Real‐time vibration monitoring and analysis of agricultural tractor drivers using an IoT‐based system

Agricultural tractor drivers experience a high amplitude of vibration, especially during soil tillage operations. In the past, most research studied vibration exposure with more focus on the vertical (z) axis than on the fore‐and‐aft (x) and lateral (y) axes. This study examines how rotary soil tillage affects the vibration acceleration and frequency, and the power spectral densities (PSDs) at the seat pan and head along three translational axes in a real‐field multiaxis vibration context. Moreover, this study aimed to identify the characteristics of the seat‐to‐head transmissibility (STHT) response to identifying the most salient resonant frequencies along the x‐, y‐, and z‐axes. Nine (9) male tractor drivers operated the tractor with a mounted rotary tiller throughout the soil tillage process. In the event of a COVID‐19 pandemic, and to respect social distancing, this study developed an Internet of Things (IoT) module with the potential to integrate with existing data loggers for online data transmission and to make the experimentation process more effective by removing potential sources of experimenter errors. The raw acceleration data retrieved at the seat pan and the head were utilized to obtain daily exposure (A(8)), PSDs, and STHT along the x‐, y‐, and z‐axes. The vibration energy was found to be dominant along the z‐axis than the x‐ and y‐axes. A(8) response among tractor drivers exceeds the exposure action value explicitly stated by Directive 2002/44/EU. PSDs along the x‐, y‐, and z‐axes depicted the low‐frequency vibration induced by rotary soil tillage operation. The STHT response exhibited a higher degree of transmissibility along the y‐ and z‐axes when compared with that along the x‐axis. The frequency range of 4–7 Hz may plausibly be associated with cognitive impairment in tractor drivers during rotary soil tillage.


| INTRODUCTION
The Internet of Things (IoT) has the potential to revolutionize agriculture and related industries by using a network of interconnected devices to gather real-time data, improve efficiency, and streamline processes (Tao et al., 2021). IoT refers to a system of interconnected devices that can transmit data and interact with each other without human intervention (Perwej et al., 2019). IoT technology has numerous applications in agriculture, from precision farming to livestock management (Elijah et al., 2018). In agriculture, smart devices equipped with sensors and actuators can be used to monitor soil moisture, temperature, humidity, and other environmental conditions (Sales et al., 2015). This information can be used to optimize crop growth and yield while minimizing the use of resources, such as water and fertilizers. In addition, IoT can be used to monitor the health and behavior of farm worker, improving their well-being and productivity (Ponnusamy & Natarajan, 2021). IoT technology has the potential to monitor and improve tractor ride comfort that can have a significant impact on the health and safety of tractor drivers.
For example, IoT systems can be installed in tractors to monitor vibration and their effects on drivers (Koene et al., 2020). This data can be analyzed to identify ways to improve tractor design for better ride comfort.
Agricultural workers operating tractors are subjected to wholebody vibrations (WBVs), which may be highly severe depending on the mounted farm implements, the tractor's forward speed, and field conditions . These vibrations are complex and variable, with multiaxis translational and rotational vibration inputs reaching various body areas through the buttocks, torso, and spinal column (Scarlett et al., 2007). Working in such an environment causes human weariness, which influences ride comfort and other health problems such as metabolic disorders, and cardiovascular and nervous system problems (Loutridis et al., 2011;Solecki, 2010). Tractor vibrations were also reported to affect the driver's performance (Bovenzi, 2010;Sam & Kathirvel, 2006;Village et al., 2012). Previous research indicated that increased WBV exposure was more severe in off-road than on-road situations (Adam & Jalil, 2017;Adam et al., 2020;Kim et al., 2018). As a result, vibration in off-road vehicles is more likely to violate the recommended health guidance caution zone of 8 h exposure within 24 h (ISO 2631(ISO -1, 1997. Increased WBV exposure among tractor drivers is a significant occupational risk factor often linked to lower back pain (Adam et al., 2020;Singh et al., 2018;Singh, Nawayseh, et al., 2022).
Additionally, WBV may elongate and shorten muscles, raising muscular tension as a stretch reflex (Ritzmann et al., 2010) which could lead to muscle fatigue. Low-frequency vibrations, or those between 1 and 10 Hz, are the real issue that tractor operators have to deal with because of uneven terrain, imbalanced rotating engine parts, a mix of different vehicle components, and other factors (Singh, Nawayseh, et al., 2019). Since different body parts have low natural frequencies, the human body is very sensitive to low-frequency vibration (Liang & Chiang, 2008). The operator's whole body is affected by the low frequencies, which causes rapid weariness, inefficient functioning, reduced working performance, and temporary or permanent harm to the driver's health. Such extended exposures may affect the operator's ability to operate the vehicle (Cutini et al., 2016;Harsha et al., 2014;Singh et al., 2018Singh et al., , 2021. WBV directly affects driver attentiveness by raising the physical and mental strain on the driver, which causes fatigue and drowsiness (Bhuiyan et al., 2022). Low-frequency WBV, especially those between 4 and 10 Hz, are the most effective at causing drowsiness (Satou et al., 2009;Zhang et al., 2018). It is the transitional period between being awake and the first stage of sleep, often known as N1 or nonrapid eye movement stage 1 (Rechtschaffen, 1968). The eyelids open and shut slightly, the muscles relax, and the eyes may roll gently. Theta waves (4-7 Hz) replace alpha waves (8-13 Hz) on the brain electroencephalogram (EEG). During the N1 stage of sleep, a person's ability to quickly react to driving conditions is significantly impaired, even though the person may appear awake and believe to be aware of surroundings.
Understanding the biodynamic responses to vibration is necessary to comprehend how it influences comfort, health, and activity performance (Griffin & Erdreich, 1991). To define the key frequency ranges of vibration under which larger deflections and hence strains of the biological system may arise, biodynamic research has been extensively employed to find the resonance frequencies of the body (Rakheja et al., 2020). Experimental biodynamic studies have greatly advanced our understanding of body mechanics and movement, the effects of posture and vibration-related factors, probable modes of vibration, resonance frequencies, potential injury mechanisms, and frequency-weighting for exposure assessments (Desai et al., 2021;Hinz et al., 2002;Mertens, 1978;Nawayseh & Griffin, 2005;Wang et al., 2004Wang et al., , 2018Zhang et al., 2015).
However, the vibration may be stronger in certain directions than others, depending on the exposed environment. Fore-and-aft and lateral vibration, whose magnitudes may be equivalent to those of the vertical in trains and off-road vehicles, has received significantly less attention in the published research on biodynamics than have the reactions to vertical vibration (Demić & Lukić, 2009;Gong & Griffin, 2018;Rakheja et al., 2008). Several studies (Boileau et al., 2002;Kumar, 2004;Newell et al., 2006;Rehn et al., 2005;Village et al., 1989) have shown that the exposure levels are significantly influenced by the activities done, the terrain conditions, and the vehicle size among other factors (Maeda & Morioka, 1998). The whole-body vertical mode, often referred to as the primary resonance of the transmissibility responses to vertical vibration, occurs in the 4-7 Hz range (Mansfield & Griffin, 2000;Matsumoto & Griffin, 2000;M-Pranesh et al., 2010). According to certain observed data, secondary vibration modes in the 8-15 Hz range were also present, although not always obvious for all the subjects (Mansfield et al., 2001;Wang et al., 2004). When seated without back support, Paddan and Griffin (1988) observed that the seat-to-head transmissibility (STHT) magnitude peaked at 3 and 1.5 Hz under both x-and y-axes vibrations. When sitting without back support and with the x-and y-axes vibrating, Hinz et al. (2010) discovered primary resonance at 1 Hz. The fore-and-aft STHT responses assessed with back support exhibited an additional peak at 8 Hz, while the main peak shifted toward 2 Hz. However, the back support showed no effect on the frequency of the lateral STHT.
The previous research focused mostly on on-road vehicles while performing controlled biodynamic investigations on lab simulators. There is a dearth of knowledge of the biodynamic response to vibration encountered in off-road vehicles, like, tractors. Additionally, no effort has been made to observe the STHT behavior of tractor drivers in a real-field context, particularly during soil tillage operations. The possible association of WBV with cognitive behavior has received comparatively little attention (Bhuiyan et al., 2022). To explore the transmissibility behavior among tractor drivers during actual field rotary tillage operations, the current research makes an effort to investigate the WBV levels thoroughly. Investigating the dominant resonance frequencies that can affect the driver's cognitive function is another key component of this research.
This research developed an IoT module that allowed for the realtime collection of more naturalistic data and facilitated social distancing during the COVID-19 pandemic without the need for human labor. Moreover, the IoT module has the potential to interface with existing data loggers for online data transmission. This is an expanded module of the Autonomous Quality Management System (AQMS), developed before and based on Quality 4.0 (Singh, Ahuja, et al., 2022). It led to remote data collection utilizing an Android application. Overall, the IoT module developed in this work was used to remotely collect data needed to provide (a) detailed information related to WBV levels at seat pan along the three translational axes in actual field tillage contexts, (b) frequency component characterization in terms of power spectral density (PSD) at both measurement locations, and (c) STHT to extract the frequencies at which the head vibrates during field tillage operation.

| RESEARCH METHODOLOGY
The methodology of the present study will be presented through three main sections: (a) apparatus, (b) experimental conditions and procedure, and (c) data analysis. Each section includes subsections elaborating on the content of that section as described in Figure 1.

| Apparatus
This section describes the developed IoT module, the machinery, and the sensors and data acquisition systems used in this study.

| IoT module
The recent global challenge of the COVID-19 pandemic made the authorities worldwide demand and apply safety measures including social distancing. Thus, an advanced technology with the ability to connect with the existing devices (e.g., data loggers) was needed for obtaining the data remotely without human interactions to support social distancing. This study required two more persons along with the tractor driver to manage the data loggers as presented by Singh, Nawayseh et al. (2019). So, this study has developed an IoT module for the online data transmission from the cloud to web-cum-application (Android).
The vibration data loggers (described in Section 2.1.3) were equipped with RJ45. The RJ45 interface connection has formed an input unit, which is further controlled by an ESP32-WROOM Series microcontroller unit. This unit (or chip) consists of a twocore central processing unit on the printed circuit board (PCB) that is referred to as the heart of the all-electronic components.
A Micro-SD unit with a storage capacity of 8 GB was installed to store the input data. The operating range of the ESP32 module is 2.2-3.6 V. However, the Micro-USB connector was used to deliver 5 V. A low-dropout regulator voltage regulator has already been installed for 3.3 V to maintain 3.3 V as the constant voltage.  Figure 3 shows the detailed representation of the rotavator and blades.

| Sensors and data acquisition systems
The vibration levels were measured at two different locations, that is, input location (at the seat pan) and output location (at the head) to evaluate the vibration at the seat pan and the STHT along the three translational axes. Two MEMS-based triaxial accelerometers type SV 38 V housed in SAE pads and SV 107B were used to measure the acceleration at the seat pan and the head, respectively. The accelerometer mounted on the head had a strap that was used to tighten the accelerometer to the forehead of the participants to reduce the relative motion between the skull and the forehead skin.
The accelerometers were calibrated before the experimentation. The SV 38 V and SV 107B accelerometers were connected to an SV 106 A six-channel human vibration meter and analyzer and an SV 958 four-channel sound and vibration analyzer, respectively, to record the SINGH ET AL.
| 1725 vibration at the seat pan and the head along the fore-and-aft (i.e., x), lateral (i.e., y), and vertical (i.e., z) directions. The sampling rate of both data acquisition systems was set at 6 kHz and a band-limit (low pass) weighting filter was used as per the standard (ISO 2631-1, 1997).

| Experimental conditions and procedure
This section describes the terrain, the climate conditions, the participants, the experimental procedure, and the data collection procedure. The present study was performed on a real agriculture field to comprehensively assess the actual vibration exposure levels.
It required several preliminary facilities including agricultural terrain, soil testing (soil texture, moisture level, and soil compactness), mainly to create a setup for the experimentation.

| Test terrain
The study was carried out on a postharvested wheat field of size 2.073-acre located at Punjab Agriculture University Ludhiana, Punjab, India. Before the experimentation, soil samples were collected and analyzed from the four random locations to get the soil texture using ISO 14688-1:2002. The field was tested sandy clay loam with a composition of clay 27%, sand 66.25%, and silt 9.27%. Soil moisture content was found between 48.61% and

| Climate conditions
The experimentation was carried out in the June-July month in Ludhiana, Punjab, India. The general climatic conditions related to the air temperature, atmospheric pressure, humidity, precipitation, and wind speed were noted. The air temperature varies from 32.8°C to 42.5°C on average over the experimental hours, with an atmospheric pressure of 999.5 mbar, humidity ranging from 38% to 47%, and wind speeds between 10.2 and 12.5 km/h.

| Participants
Nine male participants with mean (±standard deviation) age 28  misleading results. All the participants belong to agriculture background with a minimum experience of 7 years in tractor driving. Each participant was interviewed directly to assess health-related details.
None of the participants reported any existing issues related to musculoskeletal disorders or sensitivity toward the exposure to vibration. The study was approved by the Safety and Ethics Committee of GNDEC Ludhiana, confirming compliance with ethical principles and safety regulations. All subjects provided their written consent before their participation in the study.

| Experimental procedure
The experimental procedure is divided into pilot trial and main experiment. The pilot trial was performed by all participants so as to practise the ride before conducting the main experiment. After the health-related interview, the objectives and procedure of the experiment were explained to the participant. The participant then signed the informed consent form and conducted the pilot trial. The pilot trial conditions included three input parameters, that is, tractor speed (m/s), pulling force (kN), and average tillage depth (m).
Participants were instructed to set the tractor in first-low, that is, 1-L gear and maintain a driving speed in the range of 0.6-0.8 m/s as recommended for the rotavator operation (Bureau of Indian Standards, 1998). This operation required a standard pulling force (2, 4, and 6 kN) depending upon the field conditions during tilling.
Moreover, the pulling force was confirmed using a dynamometer mounted with the drawbar and the rotavator (Singh, Nawayseh, et al., 2019). Depending upon the pulling force, the tillage depth levels were measured manually using a scale ruler that provided an average depth of 0.10, 0.12, and 0.14 m (Singh et al., 2018). These conditions were made clear to the participants. The data collected during the pilot trials were not included in the reported results.
In the main experiment, each participant conducted soil tillage operation at the ride conditions described above concerning tractor speed, pulling force, and tillage depth. The participants were instructed to ride the tractor sitting in their preferred sitting posture to obtain more naturalistic data. All the participants chose a forward lean sitting posture without backrest support. Each participant repeated the experiment three times. The test duration of each run was 60 s. The time duration was determined by the maximum forward distance that the tractor can reach in a continuous run within the field length. The mean vibration data of the three runs was calculated for each participant at each location and direction.

| Data collection
During the experiment, the measured data were communicated to the IoT module. The ESP32 provides remote access to the measured data through its built-in wireless fidelity (Wi-Fi) and Bluetooth capabilities, enabling seamless data transfer. The module connects to a Wi-Fi network for data transmission to facilitate cloud connectivity.
This allows for efficient and reliable storage of the data. Finally, the data files were transferred to an Android application named https:// ergoaman1.web.app/#/ that can be downloaded in the form of.txt and/or.xlsx files. An overview of the complete data collection approach can be visualized in Figure 4.
F I G U R E 4 Overview of the data collection approach. IoT, Internet of Things.

| Data analysis
The vibration measured at the two locations is the time-domain raw acceleration response along the three directional axes. Due to the effect of terrain on in-field experiments, the data need to be carefully analyzed. The overall goal was to meticulously examine the STHT, which required dividing the input values (i.e., vibration data at the seat pan) by in-field conditions. However, at certain frequencies, either no input signal or a small input signal was retrieved.
Consequently, the outcome may lead to high peaks in the transfer function. Hence, the raw acceleration data were preprocessed to remove the signal shifts by removing the mean from the signals.
Then, the data were processed to get a statistical measure of the signals, that is, weighted root mean square acceleration (A w ). The time-domain data were converted into the frequency components by obtaining Fast Fourier Transform (FFT). FFT analysis will provide the information pertaining to the existing dominant frequencies at seat pan and at head locations along the three x-, y-, and z-axes. The power in that frequency bands was analyzed using PSD. Finally, STHT was evaluated to examine the possible primary and secondary resonance frequencies in the power spectral. Due to narrowband random signals, that is, vibration data, this study used the Hanning window function to minimize the spectral leakage and obtain good spectral resolution as well. The data analysis was carried out using MATLAB (version 2022a).
Overall, the data analysis provides information relating to the study outcomes, that is, A w response along the three translational axes, that is, x-, y-, and z-directions; daily exposure (A(8)) response, PSD, and STHT at seat pan and head locations. The cross-spectral density (CSD) method was used in this study to determine the transfer functions between seat and head motions (Bendat & Piersol, 1980). The CSD function uses the following Equation (1) to calculate the proportions of the output motion that are linearly linked with the input motion: where H C (f) is the transfer function determined using the CSD function approach, G xy (f) the cross-spectrum of the seat pan (input) and head (output) motions, and G xx (f) the power spectrum of the input, that is, seat pan motion.
Since the vibration on the seat was found dominant in the zdirection (please see Section 3), it was decided to use the z acceleration measured on the seat pan as the input for the calculation of the x, y, and z STHTs as follows (Equations 2-4): where G x z h s , G y z h s , and G z z h s represent the CSD between the z acceleration of the seat and the x, y, and z accelerations of the head, respectively, G z z s s is the PSD of the z acceleration of the seat pan, and f is the frequency.

| RESULTS AND DISCUSSION
Integrating an IoT module with vibration sensors has successfully enabled the acquisition of real-time vibration signals. The use of IoT technology in this study has provided an effective solution for collecting data remotely. It has eliminated the need for human intervention to manage the data loggers, which has resulted in increased efficiency. In this section, the results of the study will be presented and discussed taking into consideration the subobjectives of the study listed at the end of the Introduction section.
The impact of tractor ride parameters (i.e., tractor speed, pulling force, and tillage depth) on the WBV exposure among the tractor driver has been reported in previous research studies but without taking into consideration the amount of vibration transmitted through the human body (Singh et al., 2018(Singh et al., , 2021. These studies were focused on evaluating the occupational ride comfort, that is, overall vibration dose value along the nine-axis (i.e., 3 translational at the floor near to seat mounting area, at the seat pan, and at the backrest) measurement system and compressive stress on the lumbar spine (i.e., daily static equivalent compression dose (S ed )) based on recommended standards (ISO 2631(ISO -1, 1997ISO 2631ISO -5, 2004), respectively. The present research mainly focuses on examining vibration exposure levels in the frequency domain to investigate existing dominating frequencies and the effects of the vibration at those frequencies on the transmission of vibration through the human body. The present study is a step toward biodynamic response assessment during the actual field rotary tillage operation in multiaxis excitation (three translational and three rotational). The measured motion along each individual x-, y-, and z-axes is believed to have been affected by the vibration along the other axes (Nawayseh & Griffin, 2009;Nawayseh et al., 2020).
The mean A w response of the three replications was calculated for each measurement location along the three translational axes, as shown in Figure 5. The A w data values at the seat pan varied from 0.21 to 0.28, 0.27 to 0.32, and 0.68 to 0.76 m/s 2 , along the x-, y-, and z-axes, respectively. Similarly, the A w response at the head varied from 0.19 to 0.24, 0.26 to 0.37, and 0.52 to 0.61 m/s 2 along the x-, y-, and z-axes, respectively. A w values were found to lie within the range of the weighted acceleration exhibited in tractor-related research studies (Okunribido et al., 2006;Scarlett et al., 2007).
However, these studies considered test tracks or different terrain conditions for tractor rides without any actual function of tillage implementation. Furthermore, the studies were mainly confined to reporting A w information based on seat pan location. The maximum A w values obtained in the current study were lower than those reported in previous studies. This might be owing to the soil-SINGH ET AL. | 1729 dampening effect caused by tillage depth. The passage of vibration energy gets slowed down due to increased stiffness and adhesion among soil particles (Kim & Lee, 2000). Since the distance between the vibration source and the driver increases as the depth of the tilling tool increases, the amplitude of vibrations that enters the driver's body gets reduced (Singh, Nawayseh, et al., 2019).
The decrease in vibration levels may also be attributed to the geometric difference between the different machines in addition to material damping. Further, the mean A(8) response was calculated to be 0.72 m/s 2 at the seat pan. At both measurement locations, the mean A w response was most prominent along the z-axis, followed by the y-and x-axes, as illustrated in Figure 5. In addition, the A(8) response at the seat pan and the head exceeded the permissible exposure action value limit, that is, 0.5 m/s 2 (Directive & Provisions, 2002). Such increased vibration exposure level could lead to occupational risks related to low back pain (Adam et al., 2020), muscle stretch or contraction (Ritzmann et al., 2010), negative effect on the performance of activities (Bovenzi, 2010;Village et al., 2012), and discomfort (Griffin, 2007).

| Power spectral density
The raw acceleration data at the seat pan and head were evaluated along the x-, y-, and z-axes to visualize and determine the dominant resonance frequencies. The PSD spectra for both measurement locations were observed in the frequency range of up to 30 Hz. The frequency spectra offer little to no observable information above 10 Hz. Therefore, PSD spectra were included in this study for the frequency component range between 0 and 12 Hz.

| PSD analysis at the seat pan
The PSD response along the x-axis at seat pan revealed several peaks in the frequency range of 0.1-2 Hz (Figure 6). The primary peak is in the range of 0.3 and 0.5 Hz among all participants. Furthermore, a substantial secondary peak has been detected nearly at 1 Hz. Most participants demonstrated a frequency peak with very high energies at 2 Hz, whereas the remaining responses showed a significantly lower intensity around 2 Hz. The spectrum revealed little or no energy in the vibration signal between 2.1 and 3.6 Hz. A minor peak of low intensity was identified between 3.8 and 4 Hz, and no discernible peaks were found up to 5.8 Hz. Furthermore, it was determined that all participants had a peak response between 6 and 7 Hz. The remainder of the spectra showed few visible peaks with lower magnitudes at 9-9.5 and 10.2 Hz.
The PSD of the acceleration measured at the seat pan along the y-axis revealed a significant peak between 0.2 and 0.5 Hz (Figure 6).
Most responses showed a notable peak at about 1 Hz. One additional peak with a high magnitude was identified around 2 Hz. Aside from that, various low amplitude peaks between 0.1 and 2 Hz can be noticed. The spectra revealed few or no discernible peaks between 2.1 and 3.7 Hz. Then, from 3.8 to 4 Hz, a modest magnitude peak was spotted. There was no or just a little visible peak between 4.1 and 5.9 Hz. Furthermore, all participants showed a dominating peak with a high magnitude between 6 and 7 Hz. Additionally, the high peak was found at 9-9.5 Hz and around 10.2 Hz.
The PSD of the acceleration at the seat pan along the z-axis indicated the primary peak at nearly 1 Hz ( Figure 6). Then, between 1.8 and 2 Hz, a relatively small magnitude peak was observed.
Furthermore, several low-intensity peaks were identified between 0.5 and 4 Hz. The frequency spectra demonstrated few or no apparent peaks between 4.1 and 5.9 Hz. A secondary dominating peak was also identified between 6 and 7 Hz. The vibration signals exhibited minimal energy between frequencies 7.1 and 8.9 Hz.
However, a notable peak with a lower magnitude than the principal peaks was noticed between 9 and 9.5 Hz and around 10.2 Hz.

| PSD analysis at the head
The PSD of the acceleration measured along the x-axis at the head revealed a common peak at 1 Hz and around 2 Hz for the majority of the participants (Figure 7). Multiple peaks with significant magnitudes were observed between 0.5 and 2 Hz. The spectra indicated a peak response near 2-3 Hz in a few cases. In most responses, however, there were few or no peaks between 2 and 5.5 Hz. The primary frequency peak was typically observed around 6-6.5 Hz and at 7 Hz. Some peaks with fairly low magnitudes were visible between 10 and 12 Hz.
The PSD of the acceleration measured along the y-axis at the head showed a primary peak between 1 and 1.5 Hz (Figure 7). A dominant secondary peak was observed between 2.5 and 3 Hz. In most cases, several peaks with relatively small magnitudes were observed along the y-axis throughout the spectra. However, in some cases, the peak response was visible between 9 and 11.8 Hz.
The PSD of the acceleration measured at the head along the zaxis demonstrated a dominant peak with a high magnitude near 1 Hz ( Figure 7). In some cases, the peak was dominant between 2 and F I G U R E 5 Mean vibration response of participants based on measurement locations and directional axis.
2.5 Hz. In most cases, a visible peak was observed near 6 and 7-8 Hz, and the frequency spectra show multiple peaks within the 4-8 Hz range. Furthermore, the vibration magnitudes were moderate, between 4 and 8 Hz. Aside from that, multiple peaks with lesser magnitudes were observed between 10 and 12 Hz.
The PSD analysis showed that the tillage operation causes lowfrequency vibration exposure at the seat pan and the head along the x-, y-, and z-axes. This is consistent with the previously reported results for vibration measured on the seat pan in the z-direction during rotary soil tillage operation (Singh, Nawayseh, et al., 2019).
F I G U R E 6 PSD of acceleration measured at the seat pan along the x-, y-, and z-axes for the nine participants. PSD, power spectral density.
The low-frequency vibration zone (1-10 Hz) has been identified as significant owing to the presence of the natural frequencies of the human body components in the low-frequency region (Kumar et al., 2001). A variety of human body organs have natural frequencies close to such low frequencies, including the spine (3-4 Hz), shoulder girdle (4-8 Hz), heart (4-6 Hz), and entrails (3-7 Hz) (Jayasuriya & Sangpradit, 2014). So, the frequency range of 2-7 Hz is the most dangerous for tractor operators due to those resonances, which may induce different problems in the human body including musculoskeletal issues, particularly in the lower back (Griffin, 2007; F I G U R E 7 PSD of acceleration measured at head along the x-, y-, and z-axes for the nine participants. PSD, power spectral density. Taghizadeh-Alisaraei, 2017). Furthermore, these frequencies may result in stomach discomfort, muscular strain, and headaches (Griffin & Erdreich, 1991).
If the driver is exposed to vibration at those low frequencies with high amplitudes for an extended period, the driver's health and safety would be compromised (Prasad et al., 1995). According to field observations, the driver's head motion seemed to be influenced by the tractor-terrain interaction during tillage operations. The tractor driver may have experienced jerks due to the uneven terrain.
Consequently, the head's frequency response exhibited several peaks along the x-, y-, and z-axes. The PSD analysis showed high vibration of the head along the three axes, namely, x, y, and z represented by several peaks in those directions. It has also been observed that vibration energy measured at the head tends to fluctuate during the tilling process over the specified frequency range. The high motion of the head could be partly due to the pitch, yaw and roll motions of the head arising from the uneven terrain and facilitated by the high flexibility of the Craniovertebral Junction (CVJ) that connects the head and neck (Osawa & Oguma, 2013). The CVJ comprises the occipital bone, the atlas (C1), and the axis (C2), as well as a network of nerve and vascular systems. The atlas, occipital bone, and axis are primarily responsible for the head's rotation, extension, and flexion (Jung & Bhutta, 2021).

| Seat-to-head transmissibility
This section comprises detailed information on tractor drivers' biodynamic responses represented by STHT (Figure 8). Transmissibility is frequency response functions that can be used to figure out the mechanical features of the human body when exposed to vibration including identifying the frequency ranges related to the resonances of different body parts. Transmissibility functions have also been used to study the dynamic properties of seats and identify whether the seat amplifies or attenuates the vibrations at certain frequencies which is reflected on the passengers' ride comfort (Toward & Griffin, 2010;Zhang et al., 2015).
This study evaluated acceleration data collected at seat pan and head locations to determine STHT for the nine tractor drivers along the x-, y-, and z-axes. Even though there was significant dispersion in the STHT of the different drivers, the first peak was typically around 0.36 Hz along the x-axis. In addition, some peaks also appeared in the frequency range of 4.5-6.9 Hz.
The secondary resonance peak was mostly found in the 8.7-11.7 Hz frequency range. Among the subjects, the STHT revealed discernible peaks of much lower amplitude at 12, 13.5, and 14.6 Hz. The STHT along the y-axis exhibited preliminary peaks at 0.73 and 3.6 Hz. The dominant resonance peak was identified in the 4.7-6.9 Hz frequency region, comparable to the x-axis. The secondary resonance peak was seen among all the participants in the 8.6-11.8 Hz frequency range. In most instances, peak moduli were also seen at 12.1 and 13.54 Hz. In most cases, the STHT in the z-axis showed peaks in the 0.73-3.6 Hz frequency range. The primary resonance was found in the 4.1-6.5 Hz frequency range. A secondary resonance peak in the 8.7-12.1 Hz frequency range was also seen in the STHT along the z-direction. In most cases, there were common peaks at 13.1 and 13.5 Hz. Overall, the primary resonance peak of the STHT response was found to lie between 4 and 7 Hz along the x-, y-, and z-axes for all participants, as shown in Figure 9a-c. The peak of the secondary resonance was also seen at a frequency of 8-12 Hz. Bhiwapurkar et al. (2016) investigated the vertical STHT response and found identical responses to the present study's primary and secondary resonance peaks. According to Wang et al. (2006), larger magnitudes of head vibration may be experienced in the 3-10 Hz frequency range along the z-axis while driving with no backrest support and hands supported on the steering wheel. Preliminary peaks in the STHT were typically between 0.73 and 3.6 Hz along the y-and z-axes. Despite this, it has been observed that the frequency of the x-axis responses is much lower, around 1.3 Hz (Lee & Pradko, 1968) and approximately 0.7 Hz (Fairley & Griffin, 1990;Mandapuram et al., 2005Mandapuram et al., , 2010. Paddan and Griffi (1988) found that the STHT magnitude peaks at 3 and 1.5 Hz while seated without back support under vibration in the x-and y-axes, respectively. This analysis supported the peaks observed along the x-axis; however, the z-axis trend was distinct. This may imply that the STHT trends may differ for the same axes depending on whether the exposure is uniaxial (e.g., lab experiment) or multiaxial field study, for example, passenger car or actual field cultivation (Nawayseh, 2018). Hinz et al. (2010) found that the STHT peaked at roughly 1 Hz along the x-and y-axes when the subject was seated without a backrest. The hands on the steering wheel may also cause a minor increase in y-axis vibration at the head at the principal resonance frequency. In this research, the STHT along the y-axis exhibited resonance peaks with large amplitudes compared with the STHT along the x-axis for all participants. The x-and y-axes peaks often overlapped with the STHT along the z-axis which may indicate that rotation motions produce those peaks. The STHT along the z-and y-axes is often greater than that along the x-axis. This could be attributed to the way the body is coupled to the steering wheel through the hand-arm system. With placing the hands on the steering wheel, the hand-arm system provides resistance to the translational motion of the body in the x-direction in addition to pitch and yaw motions of the body. On the other hand, coupling of the hand-arm system with the steering wheel does not provide high resistance to vertical, lateral, or roll motions. Mandapuram et al. (2005) showed that without back support, the normalized magnitude of the y-axis increased in the region of the resonant frequencies. Additionally, this study demon-  Griffin, 1989;Park et al., 2019;Summala, 2000). The present research found that frequencies ranging from 4 to 7 Hz and 8 to 13 Hz dominated during multiaxis vibration exposure.
Low frequencies, especially 4-10 Hz, may influence the driver's cognitive domain, causing drowsiness and reducing alertness (Kimura et al., 2017). In addition, human performance has been reported to degrade significantly in the frequency ranges of 1-2 Hz in the horizontal plane and 4-8 Hz in the vertical plane (Griffin & Erdreich, F I G U R E 8 STHT response along the x-, y-, and z-axes for the nine participants. STHT, seat-to-head transmissibility. 1991; Mabbott et al., 2001). Stamenkovic et al. (2014) found that the 5 Hz frequency had the g reatest effect on response time owing to resonance compared with the 1, 20, and 50 Hz frequencies. Newell et al. (2006) found that 1-20 Hz randomized multiaxis vibration substantially impacts the capacity to do an activity. It has been reported that 2-8 Hz significantly impacted performance (Newell & Mansfield, 2008). When compared with 1.8 Hz or no vibration, Jiao et al. (2004) found that 6 Hz caused more mental fatigue in drivers.
Similarly, Zhang et al. (2018) found that 4-7 Hz frequency range leads to driver drowsiness more quickly than with no vibration. Thus, there seems to be general agreement that WBV frequencies between 4 and 7 Hz are more likely to cause drowsiness and decreased cognitive function than other frequencies. The fact that brain waves occur in the 4-7 Hz range could be the reason to why this range of frequencies induces a sense of exhaustion and reduced alertness.
Theta waves, which predominate during the early sleep phases, have a frequency range between 4 and 8 Hz. In contrast, delta waves, between 1 and 4 Hz, are associated with deep sleep, and alpha waves, between 8 and 13 Hz, and beta waves, between 13 and 30 Hz, are related to alertness (Nguyen et al., 2017;Teplan, 2002).
WBV produces bilateral and temporally synchronized somatosensory and proprioceptive input, which is expected to strongly stimulate cortical activity. It would be likely that WBV between 4 and 7 Hz is adequate to entrain cortical theta waves and prepare the brain for the onset of the early phases of sleep (Bhuiyan et al., 2022).

| CONCLUSION AND FUTURE IMPLICATION
On the basis of the results of this study, the following conclusions and future implications are drawn: • This study has developed an IoT system that can remotely collect vibration signals in real-time. In the future, researchers may expand the system to automatically generate alerts (such as via email or text) using technologies, like, ThingSpeak. These alerts could warn if the vibration exposure level exceeds recommended thresholds. Additionally, researchers may expand the IoT system by integrating physiological sensors (such as MAX30102 PPG and heart-rate sensors, AD8232 ECG sensors, and LIS3DH accelerometers) to monitor the tractor driver's physiological health in real-time. Future studies may investigate the relationship between vibration exposure and the tractor driver's physiological parameters during various agricultural practices.
• The PSD response at the seat pan and the head demonstrates that the rotary soil tillage operation generates low-frequency vibration along the x-, y-, and z-axes. In addition to the peaks noticed in the z-axis, the PSDs in the x-and y-axes also had significant dominating frequencies. However, the vibration energy changes across the low-frequency spectrum during tillage.
• The STHT response mostly showed a predominant range of frequencies between 4-7 and 8-12 Hz along the x-, y-, and z-axes.
The magnitudes of transmissibility were greater along the y-and zaxes than along the x-axis.
• During rotary soil tillage operations, tractor drivers were more susceptible to cognitive impairment (4-7 Hz), specifically drowsiness and decreased attention.
• The research presented the actual field information pertinent to acceleration and frequency (PSDs and STHT) characteristics among the tractor drivers at the two vital sites (at seat pan and head) along the three translational axes. Tractor designers and manufacturers may use the information to dampen critical resonance frequencies and improve riding comfort.
• This research explained the possible association between lowfrequency vibration exposure and cognitive impairment during F I G U R E 9 Intersubject variability in the STHT of the nine participants along the (a) x-axis, (b) y-axis, and (c) z-axis. STHT, seatto-head transmissibility. SINGH ET AL. | 1735 field tillage. This information could be useful to automotive industry personnel, allowing them to focus more on cognitive aspects while designing and manufacturing automobiles. Additionally, future research is urged to measure and examine EEG signals to investigate the brain's overall electrical activity and evaluate cognitive function during the exposure to vibrating (Hervé et al., 2022;Wen et al., 2022).
• Future studies should measure visual acuity, hand-eye coordination, response time, and memory under the WBV conditions.
• Although this study focused on a sample of nine participants, future research could expand the sample size to improve the generalizability of the findings.