An Improved Routing Approach for Enhancing QoS Performance for D2D Communication in B5G Networks

Abstract

Device-to-device (D2D) communication is one of the eminent promising technologies in Beyond Fifth Generation (B5G) wireless networks. It promises high data rates and ubiquitous coverage with low latency, energy, and spectral efficiency among peer-to-peer users. These advantages enable D2D communication to be fully realized in a multi-hop communication scenario. However, to ideally implement multi-hop D2D communication networks, the routing aspect should be thoroughly addressed since a multi-hop network can perform worse than a conventional mobile system if wrong routing decisions are made without proper mechanisms. Thus, routing in multi-hop networks needs to consider device mobility, battery, link quality, and fairness, which issues do not exist in orthodox cellular networking. Therefore, this paper proposed a mobility, battery, link quality, and contention window size-aware routing (MBLCR) approach to boost the overall network performance. In addition, a multicriteria decision-making (MCDM) method is applied to the relay devices for optimal path establishment, which provides weights according to the evaluated values of the devices. Extensive simulation results under various device speed scenarios show the advantages of the MBLCR compared to conventional algorithms in terms of throughput, packet delivery ratio, latency, and energy efficiency.

1. Introduction

The demand for the internet is growing exponentially as the world is moving towards the digitalization era. Billions of physical devices around the world are connected to the internet, collecting and sharing data information. This is known as the Internet of Things (IoT) [1,2], which facilitates smart homes, smart cities, smart grid, and industrial automation, as well as healthcare, environmental, and transportation monitoring. IoT transfers enormous data information over a network independent of human action. Future 5G networks will handle these data and information by providing greater connectivity, higher data rate, ultra-low latency, greater energy efficiency, and greater spectral efficiency [3,4]. However, 5G networks utilize the mmWave radio frequency range for data transmission, and, thereby, the signal transmission range is limited to only a few hundred meters [5,6]. In this case, D2D communication is a prevalent emerging technology that extends network coverage and provides seamless connectivity for peer-to-peer users in 5G networks [7,8]. This technology provides opportunistic spectrum utilization, enhanced spectral efficiency, high data rate, and reduced network congestion [9,10]. In D2D communication, users in close proximity establish the network infrastructure partially with or without the involvement of cellular base stations. The source device uses the relay devices to forward data packets to the destination device in D2D communication. Meanwhile, in an orthodox mobile network, all device users are directly connected to the base station through a wireless radio medium [11,12]. Consequently, the performance of D2D technology may become worse if the relay devices are not selected via a proper mechanism. Therefore, data packet routing is a significant task in D2D communication over IoT-based 5G networks [13,14]. The application of a D2D communication scenario over IoT-based 5G networks is depicted in Figure 1.

Researchers around the globe have proposed various routing mechanisms to provide better QoS performance for D2D communication [15,16,17,18]. Mobility, battery, link quality status, and contention window size of relay devices are the main challenges that occur during the optimal route selection. First, the network topology is changing in an unpredictable manner due to the devices moving in the network from one place to another. It adversely affects the network stability, leading to the packets being dropped in the network. Second, all of the devices operate with limited battery capacity; the system is dead once the battery is down. Moreover, the link quality status of the established routes conversely affects the device movement within the network. Third, data packet traffic congestion occurs in the relay devices, owing to the simultaneous data packets being sent from the source to the same channel. All of the challenges mentioned above should be considered during the optimal route selection in order to enhance the QoS performance of D2D communication technology. Figure 2 illustrates six relay devices (2, 3, 4, 5, 6, and 7) between the source (1) and destination (8). Each relay device has different mobility, battery, link quality status, and contention window size status. Thus, selecting the best relay devices in terms of robustness and optimal route reliability is crucial.

Contribution

Based on the challenges mentioned above, this paper proposed a mobility, battery capacity, link quality status, and contention window size-aware optimal routing (MBLCR) to enhance the QoS performance of D2D communication over an IoT-based 5G network. The proposed MBLCR aims to balance battery consumption among the devices and select the optimal path amongst the multiple paths by taking advantage of the MCDM technique. The contributions of this paper are as follows:

(i)

In the route computational procedure, evaluate the mobility, battery, link quality status, and contention window size of the relay devices from the source to the destination devices.

(ii)

Assign individual weights to the relay devices according to the evaluated value by exploiting multiple criteria of the decision-making technique.

(iii)

Compare the performance results of various QoS metrics of the MBLCR routing scheme with the conventional MP-OLSRv2, MEQSA-OLSRv2, and EMBLR routing scheme using over different device speeds.

This paper is organized as follows. Section 2 presents existing works on routing schemes in wireless networks. Section 3 explains the mathematical model of the proposed MBLCR routing approach. Section 4 shows the results and discussion of the routing schemes. Section 5 concludes the paper with possible extensions and direction for future research.

2. Related Works

This section discusses recent works on routing approaches in wireless networks involving network stability, energy resource, link quality, and device fairness. D2D communication routing in 5G networks has been studied by several authors [19,20,21,22,23,24], in which some initiatives have been found to be sufficient for the development of D2D communication in IoT-based 5G networks. Optimized link-state routing (OLSR) is a well-known routing protocol widely used in wireless networks. All of the network nodes have primary routing information through periodical topological message exchange, such as HELLO and topology control (TC) messages [25]. The network nodes also have network routing information, which is used when needed. Maintaining the network routing information via the exchange of topological messages in the network increases the flooding of control overhead messages, congestion, and energy consumption, resulting in network performance degradation. In order to minimize the flooding of control overhead messages in the network, OLSR exploits an efficient multi-point relay (MPR) mechanism in the route selection process, which is the main function of the OLSR protocol. OLSR has been modified and enhanced over time by several researchers since its inception. Other than that, ref. [26] studied an efficient mechanism for the selection of MPRs, which is aimed at improving network performance by minimizing the number of signaling overheads through a simple modification in the OLSR scheme. A new degree of database information from a neighbor’s device is employed for the selection of MPRs, in which the exchange of topological message is extended to three hops. The proposed mechanism reduces the control signaling overhead in the network compared to the OLSR protocol. Another work carried out in [27] introduced parallel disjointed multipath-based OLSR (DMP_EOLSR) algorithm for wireless networks. The proposed protocol utilizes the living time of nodes and living time of links based on the nodes’ energy consumption and movement mode. Meanwhile, an iterative algorithm based on a modified Dijkstra’s algorithm is employed to discover several node-disjointed or link-disjointed paths in the multi-path selection and recovery process. The simulation results prove that the DMP_EOLSR improves network stability and transmission efficiency by mitigating the number of interrupted network nodes and links. In [28], a traffic- and energy-aware routing (TEAR) scheme to optimize the stability period in wireless sensor networks was explored. The proposed scheme exploits random initial energy and random disparity (multi-level traffic heterogeneity) of the sensor nodes to model a realistic clustering mechanism suitable for heterogeneous wireless networks. The TEAR scheme enhances the lifetime and stability of networks in a multi-heterogeneous scenario.

Meanwhile, a novel relay-based multi-hop routing to improve the energy efficacy of D2D communication networks is introduced in [29]. The proposed routing approach improves the energy efficacy of D2D communication networks with a combinatorial optimization problem to attain a plausible solution. The result demonstrates that the proposed heuristic routing improves energy efficiency and network throughput and mitigates the computational cost more than other existing techniques. Next, ref. [30] explored an energy-conscious, dual-path geographic routing (EDGR) approach to provide efficient route recovery from the routing holes. The proposed protocol exploits the nodes’ geographical location, remaining energy, and energy-consumption features in terms of route selection. It also dynamically utilizes two node-disjoint anchor lists, which are passing through the two sides of the routing holes to shorten the routing path and load balancing in the network. The EDGR improves energy efficiency, network lifetime, and end-to-end delay, compared to other existing geographical routing approaches, in a heterogeneous network scenario.

The authors in [31] presented a path-aware geographic perimeter stateless routing (PA-GPSR) to improve wireless networks’ performance. The proposed scheme uses node information from a neighbor’s table for an optimal path selection and avoids neighbors that discover local maximum on its route for each destination. Moreover, it mitigates the chance of link failure among the nodes owing to node movement in the network. Compared to existing routing schemes, the PA-GPSR scheme improves packet delivery ratio, end-to-end delay, and network yield. In [32], an energy-efficient routing protocol was examined to select control nodes with network functionality for multi-tasking software-defined wireless sensor networks. The proposed algorithm selects the control nodes while considering the nodes’ remaining energy and transmission distance with the NP-hard problem. An efficient non-linear weight particle swarm optimization algorithm is employed to maintain a cluster structure. It decreases the data transmission range and improves the network’s energy efficiency for handling the NP-hard problem. The simulation results demonstrate that the presented algorithm prolongs the network’s lifetime in comparison to other algorithms under various scenarios. In [33], the authors found the least common multiple based routing (LCMR) to balance the traffic load among multiple routes for wireless networks. The proposed routing scheme distributes the data packets on the paths with consideration of the routing time. The data packet distributed mechanism balances the load among the path by minimizing the overall routing time for data packet transmission. The best selected path among the multiple paths is based on the number of hops between the source–destination node pairs.

Meanwhile, a multipath OLSR version 2 (MP-OLSRv2) was introduced to discover the multiple disjoint routes in dynamic and high-loaded wireless networks. It utilizes the topological network information with the multipath Dijkstra’s algorithm to discover multiple disjoint paths between the source and destination nodes. The MP-OLSRv2 avoids the disjoint paths in which a single-link breakage occurs, and data packets are transmitted in parallel to increase the network’s throughput. On the other hand, Ref. [34], presented a hybrid multipath energy consumption and QoS-aware OLSRv2 (MEQSA-OLSRv2) scheme to maintain the tradeoff between the energy consumption and quality of service (QoS) performance of wireless networks. The MEQSA-OLSRv2 utilizes the node rank mechanism according to the multiple-criteria (energy consumption and QoS) value nodes for the selection of an optimal route. Moreover, a link cost assessment function is used to forward the data packets over the multiple disjoint paths during the multipath route selection mechanism. An energy-efficient MPR selection metric is exploited to minimize the flooding of topological messages in the network and extend the MPR’s lifetime. Overall, the presented routing scheme is enhanced through the QoS performance network in the MANET-WSN convergence scenario of the IoT network.

Based on the literature reviewed so far, it can be seen that there is no study combining multiple device metrics into a single metric for an optimal path selection mechanism. Most research has focused only on two or three device metrics for route selection mechanism. To the best of the authors’ knowledge, no research has been reported that combines all of the metric parameters (mobility, battery, link quality status, and contention window size of devices) into an individual route computational metric in D2D communication over IoT-based 5G networks. This paper aims to combine all of the parameter metrics into a single metric for an optimal selection mechanism by employing the MCDM technique to assess network performance. The structure and functional framework of the MBLCR scheme are illustrated in Figure 3. The proposed framework addresses the challenges of this research, followed by a few proposed solutions in order to attain the research objectives. Moreover, the MBLCR exploits the functionalities of other existing conventional routing protocols for the route computational mechanism. In short, the MBLCR framework is modified and improved with the integration of mobility, battery, link quality status, and contention window size of devices by employing the MCDM technique in an optimal path selection among the multiple paths.

3. System Model

D2D communication over IoT-based 5G networks can be modeled as a graph $G(N,L)$, in which $N$ refers to the number of devices and $L$ refers to the number of wireless links among the devices. $l(a,b)\in L$ denotes the link among two devices, i.e., $a\in N$ and $b\in N$. The transmission of data packets is feasible when two neighbors’ devices share links with each other, whereas relay devices are utilized when the two devices disassociate from each other. The proposed MBLCR scheme considers the mobility, battery, link quality status, and contention window size of relay devices in the efficient and optimal path selection from the source to destination devices. The following subsections describe the evaluation of relay devices’ mobility, battery, link quality status, and contention window size.

3.1. Device Energy Consumption Evaluation

Mobile devices have limited battery power for the vital operation of data packet transmission. Once the battery of the device is exhausted, the network will be dead. Therefore, battery resources of relay devices play a vital role in the establishment of a reliable and optimal D2D communication over IoT-based 5G networks. The energy consumption of relay devices is evaluated by exploiting the energy consumption model of wireless communication [35,36]. The battery of relay devices is mainly consumed during the transmitting and receiving process for data packets transmitted in the network. Transmission distance of signal $(\lambda )$ from the transmitter $(Tx)$ to the receiver $(Rx)$ device is evaluated using the free space model (${\lambda }^2$ power loss) and multipath fading model (${\lambda }^4$ power loss). Consequently, a free space model is applied when the $\lambda$ value is smaller than the cut-off value ($d_0$). At same time, the multipath fading model is applied when the $\lambda$ value is higher than $d_0$. Meanwhile, additional device energy is consumed in circuit operation time, such as transmission circuits and power amplification losses. Thus, when there is $l$ number of data packet information in transmission in the wireless network, the energy consumed by the device is computed as bellows:

$$E_{Tx}(l,\lambda )=\{\begin{array}{cc}l{E}_{Tx-elec}+l{\epsilon }_{f_s}{\lambda }^2& \lambda <d_0\\ l{E}_{Tx-elec}+l{\epsilon }_{amp}{\lambda }^4& \lambda >d_0\end{array}$$

(1)

$$E_{Rx}(k)=l{E}_{Rx-elec},$$

(2)

$E_{Tx-elec}$ and $E_{Rx-elec}$ are referred to as the per-bit energy of the devices consumed by the transmitter and receiver circuits. Moreover, ${\epsilon }_{f_s}$ and ${\epsilon }_{amp}$ are defined as the power amplification components of the free space and multipath radio models. In addition, the cut-off distance value $d_0$ is calculated as $d_0=\sqrt{{\epsilon }_{f_s}/{\epsilon }_{amp}}$. The Equations (1) and (2) mentioned above evaluate the energy consumption of devices when $l$ number of data packet information is in transmission in the wireless network. However, the MBLCR approach also considers the maximum battery $(R{B}_{\max })$ and residual battery $(RB)$ of the relay devices for the selection of a robust and reliable route. The $R{B}_{\max }$ and $RB$ values are evaluated by utilizing the linear battery model [37]. Therefore, the energy consumption of relay device $c$ from the source to the destination is computed as bellows:

$$D{R}_{(a,b)}^c=\frac{R{B}_c(l)}{R{B}_{\max }(l)\times \{E_{Tx}^c(l,\lambda )+E_{Rx}^c(l)\}}$$

(3)

3.2. Mobility of Relay Devices Evaluation

Mobile device movement is unpredictable in a wireless network, which frequently changes the topology of the network, adversely affecting the route and network stability. Therefore, device mobility is considered in the MBCLR approach during the route computational procedure in order to improve network performance. Device mobility pattern is predicated by exploiting the well-known random waypoint model (RWP), which is widely used in wireless networks owing to its effectiveness and better QoS performance [38]. The RWP model predicts device mobility behavior using various metrics, such as velocity, moving time, pause time, and distance between devices. The RWP model consistently predicts device mobility and, meanwhile, selects a uniformly and randomly destination point (“waypoint”) according to the aforementioned metrics. Once the devices reach their selected waypoint, they stop for a specific time duration (pause time), and the procedure is repeated for all network devices. The RWP configures the device maximum and minimum velocities ($v_{\max }$ and $v_{\min }$) along with the pause and moving time duration of the device. The device mobility distribution $f_{(a,b)}(v)$ in the network can be computed as follows:

$$f_{(a,b)}(v)=p_{mov}.\frac{1}{vIn(v_{\max }/v_{\min })}+p_{pause}.\delta (v)$$

(4)

Here, v refers to the velocity function range that varies $v\in [v_{\max }-v_{\min }]$; $p_{pause}$ and $p_{mov}=(1-p_{pause})$ refer to the device’s probabilities during the pause and the moving stage, respectively; and $\delta (v)$ refers to the Dirac delta function, which depends upon the device’s velocity, and it ranges between 0 and 1. When the device velocity is maximal, $\delta (v)$ is updated to 0, and while the device velocity is minimal, $\delta (v)$ becomes 1. The device pause time is considered to be $t_p\ge 0$, and the pause time $p_{pause}$ probability value is computed as follows:

$$p_{pause}=\frac{t_p}{t_p+E[D]\frac{\mathrm{In}(v_{\max }/v_{\min })}{(v_{\max }-v_{\min })}}$$

(5)

Here, $E[D]$ refers to the expected distance covered by the device movement in the whole network. When a device initiates from a pause state $p_{pause}$, the pause time of the device is set to be $t_p$, while the device velocity is set to be in a range of $v\in [v_{\max }-v_{\min }]$ when the device starts from the moving state $p_{mov}$. In other words, $p_{pause}$ is the device probability in the pause state, whereas $t_p$ indicates the time taken by the device in a pause state. Therefore, the device mobility in the network is computed as follows:

$$\mathrm{Mobility}=\underset{v\in [v_{\max }-v_{\min }]}{\min }{\displaystyle \sum _{(a,b)\in N}{f}_{(a,b)}(v)}$$

(6)

Established route stability is depending upon the relay devices between the source and the destination devices. Moreover, a route goes unstable when the constituent relay device mobility is high. Therefore, the MBLCR approach selects relay devices with less mobility in the optimal route in order to maintain the route and network stability for D2D communication over IoT-based 5G networks.

3.3. Link Quality Evaluation of Relay Devices

An optimal path selection that guarantees the successful and efficient data transmission routing mainly depends upon the link quality status of the relay devices between the source and destination. An Expected Number of Transmissions (ETX) evaluates the link quality status of relay devices, which is the number of transmissions and retransmissions required for the successful delivery of data packets over the dedicated link to the destination device [39,40]. Meanwhile, link failure occurs among relay devices in the established path owing to frequent changes in the network topology. Because of this, topological packets increase in the network in order to find a new path. It induces a storm of route request (RREQ) packets in the network, which significantly degrades the overall network performance. Hence, the main objective of the ETX metric is to minimize the flooding of RREQ packets during the route discovery procedure. With these factors, the packet delivery ratio and the window size of the packets w (window size is set to be ten times the probe message interval) are utilized where the size of the window time scale is equally discretized $[t_{n-1},t_n]$. The data packets delivered and received over the dedicated link through the relay device c is defined as ${\lambda }_c$ and ${\mu }_c$, respectively, and their values are evaluated through the probe message sent prior to the data packet transmission. Every device broadcasts the probe message to its neighbor’s device for an interval of $\xi$ seconds, i.e., $w>\xi$, and the neighbor’s device records the number of received probe message, i.e., $n_w$, for a period of w second. Thus, the probability of successful data packet delivery by the source device to its neighbor’s device is computed as follows:

$${\lambda }_c=\frac{n_w}{w/\xi }$$

(7)

Moreover, the data packet transmission probability of the relay device c from at least one of its parent devices is $p_r=1-{\displaystyle {\prod }_{c>a}(1-{\mu }_a(t_n){\lambda }_c(t_n))}$. Similarly, the data packet transmission probability from the relay device c to at least one of its child devices is $p_d=1-{\displaystyle {\prod }_{b>c}(1-{\mu }_c(t_n){\lambda }_b(t_n))}$. Therefore, the ETX required to deliver data packet transmission by the device c at a time interval n is calculated as follows:

$$ETX=\frac{d_{(a,b)}}{\left(1-{\displaystyle {\prod }_{c>a}(1-{\mu }_a(t_n){\lambda }_c(t_n))}\right)\left(1-{\displaystyle {\prod }_{b>c}(1-{\mu }_c(t_n){\lambda }_b(t_n))}\right)}\frac{S}{B},t\in [t_n,t_{n+1}]$$

(8)

Here, $d_{(a,b)}$ refers to the distance between devices a and b; S defines as the size of the data packets; and B refers to the data packet transmission rate. Based on the information attached to the probe messages, each device evaluates the forward and reverse delivery ratios denoted as $p_d(t_n)$ and $p_r(t_n)$, at the previous time slot of the window, i.e., $w_{n-1}$, providing $r(t_n)=Count(t_{n-1,}{t}_n)/(w/\xi )$. Here, $Count(t_{n-1,}{t}_n)$ defines the number of probe messages received at the window, whereas $w(t_n-t_{n-1}=w_{n-1}=w)$ and $w/\xi$ refer to the summation of transmitted probe messages. Therefore, the link quality estimation based on the ETX value among the devices minimizes the link failure probability owing to the flooding impact of the RREQ messages in the network.

3.4. Contention Window Size of Relay Devices Evaluation

The contention window (CW) is defined as the waiting time frame of the devices for accessing subsequent data transmission after a collision occurrence. The waiting time frame is known as back-off time, which ranges between $\left[C_{min},C_{max}\right]$, and it increases according to the binary exponential back-off, where $C_{min}$ and $C_{max}$ are the minimum and maximum CW size of the devices, respectively. When more than one device transmits data packets simultaneously, there will be a collision between them, and the CW size will increase exponentially. The average channel access to the devices is highly dependent upon the back-off time and the CW size [41]. The device selected in the route, which has high back-off time, will experience a significant delay in the D2D network. Consequently, the CW size of the relay devices plays a vital role in the optimal route computational procedure. Therefore, selecting the relay devices in the optimum route with a smaller CW size is imperative. When the data packet transmission is successful, the CW size goes to a minimum, i.e., $CW=C_{min}$. In contrast, when a collision occurs during the data packet transmission, the CW size increases until maximum such that $CW=C_{max}$ and the binary exponential back-off time becomes two times. $\Phi$ defines the data packets that are successfully transmitted to the destination, which range varies between 0 and 1. When the data packets are successfully received at the destination, it shows 1, whereas it shows 0 when there is a collision that occurs at the time of data packet transmission. Therefore, the CW size is computed as follows:

$$CW=\{\begin{array}{cc}\min (CW\times 2,C{W}_{\max })& \Phi =0\\ C{W}_{\min }& \Phi =1\end{array}$$

(9)

Suppose $r$ denotes the total number of re-transmissions required for the successful delivery of data packets to the destination device, then the source device randomly selects a back-off time ranging from $[C_{min},{\mathrm{CW}}_s^c]$ when transmitting the data packets. The ${\mathrm{CW}}_s^c$ defines the CW size of device c at the $s^{th}$ transmission attempt, and it can be calculated according to the exponential back-off algorithm as follows:

$$C{W}_s^c=\{\begin{array}{ccc}{\sigma }^{s}C{W}_{\min },& fors\in [0,r^{*}-1]& ifr>r^{*}\\ C{W}_{\max },& fors\in [r^{*},r]& ifr>r^{*}\\ {\sigma }^{s}C{W}_{\min },& fors\in [0,r]& ifr\le r^{*}\end{array}$$

(10)

where $r^{*}=[{\log }_{\sigma }\frac{C{W}_{\max }}{C{W}_{\min }}]$ and $\sigma$ are referred to as the back-off increment factor. Therefore, the relay devices are prioritized based on the contention window size. In the situation when the CW range of a relay device is close to $C_{min}$, it will get higher priority. In contrast, if the CW value is high, it will have a lower priority in the route computational procedure. In this way, data collisions among the devices decrease, and packet drops reduce during the data packet transmission in the network.

3.5. Multi-Criteria Decision-Making Technique

A multi-criteria decision-making (MCDM) technique is employed to select the optimal route among the multiple routes according to the evaluated parameter values (mobility, battery, link quality status, and contention window size) of the relay devices. It aggregates all the relay device parameter values into an individual metric and weights the relay devices according to the evaluated parameter values. The source device periodically monitors and estimates the relay devices’ parameter values by broadcasting topological control messages at the time of path discovery and topology sensing phase. The MCDM value of relay device $c$ for the route selection process is computed as follows:

$$MCDM=\{W_{EC}\times E{C}_{(a,b)}^c(t)+W_{mob}\times Mo{b}_{(a,b)}^c(t)+W_{LQ}\times L{Q}_{(a,b)}^c(t)+W_{CW}\times C{W}_{(a,b)}^c(t)\}$$

(11)

where $W_{EC}$, $W_{mob}$, $W_{LQ}$, and $W_{CW}$ are the weights provided based on the evaluated parameter value which ranges between “0” and “1”. The relay devices which possess a higher MCDM value have a higher probability in the route computational process. In contrast, the relay device that is not selected in the route computational mechanism has a lower MCDM value. Thus, the MCDM technique exploits these parameters’ metric values in a reliable and optimal path selection among the multiple paths from the source to the destination devices. A flow chart of optimal path selection mechanism of the MBLCR approach is depicted in Figure 4.

3.6. Simulation Setup

Extensive simulations were executed to evaluate and assess the performance of the routing schemes under various device speed scenarios. The network deploys 49 number of devices randomly distributed in a network area of 500 m × 500 m. The device movement in the network based on the RWP mobility model and device speed varies between 10m/s and 60 m/s. The constant bit rate (CBR) universal datagram protocol (UDP) traffic was set to generate 20 packets per second with a packet size of 512 bps, and the wireless channel capacity was set to be 11 Mbps. The wireless channel frequency was set to be 2.4 GHz, and the radio transmission distance was set to be about 270 m based on the Wi-Fi parameter setting. A two-ray-ground path loss model with the shadowing means of 4.0 dB was selected and the transmission power was set to be 31.623 mW. The simulations were run for 200 s, with every simulation being run for an iteration of 200 times to calculate the average value of the results. The simulation parameters and values are described in

Table 1. Simulation parameter values.

Simulation Parameters	Values
Routing approaches	MBLCR, MEQSA-OLSRv2, and MP-OLSRv2
Simulation time	200 s with 200 iterations
Application traffic	Constant Bit Rate
Application packet size	512 bytes
Initial battery level	3600 mAh
Generic energy model	$P_{Transmission}=1300\mathrm{mW}\mathrm{and}{P}_{\mathrm{Re}cieve}=900\mathrm{mW}$
Battery model	Linear battery model
Signal transmission power	Pt = 31.623 mW
Transmission range	270 m
Mobility	RWP model min speed 10m/s, max speed 60 m/s
Simulation area	$500 \mathrm{m}\times 500 \mathrm{m}$
Application packet size	512 bytes
Data rate	11 Mbps
Wireless channel frequency	2.4 GHz
Path loss model	Two ray ground

.

3.7. Performance Evaluation Metrics

The objective of the extensive simulation is to evaluate the performance of the MBLCR scheme by assessing the effects of well-known conventional routing schemes. The following parameter metrics are used to illustrate the performance of the routing approaches:

(i)

Throughput: the ratio between the total number of data packets received at the destination device to the time spent during the data packet transmission. It can be calculated as follows (in Kbps):

$$\mathrm{Throughput}=\frac{\left({\displaystyle \sum N_{Rec}\times 8}\right)}{{\displaystyle \sum T_{simulation}}}\times 1000\mathrm{kbps}$$

(12)

where $T_{simulation}$ is the time spent during the data packet transmission and $N_{Rec}$ refers the total number of data packets received at the destination.

(ii)

End-to-End Delay: the average time taken by the data packets to successfully transmit messages across the network between the source and the destination devices. The end-to-end delay includes route discovery latency, retransmission, buffering, and queuing delays. The end-to-end delay is calculated as follows:

$$\mathrm{EED}=\frac{1}{N_{Rec}}{\displaystyle \sum _{c\in N}Delay(c)}$$

(13)

where $Delay(c)$ combines the entirely delay that occurs during the data packet transmission in the network.

(iii)

Packet Delivery Ratio (PDR): the ratio between the total number of packets received at the destination and the total number of packets transmitted from the source to the destination devices in the network. The PDR is directly aligned to the efficacy of a multicast routing, i.e., the higher the PDR value, the more efficient the routing protocol. It can be computed as follows:

$$\mathrm{PDR}=\frac{N_{Rec}}{N_{Sent}}\times 100$$

(14)

where $N_{Sent}$ defines as the total number of packets transmitted in the network from the source device.

(iv)

Packet Drop: the total number of packets dropped before reaching the destination device. Packets drop is computed as the difference between the total number of packets transmitted and the total number of data packets received between the source and the destination device during data transmission in the network.

$$\mathrm{Packets}\mathrm{Drop}=N_{Sent}-N_{Rec}$$

(15)

(v)

Energy Consumption: the network devices’ total energy consumption for significant data packet transfer throughout the network simulation time. Energy consumption is computed as follows:

$$\mathrm{Energy}\mathrm{Consumption}=\frac{1}{N}{\displaystyle \sum _{c\in N}{E}_{Total}(c)}$$

(16)

where $E_{Total}(c)$ refers to the total amount of energy consumption by the network devices and $N$ refers to the total number of devices in the network.

(vi)

Energy Cost: the ratio between the total energy consumption of the network devices for significant data packet transmission and the total number of packets successfully transmitted to the destination. It is computed as follows:

$$\mathrm{Energy}\mathrm{Cos}\mathrm{t}=\frac{\mathrm{Total}\mathrm{energy}\mathrm{cunsumption}\mathrm{of}\mathrm{the}\mathrm{network}\mathrm{devices}}{N_{\mathit{Rec}}}$$

(17)

4. Results and Discussion

This section elaborates the results of the MBLCR scheme and compares its effectiveness to conventional routing schemes by employing different device speeds. The obtained results of the routing schemes are compared in terms of performance evaluation metrics, as stated in the methodology section. The following subsections describe the comprehensive and critical analysis of the obtained simulation results.

4.1. Throughput

A comparison of the throughput parameter of the proposed MBLCR and two existing methods, MP-OLSRv2, MEQSA-OLSRv2 and EMBLR, at various device speeds is depicted in Figure 5. From the figure, it can be seen that the MBLCR approach constantly delivers higher network throughput value than its counterpart routing approaches at all device speeds. The MBLCR approach attains higher throughput, indicating its superiority in selecting reliable and robust paths. The MBLCR reduces the flooding impacts of RREQ packets, resulting in a precise assessment of the ETX metric. Besides that, the MBLCR employs a CW size with the back-off increment factor σ that shortens the waiting time of devices in the optimal path selection, which allocates higher channel access and mitigates data packet collision among the devices. The MP-OLSRv2 exploits path recovery with loop detection methods, whereas the MEQSA-OLSRv2 exploits the network topology sensing with path computational methods for relay device selection in the path computational procedure. These methods require additional data packet transmission to establish the route, which leads to a decreased number of data packets being successfully delivered to the destination at a specific period. Meanwhile, the MBLCR employs the ETX metric, which is evaluated with the probe message prior to data packet transmission and mitigates the probability of link breakage and increases the network throughput. As the device speed increases from 10 m/s to 60 m/s, the network throughput decrements for all routing schemes. The network throughput of the MP-OLSRv2 decrements from 54.12 kbps to 44.56 kbps, the MEQSA-OLSRv2 decrements from 55.28 kbps to 47.21 kbps, the EMBLR decrements from 59.45 kbps to 50.64 kbps, and MBLCR decrements from 60.92 kbps to 51.47 kbps, when the device speed increases from 10 m/s to 60 m/s.

4.2. End-to-End Delay

As shown in Figure 6, it can be seen that the MBLCR approach provides significantly lower end-to-end delay compared to the other routing approaches. This is because the other routing approaches do not exploit the average access channel delay, which highly depends on the contention window size of the devices. In contrast, the MBLCR utilizes the contention window size factor with a lower back-off time metric that induces less delay in the network for reliable data transmission. In addition, the end-to-end delay combines the propagation and queuing delay from the source to the destination device together to calculate a retransmission delay for each relay device. Moreover, as the device speed increases, there is a chance of link failure due to the frequently changing network topology, which increases the retransmission delay of the other routing schemes for data transmission. In this scenario, the MBLCR maintains and controls the retransmission delay by exploiting the ETX value through probe packets that evaluate the link quality and minimize the relay devices in the route prior to data transmission. Overall, it can be observed that, at 30 m/s device speed, the MBLCR decreases the end-to-end delay by approximately 32.15%, 19.35% and 9.38%, compared to the MP-OLSRv2, MEQSA-OLSRv2 and EMBLR schemes.

4.3. Packet Delivery Ratio

The proposed MBLCR approach provides higher PDR value (80%) in all different device speed scenarios, which proves its performance in providing robust and reliable data transmission in wireless networks. Due to the device movement in the network, the network topology is changed in an unpredictable manner, which inconsistently causes the links to be broken and reconnected in the established network. In this situation, the MBLCR avoids forwarding the data packets to an unstable link that has a higher probability of being broken. Moreover, the MBLCR prevents devices with a high displacement, which leads to of the maximization of the PDR value for data transmission when establishing routes. The MBLCR scheme exploits the ETX value that minimizes the number of relay devices for the route computational mechanism, which mitigates the number of dropped packets in the wireless network. It can be observed from Figure 7 that the MBLCR obtains a higher PDR value compared to its counterpart routing schemes. The average values in high device speed scenarios are enhanced by 28.24%, 21.54% and 12.27% in comparison to the MP-OLSRv2, MEQSA-OLSRv2 and EMBLR schemes.

4.4. Packet Drop

The packet drop variation of the MBLCR and its counterpart routing schemes during data transmission is shown in Figure 8. It can be seen that the MBLCR has better performance than the counterpart routing schemes. The MP-OLSRv2, MEQSA-OLSRv2 and EMBLR schemes do not utilize the awareness of channel condition with back-off time exponent factor for data packet transmission. Therefore, these routing schemes have a higher collision probability for data packets. In comparison, the MBLCR exploits the back-off time exponent factor value prior to data packets being transmitted to the link, which mitigates data packet collision among the devices and the number of dropped packets. Moreover, the MBLCR limits the flooding of RREQ packets and data packet retransmission when link failure occurs among the devices, minimizing the number of packet drops in the network. As the device’s speed increases, data packet drops in the network also increase. The MP-OLSRv2’s packet drop (29.62%), the MEQSA-OLSRv2’s packet drop (23.14%), and EMBLR packets dropped (10.45%), surpass that of the MBLCR scheme at the 60 m/s device speed.

4.5. Energy Consumption

Figure 9 illustrates energy variation consumed by the network devices for the MBLCR and its counterpart routing schemes throughout the network simulation time. The simulation results show that the MBLCR performs better than the other routing schemes owing to the reason that the MBLCR selects the relay devices with the exploitation of low ETX value acquired from the probe packets. Besides, the MBLCR evaluates the link quality value with the exploitation of the ETX parameter that can reduce the adverse flooding effect of RREQ messages in the wireless network. Thus, the ETX parameter minimizes the probability of frequent route failures, resulting in lower energy consumption in the route computational procedure. Nevertheless, as the device speed increases, the energy consumed by the network devices also increases. Figure 9 shows that energy consumption increases from 54.264 mAh to 57.064 mAh for the MEQSA-OLSRv2, from 55.098 mAh to 58.247 mAh for the MP-OLSRv2, from 53.058 mAh to 55.131 mAh for EMBLR, and from 52.164 mAh to 54.1245 mAh for the MBLCR when the device speed elevates from 10 m/s to 60 m/s.

4.6. Energy Cost

The energy cost simulation result of the routing schemes at various device speeds is depicted in Figure 10. The MBLCR obtains the lowest energy cost in the exchange of topological network information during the optimal path selection. The MBLCR incorporates the MCDM decision technique, which weighs and ranks the relay devices in the optimal path selection according to the devices’ residual battery, thereby improving energy utilization during the data packet transmission. Moreover, the MBLCR exploits the ETX value prior to data transmission, which mitigates the flooding impact of RREQ messages and results in the minimization of device energy cost. From the perspective of device speed, it can be seen from Figure 10 that the MBLCR attains the lowest energy cost per packet in comparison to the counterpart routing approaches. The energy costs achieved by the MP-OLSRv2 are 24.34% and 20.14%, MEQSA-OLSRv2 are 22.48% and 19.64%, and the energy costs of the EMBLR are 16.85% and 15.26% higher than that of the MBLCR when the device speed increases from 10 m/s to 60 m/s.

5. Conclusions

This paper presents a mobility, battery, link quality, and contention window size multicriteria-aware routing (MBLCR) approach to enhance the D2D communication over IoT-based 5G network performance. In the MBLCR approach, the mobility, battery, link quality, and contention window size of relay devices are evaluated to select the optimal route. Furthermore, an MCDM technique is employed, which provides the weights among the relay devices based on an evaluation of the relay devices’ values for the selection of an optimal route among the multiple routes. The effectiveness and efficacy of the MBLCR approach are evident in the QoS performance metrics obtained from the extensive simulation using different device speeds. Overall, it can be proven that the MBLCR outperforms as compared to the MP-OLSRv2, MEQSA-OLSRv2 and EMBLR schemes, respectively. It can be concluded that D2D communication in future wireless network applications is data-hungry, demanding that various research challenges in a heterogenous network domain are met.