feedback node. In scenarios where feedback data are transmitted to the entire group of nodes, the action latency for a particular node depends on the baud rate S, the length of the addressing block, and the amount of broadcast data for the preceding sensors. Consequently, the highest action latency occurs for the last sensor.

3.3. Node Architectures

For the hardware implementation of network nodes, we propose the schematics shown in Figure 5. Figure 5a presents the schematic illustrating the main blocks of the sensor nodes. It contains a microcontroller unit (MCU) with a processor and peripherals, which are used to communicate with one or more sensor units. To communicate within the sensor network, it includes a UART module connected in a half-duplex configuration. To ensure that the processor does not miss any data received by the UART module, a buffer for storing at least the received (Rx) values until they are processed is required. A transmit buffer is optional but recommended to alleviate the processor load during communication. It is important that the transmission (Tx) line of the hardware module can be disabled to free the communication bus for other nodes to transmit signals. It is recommended to place a resistor (50 $\mathrm{\Omega }$–100 $\mathrm{\Omega }$ ) before connecting the Tx line to the communication bus (Tx/Rx). This minimizes oscillations caused by rapidly switching signals along long, inductive wiring and helps prevent hardware damage in the event of a collision. Although the proposed architecture and protocol are designed to eliminate collisions by pre-determining the data transmission sequence between the controller and follower nodes, collisions may still occur if these rules are not followed. This can happen due to factors such as faulty hardware, misconfigurations, software bugs, and electrical interference.

The schematic illustrating the main components of the gateway is presented in Figure 5b. Similar to the sensor nodes, it includes an MCU with a processor and a UART module operating in half-duplex communication. Additionally, it contains a wireless communication module for interfacing with personal mobile devices through Wi-Fi or Bluetooth. Power is provided by a lithium polymer battery, which is recharged via a universal serial bus (USB) port. But depending on the application, this could be replaced with wireless charging or energy harvesting modules.

4. Experimental Setups and Results

4.1. The Adaption of the Proposed Protocol

The protocol proposed in Section 3.2 with slight modifications can be adopted for various connectivity interfaces and use cases. In this section, we describe our implementation of the proposed protocol for interactive sensor/actuator clothing. We focus on ensuring scalability in both the number of nodes and functionality, network testing, and tracking of missing data. The detailed protocol packet structure is shown in Figure 6. It begins with an invariable structure (Figure 6a) that includes a 4-byte preamble (0xAA, 0x55, 0xAA, 0x55) and a 4-byte header containing frame metadata: the frame address (Addr), frame type (Type), and frame sequence number (S_L and S_H). The frame address can be used to address individual nodes or a group of nodes. The frame type indicates the kind of information expected to follow in the payload. Frame sequence numbers are essential for tracking packet delivery.

The overhead reduction associated with this specific packet structure when reading N sensors, each producing b bytes of data, is defined as follows:

(4)${\mu }_{\exp }={\displaystyle \frac{80·N+2·N·b}{80+2·N·b}}.$

The maximum goodput can be calculated as follows:

(5)$G_{\exp }={\displaystyle \frac{N·b·S}{80+10·N·b}}.$

where S is the baud rate.

We assessed the overhead reduction ($\mu$) and goodput (G) achieved through group addressing when reading sensor data of different lengths (b) and from varying numbers of sensors (N), using the defined packet structure. The overhead reduction is depicted in Figure 7a, while Figure 7b presents the maximum goodput relative to the baud rate for $N=9$ and $b=32$, which matches the scenario for the nodes used in the interactive jacket described later in Section 4.2. From the graphs, we observe that in an experimental setup with nine nodes and 32 bytes of data per sensor, an overhead reduction of nearly twofold (1.98) can be achieved.

The sensor data latency (${\tau }_{\mathrm{D}}$) and action latency (${\tau }_{\mathrm{A}}$) introduced by the packet structure used in experimental setups (Figure 6) depend on the number of bytes produced by one sensor (b), the sensor’s sequence number within the group (n), and the baud rate (S):

(6)${\tau }_{\text{D\_exp}}={\displaystyle \frac{10·b·n}{S}},$

(7)${\tau }_{\text{A\_exp}}={\displaystyle \frac{80+10·n}{S}},$

Both delays are proportional to the number of sensors and inversely proportional to the baud rate.

The maximum frame rate (FPS) is calculated assuming that each sensor data packet is immediately followed by a feedback intensity packet:

(8)$FPS={\displaystyle \frac{S}{160+10·N·(1+b)}}.$

4.2. Interactive Jacket

To evaluate the proposed protocol for applications in wearable interactive systems, we developed an interactive sensor/actuator jacket, as shown in Figure 8a. The jacket comprises nine sensor/actuator nodes (Figure 8d) based on the architecture in Figure 5a, points for attaching and removing these nodes (Figure 8c), and a gateway (Figure 8b) based on the architecture in Figure 5b for transmitting information wirelessly between the wired BSN and a personal mobile device or computer.

The sensor/actuator nodes include the following components: a cylindrical vibration motor (10) for tactile feedback; an STM32C031G6U6 MCU (8) for sampling sensor data and implementing the proposed protocol; a BNO055 inertial measurement unit (IMU) (9) for movement and body pose measurements; a linear voltage regulator (7) for providing a stable 3.3 V power supply to sensor components; an electric field sensor (6) for measuring surrounding electrical field intensity; and a three-contact connector (5) for connecting to the power supply (VCC), ground (GND), and data (Tx/Rx) bus. These nodes can be attached to the jacket by snapping them onto the attachment points and detached when necessary, making it convenient for repairing and cleaning.

Because all nodes in the interactive jacket share the same design, the nine sensor/actuator nodes are grouped into a single group with Group_ID = 0. This choice reduces overhead by minimizing the number of addressing blocks. Each sensor is assigned with a unique identification number (Sensor_ID) ranging from 0 to 8, which determines the order in which sensors transmit their data. Each sensor produces the same amount of data ($b=32$ bytes), which includes sensor orientation, acceleration, angular rate, magnetic field vector, electric field intensity, calibration status, and a cyclic redundancy check. Additionally, each node can deliver tactile feedback, controlled by a 1-byte intensity setting. The flowchart in Figure 9 illustrates the MCU firmware logic for executing the adapted protocol. This logic was implemented in the sensor/actuator MCU firmware using hardware abstraction layer (HAL) libraries provided by STMicroelectronics. The values of SENSOR_ID, GROUP_ID, b, N, and the order of the sensor data are individually specified for each sensor node in the flashed firmware.

The gateway node includes the following components: a USB connector (1) and circuitry for programming the ESP32-WROOM-32E module and charging the 960 mAh lithium ion battery; power management circuitry (2) providing stable supply voltage for the gateway components and sensor/actuator nodes; ESP32-WROOM-32E module (3) to implement the proposed protocol, orchestrate communication in the wired BSN, and provide a communication link with external personal mobile devices and computers; and a sensor/actuator network connector (4) for connecting to the power supply (VCC), ground (GND), and data (Tx/Rx) bus of the BSN.

To connect the sensor/actuator nodes, we used a stretchable textile cable with four integrated wires, with two wires combined to serve as the ground wire, reducing ground resistance. Nevertheless, the used cable had considerable resistance, which had to be taken into account, to ensure proper power supply voltage. Table 3 lists the wire length (L) from the node to the gateway, resistance measured between each node, and both the ground and positive power supply voltages, relative to the gateway ground voltage at each sensor node location while all nodes are connected and operational. The measurements were taken using an Agilent U1252A handheld multimeter.

By observing the $V_{\mathrm{Node}}$ voltage, we confirmed that the power supply voltage for the nodes is sufficient for the linear regulator to operate properly. Additionally, the voltage drops on the wires introduce logical level offsets that remain within acceptable boundaries.

The maximum baud rate for symbol transmission in the experimental setup is limited to 3 MBd/s by the STM32C031G6U6 MCU used in the sensor nodes (48 MHz clock). However, parasitic wire characteristics such as resistance and inductance can also potentially constrain the baud rate for signal transmission. To determine the highest baud rate at which symbol transmission occurs without errors, we performed the Gateway–Group Communication Test. In this test, the gateway acted as the leader and simultaneously addressed a group of all sensors (S0–S8) in the jacket. The sensors in the group were expected to respond with the following five bytes of data: a 1-byte individual sensor ID, followed by 4 bytes echoing the packet header transmitted by the gateway. The test was conducted with one million consecutive packets at baud rates ranging from 1 MBd/s up to 3 MBd/s in increments of 0.1 MBd/s. As a result, the maximum baud rate at which the experiment encountered zero errors (such as sensor payload mismatches or timeouts) was determined to be 3 MBd/s.

Table 4 presents the theoretical delays, calculated using Equations (6) and (7), alongside the experimentally measured data delays, action delays, and total delays (${\tau }_{\mathrm{SUM}}$) for each sensor/actuator node location:

(9)${\tau }_{\mathrm{SUM}}={\tau }_{\mathrm{D}}\left(S8\right)+{\tau }_{\mathrm{A}}.$

The total delay, ${\tau }_{\mathrm{SUM}}$, is calculated under the assumption that feedback is sent only after the complete data packet is received. Consequently, the sensor data delay, ${\tau }_{\mathrm{D}}$, remains constant and is equal to the data delay of the last sensor in the interactive jacket (S8). Thus, the total delay for each node depends on the action delay, ${\tau }_{\mathrm{A}}$.

To experimentally measure the delays, a Rigol DG4062 function waveform generator was used to apply a square signal (26 Hz, 50 Hz duty cycle, high voltage = 3.3 V, low voltage = 0 V) to a dedicated digital input of a sensor node. The same packet payload structure used for sending real sensor data ($b=32$) was then employed to transmit the sampled input logical level to the gateway. After receiving data from all sensors, the gateway immediately responded with a packet containing feedback intensity settings for all nine sensors. The delay was measured using Rigol MSO4032 digital oscilloscope as the time difference between the signal pulse edges at the generator output and the individual sensor motor control inputs as shown in Figure 10.

From Table 4, we can see that at a baud rate of $S=3$ MBd/s, the maximum theoretical delay introduced by the packet structure in the protocol is approximately 1.02 ms. Based on experimental measurements using our interactive jacket implementation—which introduced additional systematic delays between data packets—the total delay for the last sensor is 2.27 ms ± 0.45 ms.

Considering the minimum time required to send data and feedback packets, our experimental evaluation achieved a maximum frame rate of 435.4 Hz. The theoretical limit for this particular packet structure, calculated using Equation (8), is 958.5 Hz.

Although we cannot directly compare the performance of our implementation to state-of-the-art products in Table 1 due to differences in the number of sensors, functionalities, and the lack of access to proprietary implementation details, our performance remains adequate for real-time interactive applications and post-processing. Additionally, since our implementation achieves less than half of the theoretically calculated performance, there is considerable potential for further improvements.

4.3. Long Cable Test

As with the interactive jacket, we did not observe the baud rate limitation introduced by connecting wires; we performed tests in a separate experimental setup with total wire lengths between the gateway and sensor nodes up to 15 m (as shown in Figure 11). This setup enabled us to determine the feasible symbol rates at longer cable lengths and assess the impact of different physical network topologies on the protocol’s performance.

The experimental setup shown in Figure 11 comprises a wireless gateway node (G) (refer to Figure 8b) acting as the leader and supplying power to the wired network nodes, 5 sensor (follower) nodes (S0–S4) (refer to Figure 8d), and unshielded copper cables (HELUKABEL® part no. 18003, 26 AWG, Shanghai, China) that carry the ground (GND), voltage supply (VCC), and data lines (Tx/Rx). Three of the four wires in the cable are used, with one wire left unused. A personal computer (PC) wirelessly logs data from the gateway node and controls its data transmissions.

Feasible symbol rates for the communication were determined using the Gateway–Group Communication and Gateway–Group Communication tests described below:

Gateway–Group Communication: The objective of this test was to evaluate the basic communication reliability between the gateway and individual sensors at different symbol rates. This communication is crucial in the proposed protocol for both transmitting feedback data and reading sensor data. The gateway requires all sensors in the addressed group to successfully receive the preamble and header to synchronize sampling and initiate sensor data transmission.

During the test, the gateway individually addressed each sensor node using the packet structure shown in Figure 6a. Each addressed sensor was expected to echo back a 4-byte payload containing the same information as the packet header transmitted by the gateway. The test was conducted with ${10}^6$ consecutive packets at baud rates ranging from 1 MBd/s up to 3 MBd/s in steps of 0.1 MBd/s.

As a result, the maximum baud rate at which the experiment encountered zero errors (sensor payload mismatches or timeouts) was recorded to be 1.3 MBd/s.

Gateway–Group Communication: The purpose of this test was to evaluate the proposed protocol for data acquisition from a group of nodes at different symbol rates. The gateway acted as the leader and addressed a group of all sensors (S0–S4) simultaneously. Sensors in the group were expected to respond with the following data: a 1-byte individual sensor ID, followed by 4 bytes echoing the packet header transmitted by the gateway. The test was conducted with ${10}^6$ consecutive packets at baud rates ranging from 1 MBd/s up to 3 MBd/s in steps of 0.1 MBd/s. As a result, the maximum baud rate at which the experiment encountered zero errors (sensor payload mismatches or timeouts) was obtained for different gateway locations. These locations were achieved by swapping the gateway node (G) with sensor nodes S0–S4 in the wire topology shown in Figure 11.

The results of the Gateway–Group Communication test revealed that the maximum baud rate for reliable data acquisition from a group of sensors at the default node locations in the experimental setup was 1.2 MBd/s, and changing the gateway location did not affect this rate.

5. Discussion

In this article, we introduced a wired approach with a novel group-based addressing protocol for interactive wearable systems, designed to support numerous nodes distributed across the body. Although our primary focus was on highlighting the potential for overhead reduction, node architectures, and implementation using a UART half-duplex configuration, the protocol also brings attention to considerations for sensor grouping.

In real-time scenarios, the primary consideration is to efficiently use communication time to meet data rate and latency requirements. From an efficiency standpoint, larger groups generally reduce overhead, as fewer addressing blocks are needed. Additionally, if certain groups’ sampling times do not fully align but occasionally overlap, forming a new group that includes nodes from both sets allows simultaneous data collection without increasing the overhead introduced by additional addressing blocks.

However, smaller groups provide more robust communication, as timeouts triggered by sensors that do not adhere to the group data order affect fewer nodes. In the scenario depicted in Figure 1, where a single large group (Group 1) may require lengthy data transfers, dividing it into smaller groups can help avoid delays for time-critical data blocks.

Within the scope of this article, we focused on the case of grouping all nodes into a single group to examine the resulting overhead reduction more closely. This choice was made because, for the chosen use case, the sensor/actuator nodes shared the same design, allowing us to study the impact of group addressing under uniform node conditions.

We found that the proposed group-based addressing scheme for wired communication in interactive wearable sensor networks significantly reduces overhead compared to conventional individual addressing methods. In our interactive jacket prototype with nine sensor/actuator nodes, each producing 32 bytes of data and using UART communication with 8 data bits, one start bit, and one stop bit, we achieved an overhead reduction of 1.98 times (almost twofold). This reduction becomes more pronounced as the number of sensors increases and the amount of data per sensor decreases, demonstrating the scalability of our approach.

To evaluate the proposed protocol, we developed an interactive sensor/actuator jacket comprising nine sensor/actuator nodes and a gateway for wireless communication with a personal mobile device or computer. The nodes were connected in a bus topology using three wires—ground, positive voltage supply, and data. Compared to I2C, which would require four, the enhanced daisy chain configurations proposed in [28] would require five wires to provide power supply and bidirectional communication. Also, compared to multi-line and multi-bus topologies used in other prototypes [7] or products [39,45], our approach offers greater flexibility for network physical topology, and the interactive jacket prototype had no overlapping cables to connect all nodes, resulting in more efficient wiring. The maximum distance between the gateway and sensors was 1.56 m, and the wires had a resistance of up to 3.95 $\mathrm{\Omega }$. By monitoring the sensor supply voltage for each node, we confirmed that the power supply was sufficient for the linear regulators to operate properly, providing a stable 3.3 V for the sensor node components. Voltage drops on the wires introduced logical level offsets that remained within acceptable boundaries.

Communication baud rate tests revealed that the maximum baud rate for the interactive jacket prototype was 3 MBd/s, limited by the maximum MCU capabilities in the sensor/actuator nodes. At this rate, the total delay for data acquisition and feedback was measured at 2.27 ms ± 0.45 ms, resulting in a maximum frame rate of 435.4 Hz, which is adequate for real-time applications and high-resolution data acquisition for post-processing. However, the limit set by theoretical boundaries offers twice the speed, with a total delay of 1.02 ms and a maximum frame rate of 958.5 Hz, leaving significant room for optimization in the implementation and reducing systematic delays. The achieved values could not be compared directly to performance metrics claimed by commercial products (Table 1); however, the extrapolated values indicate great potential for the proposed approach to outperform them.

The specific firmware sections responsible for the delays were not analyzed in detail. However, developing more optimized code to replace the functionalities provided by the STMicroelectronics HAL libraries could reduce the additional delays introduced by the firmware’s implementation of the communication logic.

To evaluate feasible baud rates and the impact of node locations, we conducted tests with longer cables, extending the wire length between nodes up to 15 m. We examined communication using both individual and group addressing schemes. For individual addressing, the maximum baud rate for error-free communication was slightly higher at 1.3 MBd/s across all node locations. With group addressing, the highest baud rate without errors was 1.2 MBd/s, which remained unaffected by the gateway’s location. This difference may be attributed to the fact that individual addressing transmits fewer bytes for a fixed number of packets since only one sensor response is expected, thereby reducing the likelihood of communication errors. Nevertheless, results indicate that wire length impacts the maximum reliable baud rate; specifically, longer wires lead to decreased baud rates for reliable communication. Also, we found no evidence that changing node locations or the gateway position affects the maximum baud rate using group addressing.

Building on the demonstrated performance, our proposed communication approach and node architectures have the potential to serve as foundational building blocks for new products or research in various interactive applications, such as smart workwear, sports clothing, and smart textiles for daily use. We envision that this approach could also be beneficial for applications beyond wearables—for example, in robotic systems where efficient communication networks are crucial. Currently, the protocol’s functionality is basic, focusing on reading data and providing feedback with pre-programmed group configurations. Introducing additional features like node discovery, dynamic grouping, collision management (for multi-controller applications), and node reprogramming could enhance versatility and ease of use. Furthermore, exploring the scalability of the proposed approach with a larger number of nodes may unlock more sophisticated and data-rich use cases. However, in this paper, scalability was not extensively tested beyond nine sensor/actuator nodes and 15 m cables. Additionally, our implementation achieved only half the frame rate indicated by theoretical analysis, highlighting areas for future research in algorithm optimization. Finally, we note that the proposed group addressing approach can be adapted for other semi-duplex serial interfaces beyond UART, offering potential advantages for a wider range of applications.

Author Contributions

Conceptualization, A.A. and M.G.; methodology, A.A.; software, A.A. and S.K.; validation, A.A. and S.K.; formal analysis, A.A. and M.G.; investigation, A.A.; resources, A.A.; data curation, S.K.; writing—original draft preparation, A.A.; writing—review and editing, A.A. and M.G.; visualization, A.A.; supervision, M.G.; project administration, M.G.; funding acquisition, M.G. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:

STD	Standard deviation
BSN	Body sensor network
MCU	Microcontroller unit
I2C	Inter-integrated circuit protocol
UART	Universal asynchronous receiver–transmitter
SPI	Serial Peripheral Interface
IMU	Inertial measurement unit
EMG	Electromyography
PPG	Photoplethysmogram
TEMP	Temperature
BP	Blood pressure
OX	Pulse oximetry
ECG	Electrocardiogram
BR	Breathing rate
EMS	Electrical muscle stimulation
TENS	Transcutaneous electrical nerve stimulation
FES	Functional electrical stimulation

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. The overall architecture of a wired wearable interactive system.

Figure 2. Communication frame diagrams using different addressing approaches: (a) individual addressing, (b) group addressing.

Figure 3. The format of a UART data block.

Figure 4. The proposed protocol’s overhead reduction for the UART multiprocessor communication, with orange lines representing the use of idle blocks and green lines indicating the use of the address bit. Parameters: 1 group, 8 data bits, 0 parity bits, 1 start bit, and 1 stop bit per UART data block.

Figure 5. Proposed schematic designs for network nodes: (a) sensor node. (b) Gateway.

Figure 6. Proposed protocol packed structure: (a) preamble and header structure including synchronization word, address (Addr), type, and the low byte S_L and high byte S_H of the sequence number; (b) payload structure for sensor data readout; (c) payload structure for actuator feedback intensity setting (FI) for each node in the group.

Figure 7. Characteristics of the adapted configuration of the proposed protocol used in the experimental setups: (a) overhead reduction ratio, (b) maximum achievable goodput ([Forumla omitted. See PDF.], [Forumla omitted. See PDF.]).

Figure 8. Interactive sensor jacket prototype: (a) node numbering and layout (components that are not visible from the beginning are shown with dashed lines); (b) gateway components: 1—USB connector, 2—battery charging and power management, 3—ESP32-WROOM-32E wireless module, 4—sensor/actuator network connector; (c) sensor/actuator node attachment point; (d) sensor/actuator node components: 5—connector for VCC, GND, Tx/Rx, 6—electrical field sensor circuit, 7—linear voltage regulator, 8—STM32C031G6U6 microcontroller unit, 9—BNO055 inertial measurement unit, 10—cylindrical vibration motor.

Figure 9. The logical flowchart of the sensor firmware to provide group sensor data readout and feedback intensity settings.

Figure 10. Oscilloscope signals used to measure the total delay of the experimental setup.

Figure 11. Experimental setup for testing the proposed wired communication system: overview of the main components.

Table 2

Requirements for interactive wearable systems.

No.	Requirements	Rationale
1.	Multiple sensor/actuator nodes distributed throughout the body, grouped based on their sampling rate.	Based on the sensors listed in Table 1. BSNs have many similar nodes but are placed on different body parts.
2.	Flexible wire routing to nodes, allowing for optimized wire usage or redundant connections to enhance reliability.	Approaches reviewed in Section 2 often resulted in overlapping cables to connect all nodes on the body [7,39,45].
3.	Centralized power supply and a single wireless gateway to enable communication with external devices without limiting freedom of movement	Common approach for mixed BSN architectures listed in Table 1.

Table 3

Wire resistance and power supply voltage measurements at node locations.

Node	L, m	${\mathit{R}}_{\mathbf{GND}}$, $\mathbf{\Omega }$	${\mathit{R}}_{\mathbf{VCC}}$, $\mathbf{\Omega }$	${\mathit{R}}_{\mathbf{Tx}/\mathbf{Rx}}$, $\mathbf{\Omega }$	${\mathit{V}}_{\mathbf{GND}}$, V	${\mathit{V}}_{\mathbf{VCC}}$, V	${\mathit{V}}_{\mathbf{Node}}$, V
G	0	0	0	0	0	5.48	5.48
S0	0.62	0.85	1.49	1.45	0.31	4.84	4.53
S1	0.68	0.98	1.75	1.67	0.36	4.68	4.32
S2	0.76	1.06	2.03	1.9	0.41	4.58	4.17
S3	0.91	1.28	2.42	2.44	0.45	4.51	4.06
S4	1.27	1.73	3.27	3.45	0.5	4.41	3.91
S5	1.56	2.11	4	4.28	0.52	4.37	3.85
S6	0.91	1.33	2.44	2.23	0.46	4.52	4.06
S7	1.27	1.74	3.23	3.15	0.51	4.42	3.91
S8	1.56	2.08	3.95	3.91	0.53	4.39	3.86

Table 4

Theoretical minimum and experimentally measured delays.

Node	${\mathit{\tau }}_{\mathbf{D}}$, $\mathsf{\mu }$s	${\mathit{\tau }}_{\mathbf{A}}$, $\mathsf{\mu }$s	${\mathit{\tau }}_{\mathbf{SUM}}$, ms	${\mathit{\tau }}_{\mathbf{SUM}}$ *, ms	STD *, ms
S0	106.67	30	0.99	2.251	0.44
S1	213.33	33.33	0.99333	2.272	0.4503
S2	320	36.67	0.99667	2.247	0.4457
S3	426.67	40	1	2.254	0.4386
S4	533.33	43.33	1.00333	2.267	0.4472
S5	640	46.67	1.00667	2.263	0.4459
S6	746.67	50	1.01	2.262	0.4454
S7	853.33	53.33	1.01333	2.28	0.4432
S8	960	56.67	1.01667	2.27	0.4452

* Experimental measurements.

References

1. Shajari, S.; Kuruvinashetti, K.; Komeili, A.; Sundararaj, U. The Emergence of AI-Based Wearable Sensors for Digital Health Technology: A Review. Sensors; 2023; 23, 9498. [DOI: https://dx.doi.org/10.3390/s23239498]

2. Astroskin—The Most Advanced Ambulatory Vital Signs Monitoring Platform for Medical Research in the World. Available online: https://hexoskin.com/pages/astroskin-vital-signs-monitoring-platform-for-advanced-research (accessed on 7 November 2024).

3. Dheman, K.; Giordano, M.; Thomas, C.; Schilk, P.; Magno, M. I-CardiAx: Wearable IoT-Driven System for Early Sepsis Detection Through Long-Term Vital Sign Monitoring. Proceedings of the 2024 IEEE/ACM Ninth International Conference on Internet-of-Things Design and Implementation (IoTDI); Hong Kong, China, 13–16 May 2024; pp. 97-109. [DOI: https://dx.doi.org/10.1109/IoTDI61053.2024.00013]

4. Zhang, X.; Gu, Y.; Yan, H.; Wang, Y.; Dong, M.; Ota, K.; Ren, F.; Ji, Y. Wital: A COTS WiFi Devices Based Vital Signs Monitoring System Using NLOS Sensing Model. IEEE Trans. Hum.-Mach. Syst.; 2023; 53, pp. 629-641. [DOI: https://dx.doi.org/10.1109/THMS.2023.3264247]

5. Sethuraman, S.C.; Kompally, P.; Mohanty, S.P.; Choppali, U. MyWear: A Novel Smart Garment for Automatic Continuous Vital Monitoring. IEEE Trans. Consum. Electron.; 2021; 67, pp. 214-222. [DOI: https://dx.doi.org/10.1109/TCE.2021.3085888]

6. Neseem, M.; Nelson, J.; Reda, S. AdaSense: Adaptive Low-Power Sensing and Activity Recognition for Wearable Devices. arXiv; 2020; arXiv: 2006.05884

7. Ancans, A.; Greitans, M.; Cacurs, R.; Banga, B.; Rozentals, A. Wearable Sensor Clothing for Body Movement Measurement during Physical Activities in Healthcare. Sensors; 2021; 21, 2068. [DOI: https://dx.doi.org/10.3390/s21062068] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33809433]

8. Liang, W.; Wang, F.; Fan, A.; Zhao, W.; Yao, W.; Yang, P. Extended Application of Inertial Measurement Units in Biomechanics: From Activity Recognition to Force Estimation. Sensors; 2023; 23, 4229. [DOI: https://dx.doi.org/10.3390/s23094229] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37177436]

9. Linnamo, V. Sensor Technology for Sports Monitoring. Sensors; 2023; 23, 572. [DOI: https://dx.doi.org/10.3390/s23020572] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36679367]

10. Arlotti, J.S.; Carroll, W.O.; Afifi, Y.; Talegaonkar, P.; Albuquerque, L.; Burch, R.F.; Ball, J.E.; Chander, H.; Petway, A. Benefits of IMU-based Wearables in Sports Medicine: Narrative Review. Int. J. Kinesiol. Sports Sci.; 2022; 10, pp. 36-43. [DOI: https://dx.doi.org/10.7575/aiac.ijkss.v.10n.1p.36]

11. Seckin, A.C.; Ates, B.; Seckin, M. Review on Wearable Technology in Sports: Concepts, Challenges and Opportunities. Appl. Sci.; 2023; 13, 10399. [DOI: https://dx.doi.org/10.3390/app131810399]

12. Morais, J.E.; Oliveira, J.P.; Sampaio, T.; Barbosa, T.M. Wearables in Swimming for Real-Time Feedback: A Systematic Review. Sensors; 2022; 22, 3677. [DOI: https://dx.doi.org/10.3390/s22103677]

13. Islam, M.S.; Lim, S. Vibrotactile feedback in virtual motor learning: A systematic review. Appl. Ergon.; 2022; 101, 103694. [DOI: https://dx.doi.org/10.1016/j.apergo.2022.103694]

14. Caserman, P.; Garcia-Agundez, A.; Gobel, S. A Survey of Full-Body Motion Reconstruction in Immersive Virtual Reality Applications. IEEE Trans. Vis. Comput. Graph.; 2020; 26, pp. 3089-3108. [DOI: https://dx.doi.org/10.1109/TVCG.2019.2912607] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31021797]

15. Hajar, M.S.; Al-Kadri, M.O.; Kalutarage, H.K. A survey on wireless body area networks: Architecture, security challenges and research opportunities. Comput. Secur.; 2021; 104, 102211. [DOI: https://dx.doi.org/10.1016/j.cose.2021.102211]

16. IEEE Std 802.15.6-2012 IEEE Standard for Local and Metropolitan Area Networks—Part 15.6: Wireless Body Area Networks; IEEE: Piscataway, NJ, USA, 2012; [DOI: https://dx.doi.org/10.1109/IEEESTD.2012.6161600]

17. Ancans, A.; Ormanis, J.; Cacurs, R.; Greitans, M.; Saoutieff, E.; Faucorr, A.; Boisseau, S. Bluetooth Low Energy Throughput in Densely Deployed Radio Environment. Proceedings of the 2019 23rd International Conference Electronics; Palanga, Lithuania, 17–19 June 2019; pp. 1-5. [DOI: https://dx.doi.org/10.1109/ELECTRONICS.2019.8765577]

18. Hassan, W.H.W.; Ali, D.M.; Sultan, J.M.; Kassim, M. A review of wireless body area networks. Indones. J. Electr. Eng. Comput. Sci.; 2024; 33, pp. 167-178. [DOI: https://dx.doi.org/10.11591/ijeecs.v33.i1.pp167-178]

19. Preethichandra, D.; Piyathilaka, L.; Izhar, U.; Samarasinghe, R.; De Silva, L.C. Wireless Body Area Networks and Their Applications—A Review. IEEE Access; 2023; 11, pp. 9202-9220. [DOI: https://dx.doi.org/10.1109/ACCESS.2023.3239008]

20. Xiang, H.; Li, Y.; Liao, Q.; Xia, L.; Wu, X.; Zhou, H.; Li, C.; Fan, X. Recent Advances in Smart Fabric-Type Wearable Electronics toward Comfortable Wearing. Energies; 2024; 17, 2627. [DOI: https://dx.doi.org/10.3390/en17112627]

21. Nie, M.; Li, B.; Hsieh, Y.L.; Fu, K.K.; Zhou, J. Stretchable One-Dimensional Conductors for Wearable Applications. ACS Nano; 2022; 16, pp. 19810-19839. [DOI: https://dx.doi.org/10.1021/acsnano.2c08166] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36475644]

22. Guo, Z.; Yu, P.; Li, B.; Xuan, F.; Zhao, J. Direct wire writing technique benefitting the flexible electronics. Virtual Phys. Prototyp.; 2023; 19, e2286514. [DOI: https://dx.doi.org/10.1080/17452759.2023.2286514]

23. Dang, C.; Wang, Z.; Hughes-Riley, T.; Dias, T.; Qian, S.; Wang, Z.; Wang, X.; Liu, M.; Yu, S.; Liu, R. et al. Fibres-threads of intelligence-enable a new generation of wearable systems. Chem. Soc. Rev.; 2024; 53, pp. 8790-8846. [DOI: https://dx.doi.org/10.1039/D4CS00286E] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39087714]

24. Hermanis, A.; Cacurs, R.; Greitans, M. Acceleration and Magnetic Sensor Network for Shape Sensing. IEEE Sens. J.; 2016; 16, pp. 1271-1280. [DOI: https://dx.doi.org/10.1109/JSEN.2015.2496283]

25. Marta, G.; Pedrocchi, A.; Ferrante, S.; Costa, A.; Bortolotti, D.; Sorti, S.; Vitale, F.; Bocciolone, M.; Braghin, F.; Masiero, S. Wearable Biofeedback Suit to Promote and Monitor Aquatic Exercises: A Feasibility Study. IEEE Trans. Instrum. Meas.; 2020; 69, pp. 1219-1231. [DOI: https://dx.doi.org/10.1109/TIM.2019.2911756]

26. Sortino, D.M.; Rapetti, L.; Valli, E.; Pucci, D. Towards a real-world application of wearable sensors for musculoskeletal disorders prevention: The iFeel Wired Suit. Proceedings of the 2023 IEEE 19th International Conference on Body Sensor Networks (BSN); Boston, MA, USA, 9–11 October 2023; [DOI: https://dx.doi.org/10.1109/BSN58485.2023.10331259]

27. Mourkani, S.S. IMU-Based Suit for Strength Exercises: Design, Calibration and Tracking. Doctoral Thesis; Technische Universität Kaiserslautern: Kaiserslautern, Germany, 2021; [DOI: https://dx.doi.org/10.26204/KLUEDO/6470]

28. Hermanis, A.; Cacurs, R.; Nesenbergs, K.; Greitans, M. Efficient real-time data acquisition of wired sensor network with line topology. Proceedings of the 2013 IEEE Conference on Open Systems (ICOS); Kuching, Malaysia, 2–4 December 2013; pp. 133-138. [DOI: https://dx.doi.org/10.1109/ICOS.2013.6735062]

29. AiQ.Synertial Products. Available online: https://www.synertial.com/products (accessed on 7 November 2024).

30. TrackX Inertial System—The Ultimate Solution for Inertial Motion Tracking. Available online: https://www.cometasystems.com/trackx-inertial-system/ (accessed on 7 November 2024).

31. Wave Track Inertial System—Complete, High Performance IMU Sensor Synchronised with EMG. Available online: https://www.cometasystems.com/wavetrack-inertial-system/ (accessed on 7 November 2024).

32. Wicaksono, I.; Tucker, C.I.; Sun, T.; Guerrero, C.A.; Liu, C.; Woo, W.M.; Pence, E.J.; Dagdeviren, C. A tailored, electronic textile conformable suit for large-scale spatiotemporal physiological sensing in vivo. npj Flex. Electron.; 2020; 4, 5. [DOI: https://dx.doi.org/10.1038/s41528-020-0068-y] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38624354]

33. Nandi X—How It Works. Available online: https://www.wearablex.com/pages/how-it-works (accessed on 7 November 2024).

34. Nansense—Suits & Full Body Tracking. Available online: https://nansense.com/suits/ (accessed on 7 November 2024).

35. Ultium Motion—Wearable 3D Motion Capture. Available online: https://www.noraxon.com/our-products/ultium-motion/ (accessed on 7 November 2024).

36. Owogame Technology—Feel Different Worlds with the OWO Skin. Available online: https://owogame.com/technology/ (accessed on 7 November 2024).

37. Perception Neuron 3—A Mocap Suit You Can Take Anywhere. Available online: https://neuronmocap.com/pages/perception-neuron-3 (accessed on 7 November 2024).

38. Perception Neuron Studio—Professional and Precise Motion Capture System. Available online: https://neuronmocap.com/pages/perception-neuron-studio-system (accessed on 7 November 2024).

39. Rokoko Smartsuit Pro II Full Performance Capture Bundle. Available online: https://www.voosestore.com/en/studio/rokoko-smartsuit-pro-ii?srsltid=AfmBOorlaJcToNzn_KsxFXGY-m6P2NKhe9plsRBnF6q81Rg-pNr2rGg- (accessed on 7 November 2024).

40. Shadow Motion—Technical Data. Available online: https://www.motionshadow.com/specs (accessed on 7 November 2024).

41. iSen Inertial Motion Capture—The Solution for Motion Capture and Analysis with Inertial Sensors. Available online: https://motio.stt-systems.com/isen-inertial-motion-capture-and-analysis/ (accessed on 7 November 2024).

42. Tesla Suit—A Breakthrough in Human Performance Training. Available online: https://teslasuit.io (accessed on 7 November 2024).

43. Lee, K.; Tang, W. A Fully Wireless Wearable Motion Tracking System with 3D Human Model for Gait Analysis. Sensors; 2021; 21, 4051. [DOI: https://dx.doi.org/10.3390/s21124051] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34204656]

44. MTw Awinda—Specifications. Available online: https://www.movella.com/products/wearables/xsens-mtw-awinda#specs (accessed on 7 November 2024).

45. Xsens Link—Specifications. Available online: https://www.movella.com/products/motion-capture/xsens-mvn-link#specifications (accessed on 7 November 2024).

46. Gu, C.; Lin, W.; He, X.; Zhang, L.; Zhang, M. IMU-based motion capture system for rehabilitation applications: A systematic review. Biomim. Intell. Robot.; 2023; 3, 100097. [DOI: https://dx.doi.org/10.1016/j.birob.2023.100097]

47. Al-Amri, M.; Nicholas, K.; Button, K.; Sparkes, V.; Sheeran, L.; Davies, J.L. Inertial Measurement Units for Clinical Movement Analysis: Reliability and Concurrent Validity. Sensors; 2018; 18, 719. [DOI: https://dx.doi.org/10.3390/s18030719]

48. Franček, P.; Jambrošić, K.; Horvat, M.; Planinec, V. The Performance of Inertial Measurement Unit Sensors on Various Hardware Platforms for Binaural Head-Tracking Applications. Sensors; 2023; 23, 872. [DOI: https://dx.doi.org/10.3390/s23020872]