Achieving Energy Efficiency and Impact of SAR in a WBAN Through Optimal Placement of the Relay Node

Abstract

Wireless Body Area Network (WBAN) is an emerging and promising specialized area in Wireless networks that deals with crucial health-related datasets. Unlike other wireless networks, as this type of network deals with medical facts, losing it is fatal. WBAN is a highly constrained network. Reducing energy consumption and enhancing lifetime are the two most important challenges of WBANs. One way to achieve these is by deploying relay nodes optimally in WBANs. Generally, a relay node is placed at the midpoint of the line joining the source and the destination (D) nodes. We show that such simplistic deployment of the relay nodes is not the optimal deployment, which can hamper the overall lifetime of WBANs. In this paper, we have investigated the best location to deploy a relay node on a human body. We assume that an adaptive decode and forward relay node (R) can move linearly between the source (S) and the destination (D) nodes. Moreover, the assumption is that a relay node can be deployed linearly and that the body part of a human is a flat surface and hard. We have investigated the most energy-efficient data payload size based on the optimally placed relay location. The impact of such a deployment on different system parameters, such as distance (d), payload (L), modulation scheme, specific absorption rate, and an end to end outage ($\mathcal{O}$) are examined as well. It is observed that in every aspect optimal deployment of the relay node performs an important role to enhance the lifetime of wireless body area networks. Sometimes linear relay deployment is very difficult to implement, especially on the different body parts of the human body. To address these issues, we have examined the optimal region for the relay node based on a 3D non-linear system model. The paper provides guidance for both linear and non-linear relay deployment along with the optimal data payload size under various circumstances and also considered the impact of specific absorption rates on the human body.

Introduction

Wireless Body Area Network is one of the most promising in the health-care and assisted living paradigm. A WBAN node has a very tiny lithium (Li-ion) battery for both in, on, or around the body communication. The battery lifetime is small and cannot always be changed unobtrusively. This is one major obstacle to the large-scale deployment of WBAN. A WBAN architecture comprises sensors that send its data along with forwarding others to a centrally located point called the coordinator. As we have already mentioned these sensors are energy-constrained. Transmitting other parameters become tedious and energy-consuming. So, this job is being transferred to a special node called to relay that functions only by forwarding the health database from sensor to coordinator if required. Now, with the introduction of relays in the network, the major concern is where to deploy them to have efficient communication. Deployment of relay (R) nodes optimally between the source (S) and the destination (D) is one of the prime solutions to address the issue of maximizing the network lifetime.

Authors in [1] reduced the energy consumption of cooperating WBANs by using an optimal packet size. They did not consider any particular MAC protocol. Authors in [2] evaluated optimal packet sizes for different WBAN communication scenarios. They concluded that based on the acknowledgment policy used, packet size selection has to be considered for energy-efficient communication in WBAN. The overhead in terms of delay incurred due to the process of selection has not been mentioned. In [3], WBAN nodes incur variable path loss caused by shadowing as the nodes are mobile. In [4] authors constrained the outage probability and allocated optimal power to improve the energy efficiency of WBAN. Authors have compared the outage performance and energy efficiency of star topology in WBAN in [5]. They have considered both single and multi-hop relay cooperative schemes. Authors in [6] evaluated the impact of different components like beacon order, super-frame order, and back-off exponent on network performance in WBAN. In [7], the authors further studied the influence of the position of the coordinator, the formation of the tree for an association, and guaranteed time slot allocation on a network. The authors in[8] considered beacon collisions while forming a piconet by more than one device.

The WBAN is mainly used for the collection of biomedical data based on tiny sensors, the authors [9] discussed different aspects of information security in wireless body area networks, but they have ignored the energy optimization of WBAN networks. Similarly in [10] the authors discussed the performance issue of WBAN networks without considering the importance of SAR on body network. Here the authors [11] applied WBAN networks and different machine learning algorithm approaches to take the valuable information to predict the early information of the coronavirus.

All the cooperative network enhances reliability at the cost of data rate and energy consumption [12]. The connectivity-based mechanism adapts transmission power for each link and performs well for a distributed network [12]. In a star topological network like WBAN where the number of nodes is less, this mechanism is unfruitful. The overhead delay incurred in the PRR-based network in [13] is unsuitable for a healthcare network like WBAN. The RSSI-based prediction model of [14] is inappropriate for WBAN due to the high link breakage rate for mobility. The link-state which is defined by path loss, fading, and shadowing in WBAN changes dynamically due to motion in the human body. In [15], the authors performed experiments to measure RSSI variation for different postures. The authors varied the transmission power by comparing the average RSSI value with a predefined threshold determined from the experiments. Whereas in[16], using a game-theoretic approach for the issue to reach the Nash Equilibrium. This is unrealistic in WBAN as it requires extensive packet exchanges among themselves. Authors in [17] proposed a heuristic approach to enhance WBAN network lifetime by using a limited number of relay nodes. Since the positions of the relay nodes are fixed beforehand, the total cost of the network is not minimal. In [18], authors considered human mobility while determining stable links for a relay-based WBAN structure. Similar to [17], the number of relays is fixed and their positions are pre-determined and static. This is a major problem in optimizing the overall network cost. Zhang et al. in [19] developed a reliable routing protocol to enhance the network lifetime by dynamically calculating the coordinates of relay nodes. However, the cost is not minimized due to a fixed number of relay nodes. In [20], authors have proposed another framework to minimize the relaying cost while maximizing energy efficiency. They also have considered only stationary positions for relay nodes. Later, in [21] authors have considered different body postures for a realistic dynamic approach. In [22] the authors investigated the optimal location of the relay based on two-hop WBAN scenarios but they neglected the optimal payload and the thermal impact on the human body. The thermal impact of sensors on human body tissues cannot be neglected while designing for WBAN. The successful design of WBAN and its impact on the healthcare domain will be realistic only when we consider the impact of SAR on the human body. Suwansin [23, 24] represents some of the current state of the art on the investigations of SAR. These papers have studied the physical design of the antenna and the effect of EMF radiation on the human body. Whereas in [25] the authors work with TCDMA-based algorithm techniques to achieve better throughput and packet delivery ratio [26] without considering the effect of SAR on the human body.

In [24], the authors investigated the effect of different input powers of UWB antennas on SAR value. It is observed that tissues with high water content (like muscles and fat) get heated up faster. During packet transmission, the distance between the source from the observing point and transmission power significantly affects the rise in temperature causing the SAR level to cross the threshold. Improvised MAC and routing layer along with intelligently designed physical layer can reduce the impact of SAR and [27] focused on human body communication (HBC). They compared various models of the human body like the Muscle Model, 2/3 Muscle Model, and Layered human body model to analyze the Finite Difference Time Domain (FDTD). The authors in [28], propose a MAC protocol that opportunistically schedules transmission based on the predicted channel fluctuation.

Though a lot of research work is still on, there is no significant contribution to the impact of positioning relay node deployment in WBAN architecture.

The contribution of this paper is as follows.

• Find the best location of the relay for on-body WBANs based on the linear and non-linear model.

• Investigate the effect of different design parameters; like distance (d), payload (L), an end to end outage probability ($\mathcal{O}$), and different modulation schemes in the WBAN network.

• Analyze the process to mitigate the harmful effect of a specific absorption rate (SAR).

In this paper, we examined the optimal relay location for energy-efficient WBANs. We justified the reason for the importance of optimal relay deployment instead of the middle position between two nodes for the linear and non-linear system models. We have considered a non-cooperative system model and its works over well accepted Rayleigh fading channel as shown in Fig. 1.

Fig. 1.

A linear WBAN system model

The remaining paper is structured as follows. The background information on the proposed work is described in Sect. 2. The proposed energy power consumption model is explained in Sect. 3, which is followed by analytical and simulation results in Sect. 4. Lastly, Sect. 5 concludes the article.

Background Information

We have considered a linear topology of a single source (S) and destination (D) node pair in our WBAN network. Positioning the deployment of a relay (R) node in between these two is the primary objective of this paper. As in-vivo WBAN nodes are installed to the uneven structure of the human body, we have also examined the best location of the relay-based on a non-linear 3D coordinate topology model.

In the process, we have shown that the energy consumption of the sensor nodes is reduced and hence lifetime of the WBAN network is enhanced if the relay nodes are deployed correctly. As the sensors are battery-operated, the power model plays a vital role. We have analyzed the positive impact of the prime deployment of the relay, like enhancing the battery lifetime and minimizing the effect of harmful specific absorption rate (SAR), which deals with the effect of radiation on body tissues. In this section, we also touch upon the communication model, the law related to battery power usage, and specific absorption rate, a metric that is used to measure the effect of radiation on body tissues.

The Communication Model

As per IEEE 802.15.6 standard, devices can be operated in various frequency bands using various channels. On-body sensors can transmit data in frequency bands 13.5, 50, 400, 600, 900–950 MHz, 2.4, 3.1–10.6 GHz using CM3 Radio channel link to the coordinator across tissues in the human body. In this paper, CM3 is used in a 900–950 MHz frequency band for communication between the on-body sensor and the coordinator. We also note that the volume of traffic towards the coordinator in a WBAN is more than the volume of data in the opposite direction [29]. So, we have adopted half-duplex communication in this work. In a WBAN, sensors perform in a duty cycle mode comprising ACTIVE, SLEEP, and TRANSIENT modes [30]. Switching time from sleep to active mode is negligible as the duration of active mode [29, 31] is much more.

The ACTIVE mode time ($T_{\mathit{ac}}$) depends on the frame format of IEEE 802.15.6 as shown in Fig. 2 along with the size of payload (L) and ACK/NACK frames. We have assumed a stop and wait for ARQ protocol. Once a data packet is sent, the receiver will send either an ACK frame if it has received the packet successfully or it will send a NACK frame. After transmitting the sensed data, the transceiver switches to SLEEP mode for a duration of $T_{\mathit{sl}}$ time. So, the energy required to transmit a frame is the sum of ACTIVE mode power consumption ($P_{\mathit{ac}}$), transition period energy ($P_{\mathit{sl}}$) from both sleep mode to active mode and reverse, and SLEEP mode power consumption. We have neglected the very small value of SLEEP mode power consumption [32].

Fig. 2.

Standard physical and MAC frame format of 802.15.6

Linear and Non-Linear WBAN System Model

Primarily we have investigated the optimal relay location for a simple linear three-node model shown in Fig. 1, where the relay position ($d_{\mathit{SR}}$, 0) is located in the same line between the source (0, 0) and destination (d, 0). Based on this model, we have examined the best deployment location of the relay, determining optimal payload as well as analyzed SAR. Due to the uneven structure of the human body, linear relay node deployment is practically difficult to implement. We examined another topology to investigate a non-linear relay deployment position and investigate an optimal relay position based on the 3D coordinate system [21]. The relay position can change in three directions. Based on the coordinate system, position of source is taken as (0, 0, 0) and that of destination is (0, 0, d) as shown in Fig. 3. This indicates that source and destination nodes are placed d distance apart in terms of the depth of human tissue. Optimal relay location not only can maximize the network lifetime of WBAN but, it can also reduce the possibility of damaging body tissue by energy assimilation. In our non-linear system model, the X-axis represents the distance between the source (S) and the destination (D). The relay can be deployed based on the Y-axis from $-d$ to $+d$ distance. The distance of two nodes is calculated using Euclidean formula $\sqrt{(}{(x_1-x_2)}^2+{(y_1-y_2)}^2+{(z_1-z_2)}^2)$. X and Y-axis represent the coordinates of a 2D plane. Z coordinate represents the third dimension to represent the depth of the human body as shown in Figs. 3 and 4. This depth of the human body is correlated to body position and movement. We have examined the optimal location of the relay considering all three axes and provided guidelines to deploy the relay at a prime location to maximize WBAN lifetime and minimize SAR (Fig. 5).

Fig. 3.

3D representation of non linear WBAN system model

Fig. 4.

2D (X-Y axis) representation of non linear WBAN system model

Fig. 5.

Comparison of energy consumption in dual-hop and single-hop WBAN network

Energy Consumption Model

Based on the Physical and MAC protocol used in WBAN and the assumptions mentioned in the earlier section, the energy consumed to complete one cycle (including ACK/NAK) of frame transmission is given by 1.

$$\begin{array}{c}\hfill E_F=P_{\mathit{ac}}{T}_{\mathit{ac}}+2P_{\mathit{sl}}{T}_{\mathit{sl}}\end{array}$$ 1

Here factor 2 is taken for the transition period ($T_{\mathit{sl}}$) considering both the source and destination node.

The power consumed in ACTIVE mode is a combination of power amplifier power $P_{\mathit{PA}}$ and constant circuit power $P_C$ according to the following expression [30].

$$\begin{array}{c}\hfill P_{\mathit{ac}}=P_{\mathit{PA}}+P_C\end{array}$$ 2

The power amplifier power depends on the transmission power $P_t$, the ratio of peak to average ratio (PAR) and drain efficiency $\rho$ in the following manner.

$$\begin{array}{c}\hfill P_{\mathit{PA}}=\frac{\mathit{PAR}}{\rho }{P}_t\end{array}$$ 3

In the WBAN on-body network model, the path loss values and parameters depend upon the distance between nodes and the frequency band. In this paper, we have considered the frequency band to be 950 MHz [33]. Based on that the path loss model an be expressed as follows.

$$\begin{array}{c}\hfill PL[dB]=a\_losslgd+b\_loss+N\_loss\end{array}$$ 4

$a_{\mathit{loss}}$ and $b_{\mathit{loss}}$ are coefficients of linear fitting, d as the distance between $Tx-Rx$, and $N_{\mathit{loss}}$ denotes distributed variable with standard deviation ${\sigma }_N$ in dB.

The average transmission power $P_t$ in terms of the received signal power $P_r$ can be written as [30] Eq. (5).

$$\begin{array}{c}\hfill P_t=P_r{10}^{\frac{b\_loss+N\_loss}{10}}{d}^{\frac{a\_loss}{10}}\end{array}$$ 5

The received signal power ($P_r$) at the receiver end in Eq. (5) can calculated from the required energy per bit at receiver $(E_b)$ and bit rate ($R_b$).

$$\begin{array}{c}\hfill P_t=E_b{R}_b{10}^{\frac{b\_loss+N\_loss}{10}}{d}^{\frac{a\_loss}{10}}\end{array}$$ 6

In this paper, QPSK modulation is used as a standard for WBAN. The Bit Error Rate (BER) for QPSK at the receiver is expressed as

$$\begin{array}{c}\hfill P_b=Q(\sqrt{2{\gamma }_0})\end{array}$$ 7

here $Q(·)$ represented as the Gaussian Q function and ${\gamma }_0=E_b/N_0$ denoted as the signal to noise ratio (SNR). The requisite $E_b$ can represented as

$$\begin{array}{c}\hfill E_b=\left(\frac{N_0}{2}\right){\left[Q^{-1},(P_b)\right]}^2\end{array}$$ 8

here $Q^{-1}(.)$ is denoted for inverse Q function and $N_0/2$ represented as two-sided thermal noise power spectral density (PSD).

In a WBAN, the successful packet transfer rate depends on the signal-to-noise ratio (SNR). In the receiver end, output fading is one of the crucial causes of a change in SNR value. We have adopted a well-accepted Rayleigh fading model to take care of this aspect in our energy consumption model. Based on Rayleigh fading, outage probability is defined as the unit time when the average SNR ($\gamma$) falls below the minimum required SNR (${\gamma }_0$). The average SNR ($\gamma$) is represented as the ratio of required bit energy ($E_b$) and one-sided noise power spectral density ($N_0$). The outage probability can be expressed as follows [34].

$$\begin{array}{c}\hfill \mathcal{O}={\mathrm{P}}_{\mathrm{r}}[\gamma <{\gamma }_0]=1-exp\left(-,\frac{{\gamma }_0}{\overline{\gamma }}\right)\end{array}$$ 9

From 9, the required average SNR can be represented as, $\overline{\gamma }=-{\gamma }_0/ln(1-\mathcal{O})$. From 9, the required bit energy is as follows, where $N_f$ is the receiver noise figure.

$$\begin{array}{c}\hfill E_b\ge -\frac{N_0{N}_f{\gamma }_0}{ln(1-\mathcal{O})}\end{array}$$ 10

Frame Outage and Fading

In IEEE 802.15.6, a frame can be successfully transmitted if an outage does not occur while forwarding the data frame or acknowledgment frame. The frame outage probability $O_F$ is represented as follows [35] .

$$\begin{array}{c}\hfill {\mathcal{O}}_F=1-{(1-\mathcal{O})}^{L_s}\end{array}$$ 11

where $L_s=L_t/{log}_{2}M$ ($M=$ modulation order) symbols are sent in one transmission cycle per frame and $L_t$ denotes the total number of bits transmitted including ACK or NACK frame during one transmission cycle.

The end-to-end outage probability [35] along the entire relayed path is given by Eq. (12).

$$\begin{array}{c}\hfill O_R={\mathcal{O}}_F+(1-{\mathcal{O}}_F){\mathcal{O}}_F=2{\mathcal{O}}_F-{\mathcal{O}}_F^2,\end{array}$$ 12

Energy Consumption Per Bit Calculation

The power consumed in the active mode for transferring a frame through the relayed path is given by Eq. (13).

$$\begin{array}{c}\hfill \begin{array}{c}{\displaystyle P_{bt,relay}={\mathcal{O}}_F\left(\frac{P_{PA,SR}+P_{\mathit{TX}}+P_{\mathit{RX}}}{R_b}\right)}\hfill \\ +(1-{\mathcal{O}}_F)\hfill \\ {\displaystyle \times \left(\frac{P_{PA,SR}+P_{PA,RD}+2P_{\mathit{TX}}+2P_{\mathit{RX}}}{R_b}\right),}\hfill \end{array}\end{array}$$ 13

Where, $P_{PA,SR}$ and $P_{PA,RD}$ are active mode transmitting power for S-R link and R-D link receptively. We can derive its value from Eqs. 3 to 6, and the $P_{PA,SR}$ and $P_{PA,RD}$ are represented as follows.

$$\begin{array}{cc}\hfill P_{PA,SR}& =\frac{\mathit{PAR}}{\rho }{E}_b{R}_b{10}^{\frac{b\_loss+N\_loss}{10}}{d}_{\mathit{SR}}^{\frac{a\_loss}{10}}\hfill \end{array}$$ 14

$$\begin{array}{cc}\hfill P_{PA,RD}& =\frac{\mathit{PAR}}{\rho }{E}_b{R}_b{10}^{\frac{b\_loss+N\_loss}{10}}{d}_{\mathit{RD}}^{\frac{a\_loss}{10}}\hfill \end{array}$$ 15

Average energy consumption per bit for successful transmission over the relay path is denoted by Eq. (16).

$$\begin{array}{c}\hfill E_{\mathit{suc}}=\frac{1}{8L}\left[4,P_{\mathit{sl}},T_{\mathit{tl}},+,\frac{P_{bt,relay}{L}_t}{(1-O_R)}\right]\end{array}$$ 16

The first part of the equation represents the total required power of transition period energy, where four times transition period energy need for two hops (S-R and R-D) and the second part represents the data transmission period energy including retransmissions because of outage $(1/(1-O_R))$ (Tables 1, 2, 3). Next, to analyze the result of SAR, we have applied Eq. (16) to calculate the discharge current ($I_i$).

Table 1.

Notation

$P_b$	BER at receiver for OQPSK
b	Modulation level
B	Bandwidth
$N_0$	2(power spectral density of noise)
$N_f$	Received noise
PAR	Peak-to-average ratio
$\beta$	$\frac{\mathit{PAR}}{\rho }-1$
$\rho$	PA drain efficiency
$E_{\mathit{bit}}$	Energy consumed per bit
L	Size of payload
R	Data rate

Table 2.

Parameters of the Path Loss Model Covering Frequencies of 950-956 MHz for On-Body Communication [33]

Parameter	Hospital room	Anechoic chamber
$a\_loss$	15.5	28.8
$b\_loss$	5.38	23.5
${\delta }_N$	5.35	11.7

Table 3.

Parameters in a Room for On Body Communication link CM3 [30]

Parameter	Values
$f_c$	950 MHz
B	400 KHz
d	1m
$P_{\mathit{sl}}$	0.5 mW
$P_b$	${10}^{-5}$

The most energy efficient optimal position of relay $d_{\mathit{SR}}^{\ast }$ can achieve by evaluating $E_{\mathit{suc}}$ in 16 as a function ($f(d_{\mathit{SR}})$) of $d_{\mathit{SR}}$, where, $d_{\mathit{RD}}=d-d_{\mathit{SR}}$, and in the next step differentiate $E_{\mathit{suc}}$ with respect to $d_{\mathit{SR}}$.

The solution can represent as, $\partial E_{\mathit{suc}}/\partial d_{\mathit{SR}}{|}_{d_{\mathit{SR}}=d_{\mathit{SR}}^{\ast }}=0$, and the optimal location of relay expressed as,

$$\begin{array}{c}\hfill d_{\mathit{SR}}^{\ast }=\frac{d}{1+{(1-\mathcal{O})}^{-1/(a_{\mathit{loss}}/10-1)}}\end{array}$$ 17

Estimation of Battery Life

To compute the energy consumption in this paper, we have used the Peukert law to estimate the battery lifetime. Peukert’s Law tells us exactly how long a lead-acid battery of a node in the WBAN will last under any load. This Law expresses mathematically that as the rate of discharge increases, the available capacity of that battery decreases. The mathematical expression of this law is as follows.

$$\begin{array}{c}\hfill L=\frac{\mathit{aC}}{I^b}\end{array}$$ 18

L = battery lifetime in hours C = rated capacity at that discharge rate in Ampere-hours I = actual discharge current in Amperes, a is a constant ($a<=1$) and b is a constant ($b=1.051.8$).

Where discharge current,

$$\begin{array}{c}\hfill I=\frac{E_{\mathit{suc}}}{V\ast T}\end{array}$$ 19

The total time (T) to transmit a single bit is the addition of active mode time ($T_{\mathit{ac}}$) and transit mode ($T_{\mathit{sl}}$), here the required voltage (V) is considered as 3V and $E_{\mathit{suc}}$ represents the energy consumption per bit for successful transmission.

Specific Absorption Rate (SAR)

Our system is modeled to transmit the data from the sender to the receiver via the relay node positioned adaptively to minimize SAR so that less harm is caused to human health. SAR [36] is the rate of absorption of electromagnetic energy (W) per unit mass of tissue in units of watts per kilogram, measured in W/Kg as given in 20.

$$\begin{array}{c}\hfill SAR=dm\frac{d(dW)}{\mathit{dt}}\end{array}$$ 20

where, W = power in watts (W), m = mass in Kilograms (Kg) and t = time in seconds (sec). Table 4 represents the simulation parameters of SAR along with their corresponding values used.

$$\begin{array}{c}\hfill SAR=\frac{\sigma \mu \omega }{{\rho }_1\sqrt{{\sigma }^2+{ϵ}^2{\omega }^2}}{\left(\left(\frac{I_{i}dlsin\theta {exp}^{-\alpha \Re }}{4\pi }\right),\left(\frac{1}{{\Re }^2},+,\frac{mod\gamma }{\Re }\right)\right)}^2\end{array}$$ 21

SAR is calculated using Eq. (21) whose notation are given in Table 4.

Table 4.

Parameters of SAR [36]

Parameter	Notation	Values
$\sigma$	Body temperature	.866 $\frac{S}{m}$
	conductivity
$\mu$	Permeability	4$\pi X{10}^7$ $\frac{H}{m}$
$\omega$	Frequency band	950Mhz
${\rho }_1$	Density of tissue	1040 $\frac{\mathit{Kg}}{/}{m}^3$
$ϵ$	Relative permittivity	52.73 $\frac{F}{m}$
dl	Length of antenna	1m
$\Re$	Distance from source to observation point	0.1m
$\theta$	Angle between the observation point and xy plane	90

The attenuation constant is given by Eq. (22).

$$\begin{array}{c}\hfill \alpha =\omega \sqrt{\frac{\mu ϵ}{2}}{\left[\sqrt{1+{\left(\frac{\sigma }{\omega ϵ}\right)}^2},-,1\right]}^{\frac{1}{2}}\end{array}$$ 22

SAR is used to measure the amount of EMF radiation a human body absorbs from the heat and electromagnetic radiation generated from the WBAN sensors attached or implanted in the human body. International SAR measurement standards vary based on different body parts and countries. As specified by FCC, the most generally accepted SAR limit for public exposure to EMF radiation is 1.6 W/Kg. Here, discharge current $(I_i)$ is also derived from the total amount of energy dissipated for successful bit transmission. Similar to the previous section, the required voltage is 3v and the total time to transmit each bit is the addition of active mode and transition mode. Average energy consumption per bit for successful transmission over the relay path is derived later in Eq. (16).

Algorithm for Data Retransmission at the Node

In a WBAN IEEE 802.15.6 frame K, $Nod{e}_i$ have regular data with seq. no. j to be transmitted and retransmission is applied based on NACK;

Results and Discussions

This section gives an elaborate description of the setup used to simulate our model. It justifies the reason for taking the optimal relay location for a linear $S-R-D$ path instead of the middle position of the $S-D$ path in the linear WBAN system model. Based on this model, we compare the energy consumed per bit for communication through the relay (dual-hop) ($E_{\mathit{sucR}}$) with that for communication without the relay (single hop) ($E_{\mathit{suc}}$). The analytical results are established from the mathematical models, which are plotted using MATLAB, and the simulation result is shown using Castalia-3.2 (baseline MAC for WBAN) based on realistic CM3 communication link parameters.

We analyze the best location of the relay ($d_{\mathit{SR}}^{\ast }$) considering different distances between source and destination, where the nodes are situated on the body. We have also examined the optimal payload value for two positions (middle and optimal) of the relay node in non-linear WBAN. Next, we compare the energy consumed per payload bit for different end-to-end outage (O) values. We compare energy consumption per payload bit for the best position of the relay and the traditional middle position [35] between the $S-D$ path. We have investigated the energy consumed per successful bit for different payload (L) values and examined the effect of SAR for optimal relay location. In all cases, we can’t apply the linear WBAN topology model, especially to the human body. Finally, we have examined the optimal zone of a relay to enhance the lifetime based on a 3D coordinate system model.

Comparing Energy Consumption with and Without Relay

In Fig. 1 average energy consumed per bit for successful transmission has been plotted against the distance between source and destination. It is seen that when the length of the $S-R$ link is between 28-60 cm the energy consumption is better along the relayed path. The best performance is obtained when the relay position is 44 cm from the source in the $S-R$ link. We have determined the optimal location of the relay for different distances (d) between source and destination placed on the body where linear placement is possible. It is also observed that the lifetime of WBAN is enhanced by deploying the relay in the optimal location which is not the middle of the line joining S amd D [35].

Optimal Location Analysis for Different End to End Outage (O) Values

Figure 6, shows the plot of optimal relay location for different different end to end outage values. We have represented the distance between two nodes (S and D ) based on normalized form ($d_{\mathit{SR}}^{\ast }$/d).

Fig. 6.

Optimal relay location by varying the distance (d) between source and destination

The X-axis represents the energy consumption per bit (J). We have plotted the normalized distance on the $Y-axis$ keeping the path loss and other systems parameters constant varying the end-to-end outage (O) values. From Fig. 6 we can say that for all the cases, the optimal location (in terms of energy consumption) of the relay is not the middle position in the $S-D$ path. When the outage value is high (.0016 and .0014 ), the optimal location is closer to the source, whereas for lower values (.0012 and.001 ) the optimal location is closer to the middle position of the $S-D$ path. We can conclude from these results that the optimal location of the relay in WBAN is heavily dependent on the end-to-end outage between the source and the destination node.

Energy Efficiency Variation with Payload (L)

In Fig. 7, we have compared energy consumption per successful communication of bit varying payload for two different cases (relay in the middle and relay at the optimal location). The red line shows the variation of payload when the relay is placed in the middle position of the line joining the source and destination nodes for a given low end to end outage value (.001). Greenline signifies the payload variation when the relay is deployed in the proposed optimal location (44 cm). In this case, the distance between the two nodes is 1m. The result demonstrates some interesting facts. When the outage value is .001, the optimal payload value range is 38–47. The optimal relay position is more energy-efficient compared to the relay in the middle. This optimal payload depends on an end-to-end outage value; when the outage value is .0012 the optimal payload position shifts to 32–41 cm. Based on this, we achieve some guidelines of optimal data payload range to get the energy optimization instead of analysis with arbitrary data payload values [22, 37]

Fig. 7.

Compare middle relay position and optimal relay position with payload (L) variation

The battery life for two different positions of the relay is examined in Fig. 8. We have calculated the lifetime of the network based on peukert law. Primarily, the actual discharge current (I) is calculated from energy consumption per bit and applied to this value to evaluate the lifetime from Eq. (18). Battery capacity is considered as 560 mah [38]. Figure 8 shows that when the payload is less (within 78), the battery consumption of the relay node remains the same irrespective of its position. As the payload increases, the optimal location of the relay node outperforms the relay in the middle in terms of battery longevity. Thus, from Fig. 8 we can state that when relay node is deployed at the proposed optimal location, the battery lifetime is enhanced by 13% compared to conventional relay in the middle deployment. In Fig. 9, we analyze the optimal range of payload value based on total energy consumption (in joule) of the network. We can say that energy consumption heavily depends on the size of the payload. Figure 9 shows that at the payload value of 100, it takes maximum energy. This result is the same irrespective of the position of the relay node. However, if the payload is more than or less than 100, the energy consumption of optimal relay deployment position performs better compare to relay deployment in the middle position of the S-D path.

Fig. 8.

Compare battery lifetime with payload (L) variation for middle relay position and optimal relay position

Fig. 9.

Total consumed energy compare with middle relay position and optimal relay position with payload (L) variation

Comparison of SAR Value for Two Different Relay Positions Varying Payload (L)

In a WBAN, sensor nodes are generally placed on different parts of the human body or within the body. During packet transmission from one node to another node, an electric and magnetic field (EMF) radiation is generated, which may affect the human body cells or tissues. The radiofrequency absorption by human tissue per unit mass is defined by a specific absorption rate (SAR). A higher SAR value is not desirable as it may harm the healthy tissues. In most of the previous research work, [1, 38] of WBAN neglected the effect of SAR on the human body. In this paper, we have investigated the effect of the optimal location of the relay on SAR value. In Fig. 10 we have examined the SAR values for optimal relay location and relay in the middle deployment for different payloads. The SAR values mainly depend on the distance between the observation point and the source and the current required to transmit the signal. In this paper, we have taken a constant value of the distance between the observation point and the source. We have varied the payload values which reflect the required current values. Figure 10 reveals some interesting results. We can say that for low payload values (less than 55) the SAR value is almost the same for both deployments of relay nodes. If the payload value is more than 55, SAR values for optimal relay location (blue line) are much less compared to mid-position relay deployment (red line).

Fig. 10.

SAR compare for middle relay position and optimal relay position with payload (L) variation

Optimal Relay Position Analysis for Nonlinear 3D Coordinate System Model

In Figs.11 and 12 we have analyzed the most energy-efficient region for relay deployment based on 3D coordinate system, shown in Fig. 4. This WBAN 3D topology [21] supports, the sensor node can deploy in different body parts of a human body. We have analyzed the energy saving for optimal relay deployment on the human body, with respect to direct transmission(without a relay). We assumed that the relay can be deployed anywhere on the body. In Fig. 11 we have determined the optimal location based on the X-Y plane, whereas the Z-axis value is unchanged. From Fig. 11 we can say that the most optimal location region is the smaller boundaries region, where the X-axis value is from 35 cm to 54 cm, and the Y-axis value is from +30 cm to −24 cm. We have also examined that we can save near to 53 percent of energy if we placed the relay in that region. In Fig. 12, we have evaluated the optimal location based on the X-Z plane keeping Y fixed. The best boundary region location can save 75 percent of energy. This boundary region is given by Z values between +8 cm to −7 cm, and X values between 35 and 50 cm. Finally, in Fig. 13 we have represented the energy consumption per bit for a relay-based WBAN network based on a 3D coordinate system model. The bar value represents the energy consumption per bit. We can say that the prime location for relay deployment is near to the left side of the center position of the 2D plane which is shown as a blue color in Fig. 13. It is attractive to note that the optimal location of the relay is not in the center either in a 2D or a 3D plane.

Fig. 11.

Optimal relay position based on X-Y plane

Fig. 12.

Optimal relay position based on X-Z plane

Fig. 13.

Optimal relay position based on X-Y plane 3D view

Conclusions

In this paper, we have analyzed the optimal location of the relay in a WBAN in terms of energy consumption and lifetime. We have examined the energy consumed per successful transmission of a bit for different payload values, along with the energy efficiency of a relay-assisted WBAN compared to different distances between the source and destination, different modulation schemes, and different payloads. We have considered relay placement for 2-D and 3-D WBAN. From the experiments, we may conclude that the optimal position of the relay is not at the center of the source and destination nodes. If the relay is placed at the optimal location, the lifetime of the WBAN is greatly enhanced.

Biographies

B. Ghosh

was born in Birbhum, West Bengal, India. He received B.Tech. degree from Asansol Engineering College, WB, India in 2006, M.Tech. in Information Technology from NERIST, AP, India in 2008, and Ph.D (Engg) in Wireless Network from NIT Durgapur, WB, India in 2018. He is also holding the position of an Associate Professor in the Information Technology Department,Techno Main Saltlake, Kolkata, India. His research interests include on energy-efficient Wireless Network Communication and Wireless Body Area Networks.

S. Adhikary

received her B.Sc degree in Computer Science, M.Sc degree in Computer and Information Science, and M.Tech degree in Computer Science and Engineering from the University of Calcutta, West Bengal, India. She is currently working towards a Ph.D. degree at the Department of Information Technology, Jadavpur University, Kolkata, India. Her research interest focuses on energy-efficient Wireless Network Communication and Wireless Body Area Networks.

S. Chattopadhyay

received the B.Tech. and M.Tech degree in Computer Science and Engineering from the Indian Institute of Technology, Kharagpur, India, and the PhD. degree from the Jadavpur University, Kolkata, India. He is currently working as Professor at the Department of Information Technology, Jadavpur University, Kolkata, India. His research interest focuses on Algorithms, Wireless Networks, Network Security, Intelligent Computing, Bio-Informatics, Cloud Computing, and Distributed Computing.

S. Choudhury

received the B.Sc in Mathematics, B.Tech. degree in Computer Science and Engineering and M.Tech degree in Computer Science and Engineering from the University of Calcutta, Kolkata, India, and the PhD. degree from the Jadavpur University, Kolkata, India. He is currently working as Professor at the Department of Computer Science and Engineering, Calcutta University, Kolkata, India. His research interest focuses on Computer Networking, Distributed Computing, Bio-Informatics.

Funding

The authors have not disclosed any funding.

Data Availability

Enquiries about data availability should be directed to the authors.

Declarations

Conflict of interest

We have plotted the figures using MATLAB, and the simulation result is shown using Castalia-3.2 (baseline MAC for WBAN) based on realistic CM3 communication link parameters.However, except for the payload value 100, the energy consumption of the optimal relay deployment position performs much better than relay deployment in the middle position of the S-D path