Optimized design and performance analysis of wearable antenna sensors for wireless body area network applications

ABSTRACT Over the past decade, the continuous development and the high level of maturity of wireless sensor networks resulted in new networks called wireless body area networks (WBANs), which are an emerging sector of biomedical technology. Moreover, this field has gained significant attention due to its applications which mainly are toward biomedical and healthcare applications. Nowadays, small sensors that can transfer data to other devices can now be implanted anywhere on the human body to record different physiological indicators and enable further actions to be conducted, such as processing, remote procedures and decision aided. Considering this recent hot subject, the intent of this work is to present a new approach of the optimized design and the performance analysis of WBAN specifically channel modelling between wearable wireless sensors. The behaviour of these sensors on the human body is theoretically and experimentally explored in detail along this paper. A good agreement is obtained between the theoretical and the experimental results, although the complexity of the physiological behaviour of human body.


Motivation and related work
The increasing use of wireless networks and the proliferation of miniaturized sensors lead to network applications that can be applied on a human body providing a multitude of services. A wireless body area network (WBAN) is considered as a wireless sensor network (WSN) that is assembled with different intelligent devices such as sensors, actuators and nodes. In this context, the WBAN uses have known a large field of application domains, such as the medical domain (Tavera et al., 2021), the military domain (Egbogah & Fapojuwo, 2013;Saravanakumar et al., 2021), sports (Jin, 2022), (Roshini & Kiran, 2022), multimedia Hassan et al., 2017), etc. The WBAN is classified under IEEE 802.15.6 standard (Das et al., 2021). The purpose of this standard is to provide a medium access control and physical (PHY) layers for WBAN. Thus, these layers are based on rigorous propagation channel modelling and antenna design. Consequently, WBAN has attracted researchers from academia and industries to investigate the opportunities for new use cases WSN-based user scenarios for health care and study their effect on the solutions of the issues and propose viable solutions, especially during the rapid spread of certain pandemics such as Coronavirus Disease 2019  where the intervention must be very rapid. WBANs use the human body as a transmission medium for electromagnetic signals. Although human tissue represents naturally an absorbent layer, it acts as a low loss dielectric medium which can mitigate propagation signals. Thus, in AL-Asadi and Wadday (2021), the authors introduce an approach of channel modelling that can be served for the estimation of channel gain and link loss in electromagnetic wave (EMW) propagation media. In AL-Asadi and Wadday (2021) and Munoz et al. (2014), the authors present WBAN use cases and they revealed that the wireless body sensors may further monitor human vital signs or other physiological data for a variety of applications such as healthcare facilities, elderly or people with disabilities assistance, user authentication, etc. WBAN data are collected by different body sensors and processed by the coordinator. To study propagation radio waves and then analyse the performance of channel model inside or on the body surface, the modelling of the body is necessary and all transmission media should be rigorously investigated as illustrated in Figure 1. In other words, it is important to have knowledge of the electromagnetic properties of biological tissues at a given frequency. Besides, the human body is a stratified dielectric with varying physical properties significant with tissue types and frequency of work. To explore these properties, such a task can be completed either numerically or experimentally in two steps. In a first step, geometrical assumption of the human body has been used. Second, accurate measurement campaigns on volunteer human persons were performed. Furthermore, even while WBANs make it possible for anybody to connect to the global network, they also require assistance from a variety of software implementations, including database processing, remote procedures and a user interface, in addition to the network infrastructure. However, these systems must have minimal consumption and little effects on routing (Al-Janabi et al., 2017).

Contribution
In the WBAN field, our work is concerned mainly with channel modelling field, energy consumption prediction and antenna design of level 1 of Figure 1. It deals with the design and development of mathematical channel model-based human body morphology that takes into account the different body tissues specifications in order to contribute to the improvement of channel characterization field. Various types of WBAN communication mode are explored in Tavera et al. (2021), AL-Asadi and Wadday (2021) and Haddad et al. (2022). Onbody channels are the most sensitive to the presence and movement of the human body compared to off-body channels. Different existing channel models have been classified according to the body communication link (in-body, on-body and off-body) (Bharadwaj et al., 2021;Mohamed et al., 2021). During our research, we noticed that there is a lack of energy models dedicated to conceiving the energy consumption in a WBAN system and taking into consideration the different propagation channel features. These features include body tissue effects, body movements, physical activities, as well as the dynamic environment due to twisting, running, multipath and mobility. Hence, based on the existing energy models (Kaushik et al., 2019;Kumar & Singh, 2018;Ullah et al., 2021;Waheed et al., 2018;Yang et al., 2020) and our proposed channel model, we will attempt to suggest a new energy model containing propagation channel characteristics, the signal to noise ratio, and the small-scale fading characterized by the multipath. The suggested energy model is based on human body propagation characteristics and it is explained in detail along this paper. The channel characteristics for line of sight propagation along human body parts will be studied by experiments. A set of experimental data helps to predict the human body behaviour for each considered scenario. Our objective from an experimental study is to validate the mathematical formulation and optimize the link budget by using different devices and testing multiple scenarios in two different environments (anechoic chamber and indoor multipath laboratory). The obtained results are encouraging and promising and make a significant contribution compared to prior works which are illustrated in Table 3.
Finally, this paper is structured as follows. Section 2 details the proposed analytical model of WBAN. Section 3 exposes the experimental study and the measurement results. Section 4 draws the conclusions presented as the opportunities for research.

Path gain determination
In this subsection, we present the suggested model of WBAN based on the approach of a simple wireless communication between two body antennas (transceiver, T x , and receiver, R x ) at 2.4 GHz. In this context, a model of on-body communication scenario, illustrated in Figure 2, is proposed with all dimension parameters and all electromagnetic properties of mediums (I) and (II). Rigorous sequence formulas are proposed to expect the channel model based on the interaction of EMW of the free space (medium I) with the body waist (medium II).
The medium II in Figure 2 represents the human tissue and has also an additional effect on the EMW propagation depending on the variability of physiological characteristics. To ensure a rigorous investigation of on-body communication channel, the T x /R x antennas are placed closely on the human body waist, with a curvature distance equal to 30 cm. The choice of the waist as an investigated human body part defines two locations of antennas at the same cross-section of the waist. Seeing that, we aim to validate our model experimentally; these locations are favourable to use in practice, since the antennas performances are inconsistent when operating in a bent condition.
We can assume a curvilinear path close to the form of human body waist with relative permittivity (ϵ r ) and conductivity (σ) equal to 50 and 1.7 S/m, respectively, at 2.4 GHz. In this scenario, we have a model that is as similar to reality as feasible in terms of characteristics. Using 'Friis formula', we can extract the transmitted (P T x ) power with the gain (G T x ), and the received power (P R x ) with the gain (G R x ), with a known distance (d = r) between T x /R x . The Friis formula is expressed as follows: Using Equation (1) and neglecting the heights of antennas (T x /R x ), heights (h t ) and (h r ), for simplification reasons, the received power can be expressed as follows: where P T x and P R x are expressed in dBm, and G T x and G R x are expressed in dBi. A p is expressed in dB and it is considered as the consistent factor needed to optimize the link power budget between T x /R x antennas. A p evaluation depends essentially of the frequency ( f ), the distance (d) and the attenuation of propagation medium. To determine the path loss for the 2.4 GHz on-body propagation, we focus on the calculation of approximate path loss both in free space and around cylindrical or elliptical assumption of the human body for antennas polarized normal to the body surface.
2.2. Path loss of medium (I): the case of the free space According to Equation (1), the path loss in the free space (medium I, Figure 2) can be expressed as At a reference distance d 0 , the member L 0 can also be considered as a path loss and expressed as L 0 = 20 log 10 (4p/c). (4) When the frequency and the distance are in MHz and in kilometre, respectively, L 0 is equal to 32.44 dB. However, for the proposed model, the operating frequency is in the order of 2.4 GHz and the distance between T x /R x is very small, usually less than 1 m for real human body. In case where frequency ( f ) is expressed in GHz (1000 MHz), (20log 10-(1000)) equals to 60 dB. We add 60 dB to L 0 of Equation (4). Then, L 0 grows into 92.44 dB. In case of the free space, Equation (3) of the path loss becomes A = 92.44 + 20 log 10 ( f .d). (5) When the medium (I) presented in Figure 2 is considered as the free space and the operating frequency ( f ) is more than 1 GHz, the path loss will be expressed by the product of the frequency ( f ) and the distance (d) as follows:

Path loss of medium (II): the case of the human body waist
The human body waist is regarded as a complicated medium in electromagnetic characterization. Permittivity (ϵ r ), conductivity (σ), dielectric constant (TanD), and characteristic impedance (Z 0 ) are variable parameters in this medium. As a result, the body's tissues and positions cause the propagating wave to have some electromagnetic effects, including energy absorption, reflection, diffraction and shadowing. As attenuation (α B ) may be represented in Np/cm in Equation (7) and converted to dB in equation, we can roughly describe these effects in this work (Equation (8)).
a B(dB) ≈ 520.8ps According to Hall et al. (2007), at 2.4 GHz, the human body has the following electromagnetic parameters: permittivity ϵ r = 50 and conductivity σ = 1.7 S/m. For 25 cm T x /R x distance, the attenuation (α B ) is approximately 100 dB.
2.4. Formula of total path loss of the proposed model: in mediums (I) and (II) By a simple summation of Equations (5) and (8), we can obtain the expression for total path loss (A PL ) as follows: A PL = 92.44 + 20 log 10 (f ) + 20 log 10 (d) + 520.8ps 1 r √ .
Furthermore, all things surrounding and near the human body can create multipath phenomena. Usually, the multipath represents a major factor producing fading which makes the WBAN channel modelling dissimilar from the ones in the other environments. As shown in Figure 3, the proposed path loss model is illustrated as a function of the T x /R x antenna distance associated to free space path loss and body path loss separately. According to previous research in this field (Ahmed, 2020;Alves et al., 2010;Barake et al., 2016;Chandra & Johansson, 2012), the developed model is considered as a novel analytical method which is, at the same time, simple, accurate and optimized for numerical simulations. In WBAN deterministic channel modelling, the mathematical equation (9) is a novel generic model for on-body applications and can be equally extended to offbody communication. This expression will be used later for the modelling of energy consumption during wave propagation along body surfaces.

Analysis of on-body EMWs propagation
In the following part, the path gain theory introducing the transmitting and receiving power as well as the T x /R x antennas gains is rewritten within body surface at 2.4 GHz. The following scenario is taken into account to characterize wave propagation in the human body communication environment: the human body at 2.4 GHz is equipped with two antennas as T x /R x . In reality, if the considered human body parts have EMWs, it propagate as creeping waves, a curved path and a flat surface, they propagate as surface waves. All system characteristics in channel modelling with consideration of electromagnetic propagation properties of human body waist and free space are rigorously studied in this paper. At the working frequency 2.4 GHz, an analytical path gain model is suggested to predict the on-body propagation model. Starting with the most generic Friis equation (Yuce & Khan, 2011), a feasible model may be derived. The semblance of the human tissue medium, compared with the link budget for the space conditions, has an additional impact on the propagation channel modelling. The body morphology attenuation may exhibit this impact. Therefore, considering the attenuation function f(α B ) which depends both on electric properties of the body surface and free space conditions, the electric field is expressed as follows: ( 1 0 ) with E 0 is the electric field radiated at a distance by the same antenna implemented above the human body surface. G T x , P T x and k = 2π/λ 0 are the antenna gain, the feeding power and the wave number in free space, respectively. Equation (10) possibly takes into account the near field if the distance d is close to λ (d ≈ λ).
From Equation (10), we can calculate the electric field |E| at a distance d as follows: At the same distance, the power P R x provided by a G R x is expressed as follows: It is then obvious to deduce the path gain which is equal to α B(dB) is the relevant human body attenuation element and it is necessary and required to optimize the T x /R x path gain. In other words, α B(dB/cm) represent the path loss considering the whole path attenuation including body absorption losses. Therefore, Equation (14) constitutes a novel on-body propagation model including both frequency and distance dependence, as well as the attenuation of the propagation medium on the body. Indeed, dependencies according to frequency, body permittivity and body conductivity will be demonstrated in the next subsection.

Parametric studies on the path gain model
In this section, the path gains for different parameters, permittivity, conductivity and frequency bands, are studied and calculated using our proposed channel propagation model. Figure 4 illustrates the variation of the path gain with the permittivity. There is a slight improvement in the link budget when permittivity is increased. The observed increase in path gain might be explained by increasing the permittivity leads to an increase in the inductive surface of the waves, and this favours the propagation of the creeping waves around the considered body part. The attenuation becomes relatively weaker (−80 dB) for ϵ r = 50 (skin or muscles). The conductivity of the human body has also a significant effect on the link budget as depicted in Figure 5. Results are almost the same as what we saw in the case of permittivity variation. Clear improvements are exhibited in the link budget when the conductivity increases. The increase in conductivity increases the conductance of the body surface, which induces significant attenuation of the path gain. To study the variations of the different parameters, permittivity, conductivity and frequency, we considered our proposed propagation model defined by Equation (2). For a frequency of 2.4 GHz, the influence of permittivity and conductivity seems negligible. Moreover, an enhancement in the link budget up to 20 dB is attained by changing the working frequency from 2.4 GHz, 5 GHz to 10 GHz for 80 cm T x /R x distance as shown in Figure 6. On the other hand, attenuation amplitude is 80 dB at 2.4 GHz, while at 10 GHz it is 115 dB. Note   that significant attenuation at 10 GHz requires a significant amount of signal to compensate attenuation due to the body tissue of the propagation support. The link budget is given by the received signal strength on the dBm scale. G T x = G R x = 0 dBi is supposed for all frequency ranges in order that the difference in the link budget is only caused by frequency, not by the geometry and the category of the T x /R x antennas. To sum up, the link budget increases as predicted by the increase in the frequency. The electrical characteristics of the skin, muscles and organs of the human body are very similar. There is also a layer of fat that can play a role, more or less important in the EMW propagation depending on its thickness, in the ISM band and beyond. For this reason, we will introduce the notion of experimental characterization of the on-body channel propagation.

Antenna design specifications
One of the most important tasks in the WBAN physical layer setup is the sensor's antenna design to guarantee high radiated power from the sensors to the receiving antenna. It needs an antenna with stable parameters close to the human body. Impedance matching circuits, radiation efficiency, direction with their solid angle of radiation pattern and polarization are the most significant factors to examine the body. In order to investigate the influence of the human body on antenna parameters, various design flows and simulation setups are considered. The proposed antenna geometry is arranged by a CST Microwave Studio (CST MWS) using the finite element method to optimize the electromagnetic performance of the antenna. In this work, the design and electromagnetic proprieties of the antenna, along the ISM 2.4 GHz band, make it easy to be matched to the physiological parameters of the human body. The antenna is excited by an inset microstrip line, and the radiation element is a square metal patch. Both the radiation patch and the feeding transmission line are implemented on the up-side of the FR4 substrate (ϵ r = 4,6), where the ground-plane (GND) covers the other down-side, which gives a simple fabrication process with a single layer. With the aim of covering the 2.4-GHz ISM band frequency range, the antenna resonant frequency ( f r ) is principally controlled by the size of square patch and the width of the inset-feed of the microstrip (W g ), which also controls the impedance bandwidth. In addition, to reduce the effects of creeping surface waves on the human body waist, which may decrease the efficiency of the antenna, a low substrate thickness is selected as follows: h s = 1.6 mm. The proposed wearable antenna structure is illustrated in Figure 7(a).
The first task in the patch antenna design is the evaluation of its largest dimension, such as the length (L) (Ammann, 1997). To calculate the preliminary value of the length side (L) at the chosen frequency ( f = 2.4 GHz), it is easy to use Equation (15).
In this paper, we use a FR4-epoxy substrate with a thickness of h s = 1.6 mm, an electric loss of tanΔ = 0.001 and a relativity permittivity ϵ r = 4.7 at a chosen frequency f = 2.4 GHz. The light velocity is c = 3.10 8 m/s, where λ 0 is the free space wavelength. The geometry of the designed antenna is a square patch with L = 29.1 mm. The effective permittivity (ϵ eff ) can be calculated using Equation (16) and then ϵ eff = 4.261.

Antenna optimization
The target of the optimization is the identification of the optimal geometry of the antenna with high matching characteristics of the T x /R x sensor systems giving the highest detection performance. To carry out this task, we introduce a simple modification on the antenna geometry and the dimensions as in Figure 7(b). According to Equations (15) and (16), the expression of the effective length (L eff ) is written as Equation (17). It becomes equal to 30.3 mm.
The fringe factor (ΔL) can be expressed by Equation (18) and can be evaluated to ΔL = 1.1 mm.
Using Equation (18), L = L eff − 2ΔL = 28.1 mm. This value is almost matching length L as in Equation (15). The inset line into the square patch is done by H = 0.822 × ΔL/2. Table 1 provides all values of these dimension.
In this work, we consider the communication channel explored here which considers two body antennas closely located on the human body waist. For simplification reasons, the human body is modelled as a homogeneous perfect electric conductor tube, as depicted in Figure 2(a). The suggested method for antenna characterization can also be implemented for these elliptical tube assumptions with high matching performance  T x /R x antennas. At 2.4 GHz, we consider that the PEC tube associated to the body antennas has a ϵ r = 50 and σ = 1.7 S/m. In addition, the elliptical tube radius (r) is chosen as 30 cm in order to model the human body waist.

Simulation results and analysis of antenna performance (a) Return loss
When the device under test (DUT) is an antenna, there are important parameters considered such as the return loss (RL), the insertion loss, the gain and the impedance; i.e. the RL is a measure of the power reflected from the antenna to the source line. It translates the impedance mismatch between the incident power P in and the reflected power P ref . It is usually expressed in dB as The RL simulation results of the designed wearable antenna are represented in Figure 8. At the chosen frequency 2.4 GHz, S 11 of the proposed antenna in the free space is equal to −21 dB with a good matching between the antenna and the feed microstrip line, besides a frequency range of 80 MHz measured at −10 dB. However, the existence of the human body has a significant impact on the RL, as shown in Figure 8, where S 11 becomes equal to −15 dB.

(a) Voltage standing wave ratio
The voltage standing wave ratio (VSWR) is also considered as an essential parameter to control. The VSWR can be just expressed in terms of RL by Equation (20). Figure 9 reveals that the obtained value of the VSWR is almost 1.12 at 2.4 GHz, which is very efficient in a free space case. On the other hand, when the propagation medium is the human body, the VSWR becomes 1.92 at 2.4 GHz, as illustrated in Figure 9. Obviously, the suggested antenna offers a good impedance matching with a nearly 100 MHz bandwidth (|S 11 | < 10 dB, VSWR < 1.9) along the ISM 2.4 GHz frequency range (International Telecommunication Union (ITU) radio regulations). These obtained results reveal also that the human body has a significant effect on the performance of the T x /R x communication channel. VSWR = −20 log 10 1 + 10 −RL/20 1 − 10 −RL/20 .

(a) Distribution of electromagnetic fields
Further studies on the distribution of the electric field on the body surface are also necessary to investigate the potential impact of using the human body model on antenna performance. In fact, due to the body-antenna significant electric interaction differences, the determination of the E-field setup in the human body is a topic of interest in all application categories, as well as medical and non-medical domains. Consequently, it is important to understand the distribution of E-fields and current densities in various parts of the body from both the general and medical points of view. Figure 10(a,b) depicts the distribution of the E-fields and the H-field, respectively, over the proposed patch antenna. According to these obtained results, the variation in the electromagnetic vectors confirms that the proposed antenna has linear polarization. Thus, the E-field and the H-field are very intense along the edge of the square patch antenna. According to the variation in the distribution fields, the designed WBAN antennas have vertical polarization. It helps us to effectively obtain maximal T x /R x power transfer and to optimize the link budget of the suggested channel model. It is common knowledge that the horizontal component of the electrical field (i.e. tangential to the human body surface) experiences a significant attenuation compared with the perpendicular component (i.e. perpendicular direction to the human body surface). Moreover, a normal polarization to the human body surface is favoured for on-body channel communication. Hence, just the vertical polarization is considered in our thesis. The intensity of the electromagnetic field over the proposed antenna presented in Figure 11 is very matched with WBAN applications where the maximum E-field radiation is obtained vertically at θ = 0°and 180° (Figure 11(a,c)) and minimally at the horizontal plan θ = 90°as shown in Figure 11(b) where the E-field is intense just on the edges of the patch. Thus, the maximal H-field radiation is obtained horizontally at θ = 0°and 180° (Figure 11(d,f)) and minimally at the vertical plan θ = 90° (Figure 11(e)).

(a) Radiation pattern
Investigating the electromagnetic radiation effect on the human body surface is an essential issue specifically for the medical applications because of possible short and long-term health consequences and for risk factor associated constraints. Based on simulation results, this is a huge challenge because the wavelength of signal λs is much smaller than the normal size of the human body, leading to the requirement of infinitesimal discretization of complex geometrical structures and then the need of computational resources due to the long execution times. For these reasons, in our numerical study, the Finite Integration Technique (FIT) technique associated with CST-MWS simulator is employed for the methodology. It consists of studying the feasibility of creeping-wave simulation of field absorption at Industrial, Scientific, and Medical (ISM) frequencies, for numerical methods applied to the electromagnetic domain and geometries in the order of a human body with a specific size.
The square patch antenna with fundamental mode excitation has a maximal directivity at the broadside direction of the patch where θ = 90°(vertical plane), and a minimum moveaway towards the endfire direction where θ = 0°(horizontal plan). Regarding the obtained simulation results represented by the 3D radiation pattern at 2.4 GHz, the directivity of the patch antenna placed in a free space is 6.37 dBi, as shown in Figure 12(a), and it decreases to 4.86 dBi when the antenna is fixed on the human body waist, as depicted in Figure 12(b). This overall radiation at 2.4 GHz in the broadside direction also confirms that this antenna is appropriate for on-body applications to improve the connections to the gateway device that is located away from the human body waist. Moreover, there is a significant attenuation of the radiation pattern for 90°< θ < 270°, which is due to the effect of the physiological parameters of the human body, such as the high tissue conductivity (σ = 1.7 S/m).
Moreover, the radiation is significantly attenuated in the endfire direction (90°< θ < 270°) which is due to the high human body tissue conductivity (σ = 1.7 S/m). This causes considerable absorption of the incident radiation power as it is evident from the diagram of the E-field variation, as depicted in Figure 12(c,d). Moreover, this radiation reveals good results about the half power beam width, where the solid angle is around 80°at −3 dB. According to the radiation patterns illustrated in Figure 12, the proposed antenna has high efficiency radiation where the front-to-back ratio is equal to 63.21%. In other words, the suggested antenna exhibits radiation of almost 64% of the transmitted power.
In WBAN systems, antenna positioning is critical in the simulation and measurement phases. In our suggested WBAN system, the sensor transmitting antenna (T x ) and the coordinator receiving antenna (R x ) are usually identical and positioned rigorously in the optimal positions to ensure a good propagation channel, and their electromagnetic performance is listed in Table 2.
In order to clarify this issue, three TRL calibration phases (Thru, Reflect, Line) before every manipulation of the DUT are necessary (Wang et al., 2020). Once these phases and tasks are achieved, the measured scattering parameters S 11 and S 21 and radiation patterns should be close to the simulated results in Figure 8 and Figure 12, respectively. To illustrate the body-antenna interaction concept, it is obvious from the electromagnetic numerical data that the radiation efficiency and the antenna gain are significantly reduced when mounted directly on the body. On the body surface, the radiation efficiency and gain are, respectively, 63% and 4.86 dBi compared to the free space values of 68% and 6.37 dBi. Finally, the proposed antennas are in good agreement with the use in WBAN applications.

Experimental study and measurement results
During our work, all measurements were taken in both multipath indoor environments: anechoic chamber with a radio communication platform as depicted in Figure 13. Thus, the optimal measurement points are obtained using two different antennas for the ISM band: monopole and patch. By calculating their path attenuations, we cannot only use it to find the accuracy of the received signal but also try to reduce it. This analysis also helps to select the best way to propagate the signal in a WBAN from a transmitter to the network gateway or from the T x to the R x body node. Vector signal generator Agilent technology N5182A and Signal Analyzer Agilent technology N9020A are used in this platform measurement. The T x antenna is designed at 2.4 GHz with S 11 < −25 dB. The dimensions are 25.7 × 7.5 mm and it is fabricated on a Printed Circuit Board (PCB) with 1 mm thickness. Whereas the R x patch antenna is also a 15.2 × 5.7 mm PCB designed for the whole 2.4 GHz ISM band with a reflection coefficient S 11 < −10 dB. Figure 14 presents the measured received power for the centre frequency 2.4 GHz where T x /R x antennas are placed directly on the body for the on-body communication link. The experimental study was exploited comparing free space and on-body communication modes. Signal attenuation is more important in the presence of the human body. Thus, the antenna performance was reduced as an important amount of the power is dissipated to body tissues. Figure 15 shows measurement campaigns in an anechoic chamber in order to acquire the received signal constituted only by the waves propagated on the body surface. It is noticed that in an anechoic chamber, the signal became less sensitive for long separating antenna's distance. The mean attenuation of the first arriving paths was noted to lie between −25 and −45 dBm.
The obtained measurement results were also examined as a function of frequency for T x /R x antennas. Figures 16 and 17 show these results when the T x /R x distance is 20 and 50 cm. x-axis denotes the frequency range (1-2600 MHz) and y-axis denotes the received power in dBm. Each figure reveals the characteristics of propagation response of the free space and the average intra-body of the two test persons. Results illustrated in these figures reveal also a significant attenuation of the received power when the frequency increases and the spatial variations of the received power for the on-body model at 2.4 GHz are particularly interesting. Besides, the wider T x /R x gap is observed, the faster,  the reduction in the received signal power. On the other hand, the intra-body channel follows the same pace of path loss with the same altering rhythm as the air channel as extended as the radiofrequency is between 1 and 2600 MHz. The antenna performance close to the body was assessed and was established to be very stable in relation to the free space performances both in temporal and spatial variations. It is important to note that the received signal attenuation is a significant parameter because the input signal must be fairly limited and weak in order to be safe for the human body. Table 3 shows a comparison of our WBAN model with other published works. For instance, some parameters are used such as (i) protocol is a system of rules that allows two or more entities  of a communications nodes, (ii) network stability indicates the regular period of the network activities which start at the moment of packets emission (T x ) until their complete reception (R x ) as illustrated in Figure 13 and (iii) delay is the necessary time it needs for packets to reach their target which is denoted as delay. WBAN may follow any physiological parameters of human body in real case and in real time, such as the electrical activities and the beat rate of the heart. Some emergency applications require a rigorous investigation of the delay to guarantee that data arrive instantaneously. These performance parameters and others will also be undertaken more in detail in our future works in the aim to cover a wide field of WBAN applications.

Conclusion
WBAN is still a very promise technology which has a wide potential to change fundamentally biomedical applications. This work provides a theoretical channel model for on-body propagation based on optimized design antenna sensors. The proposed formulation is an enhanced WBAN channel model supporting simultaneously the electromagnetic proprieties of free space and human body. It consists of combining the basic free space model characteristics with the existence of human body. To validate the theoretical channel propagation model, we followed a measurement procedure to extract the needed data for channel propagation prediction. Measurement campaigns were performed with T x /R x antennas operating at 2.4 GHz. Experiments were conducted on real human subjects in two different environments: indoor and anechoic chamber. Thus, we will deduce the real effects of each antenna type and measurement environment. According to electromagnetic propagation study of the existing channel models, we noticed that some factors must be involved in the modelling procedure from a physical layer point of view.

Disclosure statement
No potential conflict of interest was reported by the author(s). Mokhtar Harrabi is currently PhD student in computer science, ISITCOM, university of Sousse. He is a researcher member in NOCCS research Lab of ENISo, university of Sousse, Tunisia. He was an accomplished embedded system designer with expertise in mobile development and IoT systems. He spent over 18 years teaching at the university level and made significant contributions to the industrial field. Sadly, he has passed away.
Dr. Rafik Brahem is currently professor in computer engineering, ISITCOM, Sousse, Tunisia with more than 30 years of experience in academics. He is the director of PRINCE Research Lab, ISITCOM, Sousse, Tunisia. His area of expertise includes Learning Technologies, Wireless Communication (PHY/MAC), Neural network.