Hybrid Muddy Soil Fish Optimization-Based Energy Aware Routing in IoT-Assisted Wireless Sensor Networks

Abstract

The seamless operation of interconnected smart devices in wireless sensor networks (WSN) and the Internet of Things (IoT) needs continuously accessible end-to-end routes. However, the sensor node (SN) relies on a limited power source and tends to cause disconnection in multi-hop routes because of a power shortage in the WSN, eventually leading to the inefficiency of the total IoT network. Furthermore, the density of available SNs affects the existence of feasible routes and the level of path multiplicity in the WSN. Thus, an effective routing model is predictable to extend the lifetime of WSN by adaptively choosing the better route for the data transfers between interconnected IoT devices. This study develops a Hybrid Muddy Soil Fish Optimization-based Energy Aware Routing Scheme (HMSFO-EARS) for IoT-assisted WSN. The presented HMSFO-EARS technique majorly focuses on the identification of optimal routes for data transmission in the IoT-assisted WSN. To accomplish this, the presented HMSFO-EARS technique involves the integration of the MSFO algorithm with the Adaptive $\beta$-Hill Climbing (ABHC) concept. Moreover, the presented HMSFO-EARS technique derives a fitness function for maximizing the lifespan and minimizing energy consumption. To demonstrate the enhanced performance of the HMSFO-EARS technique, a series of experiments was performed. The simulation results indicate the better performance of the HMSFO-EARS algorithm over other recent approaches with reduced energy consumption, less delay, high throughput, and extended network lifetime.

1. Introduction

The Internet of Things (IoT) has been recently made practical with the enforcement of certain existing technologies, such as intelligent sensing and wireless sensor networks (WSN) [1,2]. The applications of the IoT comprise agriculture [3], healthcare [4], security [5], transportation, inventory tracking, and smart grid networks [6]. The interconnected smart objects with embedded sensors in the IoT platform coordinate and cooperate to transfer the gathered data to a gateway sink. For IoT-related applications, such as smart homes, as well as logistics management, industrial control, and environmental sensing, the WSN becomes an indispensable part of the architecture [7]. The WSN is denoted as a graph of many interconnected sensor nodes (SNs) in which all the nodes sense pieces of information from the network and transfer them to an ultimate station [8]. The architecture of IoT-related WSN is autonomously planned without any complicated time-taking installation and configuration other than the common wired networks for various purposes. Figure 1 exemplifies the infrastructure of WSN.

WSN are networks that include several efficient, cost-effective, and multifunctional SNs operating as a package to observe an area of interest (AoI) [9]. The SN collects information from the AoI and transfers it to a base station (BS), where it can be processed. WSN help monitor remote atmospheres. They are very effective in gathering data in various inaccessible zones such as climatic changes, coastguard, war-prone areas, forestry, and many more [5]. Several SNs can be presented in WSN, which are connected along with to a BS. In remote parts, it is not possible to reinstall and charge SNs [10]. There were numerous difficulties and limitations in WSN; the primary concern is energy, as the powerhouse for WSN were batteries that can be bound by time restraints. The next problem was self-management, as WSN generally function in rough and inaccessible atmospheres [11]; numerous difficulties, which include mitigation in the medium of propagation and augmented distance among BS and SN, will be major difficulties encountered by WSN. Lastly, in WSN, there comes a design constraint, as several favourable elements were prohibited by the incorporation due to storage, energy, and other problems. The authors have solved this problem recently through cluster formation from nodes, thus enhancing its lifetime.

For multi-hop transmission in WSN, a node probably has many options for choosing a path towards a destination. Several authors have modelled several routing techniques taking routing parameters into account, namely the energy levels of nodes, security, transferring rate, etc. In IoT-related WSN, the energy consumption of sensors will be a major problem. Thus, the impacts on power utilization were examined in many legacy routing protocols. Furthermore, several routing techniques can be devised with specific focus on the elongation and energy preservation of the network lifespan. The objective of energy management was to assure that the sensors execute for longer periods and every sensor consumes its energies equally. However, it is unavoidable that certain nodes in the network do not collaborate to save their energies.

This study develops a Hybrid Muddy Soil Fish Optimization-based Energy Aware Routing Scheme (HMSFO-EARS) for IoT-assisted WSN. The presented HMSFO-EARS technique majorly focuses on the identification of optimal routes for data transmission in the IoT-assisted WSN. To accomplish this, the presented HMSFO-EARS technique involves the integration of the MSFO algorithm with the Adaptive $\beta$-Hill Climbing concept. Moreover, the presented HMSFO-EARS technique derives a fitness function for maximizing the lifespan and minimizing energy consumption. For reassuring the improvised performance of the HMSFO-EARS technique, a series of experiments was performed under a varying number of nodes.

The rest of the paper is organized as follows. Section 2 offers a detailed review of existing works and Section 3 introduces the proposed model. Next, Section 4 provides experimental validation. Additionally, Section 5 discusses the challenges and future work. Finally, Section 6 concludes the work.

2. Related Works

Kim et al. [12] present a new routing system for balancing the energy consumption amongst every node and extend the WSN lifespan that establishes a score value allocated for every node beside a path as an integration of evaluation metrics. Particularly, the scoring method assumes the data of node densities at a particular region and the node energy level for representing the significance of individual nodes from the routes. Lindsey and Raghavendra [13] examine PEGASIS (power-efficient gathering in sensor information systems), an adjacent optimum chain-based protocol that is a development on LEACH. In PEGASIS, all the nodes transmit only with nearby neighbours and turn broadcasting to BS, therefore decreasing the count of energy spent per round. Brar et al. [14] introduce a directional transmission-based energy aware routing protocol termed as PDORP. The presented protocol PDORP has the features of either a power effective collecting sensor data system or DSR routing protocol. Additionally, the hybridization of GA and a bacterial foraging optimizer (BFO) was executed for the present routing protocol for identifying the energy-efficient optimum path. In [15], the authors have explained the principal systems of Route Discovery and Route Maintenance utilized by DSR, and they are demonstrated to enable the wireless mobile node for automatically creating an entirely self-organizing and self-configuring network between them.

Subramani et al. [16] presented an energy aware clustering and multi-hop routing protocol with a multiple sink (EACMRP-MS) method for IoT-aided WSN. The study aim was to proficiently minimize the power consumption of IoT–SNs, thus raising the network efficacy of the IoT network. The recommended approach originally depends on the Tunicate Swarm Algorithm (TSA) for cluster assembly and cluster head (CH) selection, along with that for the TSA. Additionally, the type-II fuzzy logic (T2FL) method was employed for optimally selecting the multi-hop routes, with many input variables utilized for achieving this. Tandon et al. [17] devised a biologically inspired cross-layer routing (BiHCLR) method for attaining effectual and energy conserving routing in the WSN-supported IoT. Primarily, the installed SNs will be organized in a grid form according to the grid-related routing method. After that, to allow energy conservation in BiHCLR, the fuzzy logic method was performed for selecting the CH for all cells of the grid. After that, a hybrid bio-inspired technique was utilized for selecting the routing path. The hybrid technique will combine Salp Swarm and moth search optimized methods.

Kaur and Chanak [18] proposed an intelligent fault-tolerant method in which distinct faults within the WSN-assisted IoT such as a link fault and node fault will be promptly identified and tolerated. It suggestively enhances network reliability. In [19], the authors modelled a robust reinforcement learning-related mobile sink method for dynamic routing and data collection. Moreover, the Q-Learning technique was applied for inducing automatic learning by the shortest route. Integrating such techniques will preserve stability of the network and efficiently enhances the routing performance along with the reward. Alkhliwi [20] presented an energy-efficient cluster-related routing protocol with secure IDS (EECRP-SID) for WSN. This aforementioned approach includes three main stages; they are intrusion detection, cluster construction, and optimal path selection. Primarily, the type-II FL-oriented clustering (T2FC) method with three input variables can be implemented for selecting CH. Along with the CH selection, the Salp Swarm optimization (SSO) method was leveraged for selecting optimal routes for inter cluster data communication, which leads to energy-effective heterogeneous WSN.

Chouhan and Jain [21] introduced the multipath routing protocol, utilizing the devised optimized technique, called the tunicate swarm grey wolf optimizer (TSGWO) methodology in the IoT-supported WSN. The multipath was modelled by the multipath source node to numerous destinations by the multipath routing protocol. In the beginning, the nodes in IoT-supported WSN can be experimented with together and execute the CH selection through the Fractional Gravitational Search Algorithm (FGSA). After that, the multipath routing procedure can be performed based on the devised TSGWO, whereby the routing path was chosen through the consideration of the fitness variables, for example, QoS trust factors and variables. In [22], collision-aware routing utilizing the multi-objective seagull optimization algorithm (CAR-MOSOA) was modelled to address the efficacy of scalable WSN. The presented model uses the clustering procedure for selecting CHs for transferring the data from the source to destination, therefore constituting a scalable network, and enhancing the CAR-MOSOA protocol’s performance.

Though several models are available in the literature, there is still a need to enhance the energy efficiency and lifespan of the network. Most of the existing models do not take multiple input parameters for the objective function into account. Therefore, in this study, we develop a new HMSFO-EARS technique with multiple parameters in the objective function for an enhanced network performance.

Table 1. Summary of existing routing protocols in WSN.

Reference	Objective	Method	Metrics	Advantages	Limitations
Subramani et al. [16]	To develop an energy aware clustering with routing for IoT-assisted WSN	EACMRP-MS technique, TSA-based CH selection, T2FL routing	Energy efficiency	Adjust routes via mobile sink trajectories	Requires extensive experimentation
Tandon et al. [17]	To introduce a routing technique for IoT–WSN	BiHCLR technique	Packet loss, error bit rate, transmission delay, lifetime of network, buffer occupancy, throughput	Alternate solution for energy aware routing	Low execution speed
Kaur and Chanak [18]	To design an intelligent fault-tolerant data routing method for IoT-assisted WSN	Fault-tolerant scheme	Average packet delivery, energy consumption, throughput, network lifetime, delay	Node fault and link fault are detected and tolerated in a timely manner	High congestion
Krishnan and Lim [19]	To introduce a reinforcement learning-related mobile sink method for dynamic	Q-Learning	Energy efficiency	Preserve network stability and improved routing	Requires extensive experimentation
Alkhliwi [20]	To present a cluster-related routing with intrusion detection	EECRP-SID technique, T2FC, SSO based clustering	Energy efficiency, lifetime, intrusion detection	Intrusion detection	Need to improve performance
Chouhan and Jain [21]	To design a routing technique for IoT–WSN	TSGWO technique, FGSA-based CH selection	Energy, lifetime, throughput, PDR	Involves route maintenance process	Low execution speed
Jagannathan et al. [22]	To present a cluster-based routing in IoT-assisted WSN	CAR-MOSOA technique	Energy, lifetime, throughput, PDR, delay	Secure data transmission with minimal energy utilization	Requires extensive experimentation
Heinzelman et al. [23]	To present a clustering-based protocol with arbitrary CH rotation	LEACH	Energy, lifetime	Reduces communication energy	Single-hop communication, arbitrary CH rotation

provides an overall summary of the reviewed works.

3. The Proposed Model

In this study, a new HMSFO-EARS technique has been developed for an effectual routing process in the IoT-assisted WSN. The presented HMSFO-EARS technique majorly concentrates on identifying the optimal routes for data transmission in the IoT-assisted WSN. Moreover, the presented HMSFO-EARS technique derives a fitness function for maximizing the lifespan and minimizing energy consumption.

3.1. System Model

In this work, the network method is regarded as a collection of SNs randomly positioned in a network range with an N×N dimension. The nodes are under a random deployment with a Poisson distribution. All the sensors have different IDs and coordinate data. SNs gather environmental data and transfer the information to their corresponding nodes. Each node is heterogeneous with respect to energy. Furthermore, the BS is static in nature.

In this work, the first order radio energy model is used [23,24]. Once an SN receives or transmits information, it expends energy based on the distance D among the receiving and transmitting nodes. The free space communication model ($D^2$ model) for the direct or single-hop communication is adapted while the D is smaller and when D is larger the multipath models ($D^4$ model) are adapted. The amount of energy consumed $E_{TX}\left(l,D\right)$ to transmit $l$-bit packet over distance “D” is given as follows [23]:

$$E_{TX}\left(l,D\right)=\left\{\begin{array}{cc}l\ast E_{elec}+l\ast {\epsilon }_{fs}\ast D^2,& if D<D_0\\ l\ast E_{elec}+l\ast {\epsilon }_{amp}\ast D^4,& D\ge D_0\end{array}\right.$$

(1)

The received energy is defined as:

$$E_{RX}\left(l,D\right)=l\ast E_{elec}$$

(2)

In Equation (2), the size of the data packet is $l$ whereas $E_{elec}$ characterizes the energy expended for each bit. ${\epsilon }_{fs}$ characterizes the free space energy and ${\epsilon }_{amp}$ signifies the multipath energy model. The $D_0$ characterizes a threshold distance that controls states either to use ${\epsilon }_{amp}$ or ${\epsilon }_{fs}$:

$$D_0=\sqrt{\frac{{\epsilon }_{fs}}{{\epsilon }_{amp}}}$$

(3)

3.2. Steps Involved in HMSFO Algorithm

The presented HMSFO-EARS technique involves the integration of the MSFO algorithm with the Adaptive $\beta$-Hill Climbing (ABHC) concept. In this study, a novel MSFO algorithm is developed that is impacted by the search (food) analogy of cooperative/societal response or the swarming rudiment of fishes living in a turbid or muddy water habitat [25]. Generally, every single fish impacts the surroundings through its own neighbourhood efforts. Mainly, the MSFO technique is intended for solving the presented energy management problems by efficiently handling the energy flow through the networks toward decreasing the losses and production cost along with system and operation constraints amongst meeting the load demands.

Every fish that exists in the muddy water implements a localized search in a random fashion through solitary elements to search for a possible solution and it is evaluated in the following Equation (4):

$${\tau }_k^{msf}\left(t+1\right)={\tau }_k^{msf}\left(t\right)+r\times \sigma ,r\in \left[-\mathrm{1,1}\right].$$

(4)

The mobility of fish can be represented by the novel and existing locations ${\tau }_k^{msf}(t+1)$ and ${\tau }_k^{msf}(t)$, correspondingly. In addition, random number $r$ is formed by using a uniformly distributed range within [$-1$, 1] and the variable liable for the maximum step is denoted by $\sigma$. Consequently, all the fishes allow movement towards the optimum position if the fitness function at the neighbourhood position is higher than the existing position. $f({\tau }_k^{msf}(t+1))>f({\tau }_k^{msf}(t))$ positions become equivalent and are determined by ${\tau }_k^{msf}\left(t+1\right)={\tau }_k^{msf}\left(t\right);\Delta {\tau }_k^{msf}=0$ is defined by the following equation:

$$\Delta f_k=f\left({\tau }_k^{msf}\left(t+1\right)\right)-f\left({\tau }_k^{msf}\left(t\right)\right),$$

(5)

$$\Delta {\tau }_k^{msf}={\tau }_k^{msf}\left(t+1\right)-{\tau }_k^{msf}\left(t\right),$$

(6)

Furthermore, the development of exploitation capability in succeeding iterations has led to reducing the amplitude of the step $\sigma$ variable and it is estimated as follows:

$$\sigma \left(t+1\right)=\sigma \left(t\right)-\left[\frac{{\sigma }_i-{\sigma }_f}{Total Iterations}\right],$$

(7)

In Equation (7), the first and last steps are represented by ${\sigma }_i$ and ${\sigma }_f$. The first step must be higher than the last step. Figure 2 depicts the flowchart of the MSFO technique.

The algorithm begins with population initialization. Then, the fitness function is determined with foraging—impulsive and non-impulsive. Next, the current state of the soil fish can be evaluated and the optimal one is chosen as the target solution. The process gets repeated until the stopping criteria are satisfied. The foraging operators are closely connected to all the fish’s weight updating model. It can be defined by the success rate accomplished in the present iteration. The rise in fish weight maximizes the probability of acquiring an optimal region. Moreover, the weight function can be expressed by:

$$M_k\left(t+1\right)=M_k\left(t\right)+\left[\frac{\Delta f_k}{\max \left(\left|\Delta f_k\right|\right)}\right],$$

(8)

In Equation (8), $M_k(t+1)$ and $M_k\left(t\right)$ denote the mass of fish at novel and recent positions, correspondingly. The change occurs in the target functions concerning present and novel locations regarding the fish $k$ that is denoted by $\Delta f_k$, and $\Delta f_k$ is assumed as zero once the fish remain in a similar position. In addition, some additional efforts are included in the fitness assessment for guaranteeing improved convergence towards the optimal area over design space through the mass scale $\left(M_s\right)$.

• Each fish is born with a mass closer to $\frac{M_s}{2}.$
• Parameters are likely to limit the fish mass through $M_s$ within $[$1, $M_s].$

Such cooperative operators reveal the displacement pattern established by the fish, i.e., effective together with solitary movement and $f_k\ne 0$. The average weighted displacement factor $D\in R$ can be defined using the variance of fish fitness $\left(\Delta f_k\right)$ by involving the fish that have the highest fitness in the solitary movement and is given as follows:

$$\overrightarrow{D}\left(t\right)=\frac{{\sum }_{k=1}^N\Delta \overrightarrow{\tau }\Delta f_k}{{\sum }_{k=1}^N\Delta f_k},$$

(9)

In Equation (9), $N$ represents the maximal capacity of the cluster and the $\Delta {\tau }_k^{msf}$ vector denotes the variance among the novel and present locations of fish. Consequently, the present position of all the fish can be upgraded by:

$$\overrightarrow{\tau }\left(t+1\right)={\overrightarrow{\tau }}_k^{msf}\left(t\right)+\overrightarrow{D}\left(t\right).$$

(10)

This operator becomes instrumental to deal with the exploration and exploitation capability interconnected with the swarm. In these contexts, the rise in the total weight of the swarm indicates the remarkable efficiency of the search function and the additional increase in swarm size decreases the exploration of another region. Therefore, the size of the swarm can be controlled by means of centroid over its compression and expansion techniques. Furthermore, centroid $G$ is calculated on the basis of the current location of the fish and its corresponding weight can be given as follows:

$$\overrightarrow{G}\left(t\right)=\frac{{\sum }_{k=1}^N\overrightarrow{\tau }{M}_k\left(t\right)}{{\sum }_{k=1}^N{M}_k\left(t\right)}.$$

(11)

Once the swarm size is constantly rising in the present iteration, then each fish updates the current position. Likewise, the existing position of every fish is upgraded once swarm size reduces in the current iteration:

$${\overrightarrow{\tau }}^{msf}\left(t+1\right)={\overrightarrow{\tau }}^{msf}\left(t\right)-\gamma \left(\mathrm{0,1}\right)\ast {\sigma }_m\left[\frac{{\overrightarrow{\tau }}^{msf}\left(t\right)-\overrightarrow{G}\left(t\right)}{d\left({\overrightarrow{\tau }}_k^{msf}\left(t\right),\overrightarrow{G}\left(t\right)\right)}\right]$$

(12)

$${\overrightarrow{\tau }}^{msf}\left(t+1\right)={\overrightarrow{\tau }}^{msf}\left(t\right)+\gamma \left(\mathrm{0,1}\right)\times {\sigma }_m\left[\frac{{\overrightarrow{\tau }}^{msf}\left(t\right)-\overrightarrow{G}\left(t\right)}{d\left({\overrightarrow{\tau }}_k^{msf}\left(t\right),\overrightarrow{G}\left(t\right)\right)}\right].$$

(13)

In this study, the presented HMSFO-EARS technique involves the integration of the MSFO algorithm with the ABHC concept [26]. It is a new local search-based technique that is an adapted version of $\beta HC$. This study shows that $A\beta HC$ gives an improved performance when compared to others. To increase the exploitation capability and quality of the concluding solution, $A\beta HC$ is incorporated into the basic MSFO to help find the neighbourhood of the optimum solution. For the existing solution $X_i=(x_{i,1},x_{i,2},\dots ,x_{i,D}),$ $A\beta HC$ iteratively produces the improved solution $X_i^{″}=(x_{i,1}^{″},x_{i,2}^{″},\dots ,x_{i,D}^{″})$ according to the $\mathcal{N}$-operator and $\beta$-operator. First, the $N$-operator transfers $X_i$ to the new neighborhood solution $X_i^{\prime }=(x_{i,1}^{\prime },x_{i,2}^{\prime },\dots ,x_{i,D}^{\prime })$ that is determined by the following expression:

$$x_{i,j}^{\prime }=x_{i,j}\pm U\left(\mathrm{0,1}\right)\times \mathcal{N},j=\mathrm{1,2},\dots ,D$$

(14)

$$\mathcal{N}\left(t\right)=1-\frac{t^{\frac{1}{K}}}{\mathrm{M}\mathrm{a}\mathrm{x} ite{r}^{\frac{1}{K}}}$$

(15)

Whereas $U(O,1)$ indicates a random value between zero and one, $x_{ij}$ represents the value of the decision parameter in $j-\mathrm{t}\mathrm{h}$ dimensions, $t$ indicates the existing iteration, $Maxiter$ indicates the maximal amount of iterations, $N$ suggests the bandwidth distance amongst the existing solution and the neighbour, $D$ represents the dimension vector, and $K$ represents a constant. Then, the decision variable of novel solution $X_i^{″}$ is allocated from the present solution or arbitrarily from the presented range of the $\beta$-operator:

$${x_{i,j}^{\prime }}^{\prime }\leftarrow \left\{\begin{array}{cc}{x}_{i,r},& if r_8<\beta \\ x_{i,j}^{\prime },& else\end{array}\right.$$

(16)

$$\beta \left(t\right)={\beta }_{\mathrm{m}\mathrm{i}\mathrm{n}}+\left({\beta }_{\mathrm{m}\mathrm{a}\mathrm{x}}-{\beta }_{\mathrm{m}\mathrm{i}\mathrm{n}}\right)\times \frac{t}{\mathrm{M}\mathrm{a}\mathrm{x} iter}$$

(17)

Now, $r_8$ indicates a random value within zero and one $,$ $x_{i,r}$ represents another random value selected from the potential range of that specific problem dimension, and ${\beta }_{\mathrm{m}\mathrm{a}\mathrm{x}}$ and ${\beta }_{\mathrm{m}\mathrm{i}\mathrm{n}}$ denote the maximal and minimal values of probability value $\beta \in [0,1]$, correspondingly. Once the produced solution $X_i^{″}$ is superior to the existing solution, $X_i$ can be replaced with $X_i^{″}$. The overall process of the ABHC concept is given in Algorithm 1.

Algorithm 1: Adaptive $\beta$-Hill Climbing Algorithm

1: Initialization of the variables ${\beta }_{\mathrm{m}\mathrm{a}\mathrm{x}},$ ${\beta }_{\mathrm{m}\mathrm{i}\mathrm{n}}$, and K 2: Input the existing solution $X_i=(x_{i,1},x_{i,2},\dots ,x_{i,D})$ 3: Compute the fitness value $F\left(X_i\right)$ 4: While $t\le$ Maxiter do 5: Produce the neighbouring solution $X_i^{\prime }$ based on Equation (14) 6: For $j=1$ to $D$ do 7: If $r_8<$ then 8: $x_{i,j}=x_{i,r}^{}$ 9: Else 10: $x_{i,j}^{″}=x_{i,j}^{\prime }$ 11: End If 12: End For 13: Compute the fitness value $F\left(X_i^{″}\right)$ 14: If $F\left(X_i^{″}\right)<F\left(X_i\right)$ then 15: $X_i=X_i^{″}$ 16: End If 17: $t=t+1$ 18: End while

3.3. Derivation of Routing Process

The major objective of the HMSFO-EARS approach is to minimize energy utilization and maximize the lifespan of the network of each SN [27]. Consider $h_1$ as an objective function so that the next-hop relay node with larger residual energy is chosen to route the data and thus the network lifespan can be maximized, for example, $h_1$ is maximization. Consider $h_2$ as an additional main function that is the minimal distance to next-hop to $BS$. Consider $h_3$ as the third main function so that the relay node selects the next-hop relay node with less node degree. The proposed model executes an objective function for the maximization of network lifetime and minimization of energy, given as follows:

$$b_{ij}=\left\{\begin{array}{cc}1& if next-hop\left( \right)=C{H}_j,{\forall }_{i,j}1\le i,j\le m\\ 0& Otherwise\end{array}\right.$$

(18)

$$Minimize F=\frac{1}{h_1}\times {\beta }_1+h_2\times {\beta }_2+h_2\times {\beta }_3$$

(19)

Subjected to,

$$dis\left(N_i,N_j\right)\times \le d_{\mathrm{m}\mathrm{a}\mathrm{x}}RN\epsilon \left\{C+BS\right\}$$

(20)

$$\sum _{j=1}^m{b}_{ij}=1 and 1\ne j$$

(21)

$$0<{\beta }_1,{\beta }_2,{\beta }_3<1$$

(22)

4. Experimental Validation

This section examines the performance of the HMSFO-EARS model under a varying number of nodes. The proposed model is simulated using the MATLAB tool. A detailed comparative result analysis took place with recent models such as

• Node status and score-based route optimization protocol (NSSROP) [12]: involves DSR and OLSR with the novel scoring mechanism for path selection;
• Power-Efficient Gathering in Sensor Information Systems (PEGASIS) Routing Protocol (PRRP) [13]: a chain cluster-based routing protocol;
• Proposed directional transmission-based energy aware routing protocol (PDORP) [14]: integrates PEGASIS and DSR for route selection;
• Dynamic source routing (DSR) [15]: forms a route on-demand when a transmitting node requests one;
• Low energy adaptive clustering hierarchy (LEACH): a classical clustering scheme.

Table 2. Parameter Settings.

Parameter	Value
Number of nodes	50–250
Traffic type	CBR
Simulation time	100 s
Network size	250 × 250 m2
Packet size	4000 bits
MAC protocol	IEEE 802.15.4
Transmission range	30 m
Sensing range	10 m
Location of BS	125 × 125 m2
Node distribution	Random

offers the parameter settings involved in the simulation process.

Table 3. AECN analysis of HMSFO-EARS system with existing approaches under different nodes.

Average Energy Consumption (J)
No. of Nodes	HMSFO-EARS	NSSROP	PRRP	PDORP	DSR	LEACH
50	5.21	6.10	9.71	8.00	9.37	9.37
100	5.48	6.10	6.10	7.39	16.45	16.45
200	5.28	6.57	5.55	7.46	9.71	24.21
250	5.35	6.51	5.62	7.87	10.73	27.75

and Figure 3 report a comparative average energy consumption (AECN) analysis of the HMSFO-EARS model with recent models. The outcomes specified that the LEACH model has shown poor performance with higher AECN values. Then, the PDORP and DSR models have indicated moderately closer and reduced AECN values.

Though the PRRP model has reached reasonable AECN values, the HMSFO-EARS model has obtained the least AECN values under all nodes. For example, with 50 nodes, the HMSFO-EARS model has presented the least AECN of 5.21 J whereas the NSSROP, PRRP, PDORP, DSR, and LEACH models have attained an increased AECN of 6.10 J, 9.71 J, 8 J, 9.37 J, and 9.37 J, respectively.

A comparative number of alive nodes (NAN) study of the HMSFO-EARS model is portrayed in

Table 4. NAN analysis of HMSFO-EARS system with existing approaches under different rounds.

No. of Alive Nodes (%)
No. of Rounds	HMSFO-EARS	NSSROP	PRRP	PDORP	DSR	LEACH
1	100	100	99	99	98	99
2	99	98	97	98	96	96
3	97	95	89	88	84	86
4	82	77	68	63	56	61
5	70	66	45	40	26	32

and Figure 4. The results inferred the betterment of the HMSFO-EARS model with increasing values of NAN. For instance, in Round 1, the HMSFO-EARS model has revealed a higher NAN of 99% whereas the NSSROP, PRRP, PDORP, DSR, and LEACH methods have portrayed a lower NAN of 98%, 97%, 98%, 96%, and 96%, respectively. In addition, in Round 5, the HMSFO-EARS method has portrayed a higher NAN of 70% while the NSSROP, PRRP, PDORP, DSR, and LEACH models have shown the lowest NAN of 66%, 45%, 40%, 26%, and 32%, correspondingly.

A comparative throughput (THRO) study of the HMSFO-EARS method is demonstrated in

Table 5. Throughput analysis of HMSFO-EARS system with existing approaches under different pause times.

Throughput (KBPS)
Pause Time	HMSFO-EARS	NSSROP	PRRP	PDORP	DSR	LEACH
1	43	40	38	30	9	3
2	44	41	37	34	10	3
3	45	41	37	37	12	4
4	46	42	37	39	14	4
5	46	42	37	39	17	4
6	47	42	38	39	17	5
7	47	43	38	39	18	4
8	47	44	38	40	19	6
9	47	44	39	39	18	9
10	47	45	40	39	19	9

and Figure 5 under different pause times. The pause time refers to a period of time during which a node temporarily stops transmitting or processing data. This is often used as a means of power management, as transmission and processing activities consume a significant amount of energy in WSN. The outcomes show the betterment of the HMSFO-EARS approach with increasing values of THRO. For example, for Pause Time 1, the HMSFO-EARS technique has revealed a higher THRO of 43 KBPS whereas the NSSROP, PRRP, PDORP, DSR, and LEACH methods have portrayed a lower THRO of 40 KBPS, 38 KBPS, 30 KBPS, 9 KBPS, and 3 KBPS, correspondingly. Furthermore, for Pause Time 10, the HMSFO-EARS method has revealed the higher THRO of 47 KBPS whereas the NSSROP, PRRP, PDORP, DSR, and LEACH techniques have depicted a lower THRO of 45 KBPS, 40 KBPS, 39 KBPS, 19 KBPS, and 9 KBPS, correspondingly.

Table 6. Delay analysis of HMSFO-EARS system with existing approaches under different pause times.

Delay (ms)
Pause Time	HMSFO-EARS	NSSROP	PRRP	PDORP	DSR	LEACH
1	1.47	4.20	5.85	9.84	26.20	21.72
2	1.67	4.20	6.53	9.75	26.01	22.40
3	2.15	4.20	6.92	10.82	24.93	22.50
4	1.86	4.59	6.34	10.72	24.93	23.77
5	1.47	3.81	4.98	11.21	24.06	24.16
6	1.76	3.42	5.27	11.60	22.89	25.52
7	1.28	3.71	5.17	12.47	22.89	25.52
8	1.28	2.74	4.68	12.47	22.70	25.42
9	1.97	3.03	4.59	12.47	23.86	25.03
10	2.15	3.81	4.88	11.99	22.99	25.23

and Figure 6 show a comparative delay (DEL) inspection of the HMSFO-EARS with recent techniques. The outcomes show that the LEACH approach has demonstrated poor performance with high DEL values. Then, the PDORP and DSR techniques have showed moderately closer and reduced DEL values. However, the PRRP method has obtained reasonable DEL values, and the HMSFO-EARS approach has attained the lowest DEL values under all pause times. For example, with 10 pause times, the HMSFO-EARS model has provided the minimum DEL of 2.15 ms while the NSSROP, PRRP, PDORP, DSR, and LEACH models have attained an increased DEL of 3.81 ms, 4.88 ms, 11.99 ms, 22.99 ms, and 25.23 ms, correspondingly.

From these values, it is reassured that the HMSFO-EARS model has accomplished better performance. The enhanced performance of the proposed model is due to the incorporation of the MSFO algorithm with the ABHC concept. Additionally, the inclusion of multiple parameters in the objective function helps in the optimal selection of routes to their destination.

5. Challenges and Future Work

Despite the benefits of the proposed model in the routing process, a few challenging issues that need to be addressed are discussed here. The proposed model uses a random distribution of nodes in the network, which poses some demerits such as uneven node distribution, uneven coverage, limited range, resource consumption, and complex network management. In addition, in a connection-based mesh topology, every node acts as a relay or router, transmitting data between other nodes in the network. This type of topology can be used to create a highly redundant and fault-tolerant network, as data can be routed through multiple paths in the network if one path fails. As the number of nodes in the network increases, the number of connections and routing tables required can become overwhelming. It makes it complex to scale the network, particularly if new nodes need to be added frequently. On the other hand, in many cases, the connections in a mesh topology are dynamic and can vary depending upon node movement, topology changes, or routing demands. Dynamic connections in a mesh network can provide greater flexibility and adaptability, allowing the network to reconfigure itself in response to changing conditions. However, this can also increase the complexity of the network, as nodes must constantly monitor and update their connections and routing tables. In DSR, the source node of a data packet determines the complete route of the packet and includes this information in the packet header. While DSR can provide an efficient and flexible way to route data in the networks, it can also result in increased energy consumption at the source node. This is because the source node must actively participate in the routing process by selecting the appropriate route and including this information in the packet header. This can require significant processing and communication resources, which can increase energy consumption and reduce the battery life of the source node.

6. Conclusions

In this study, a new HMSFO-EARS technique has been developed for an effectual routing process in the IoT-assisted WSN. The presented HMSFO-EARS technique majorly concentrated on identifying the optimal routes for data transmission in the IoT-assisted WSN. To accomplish this, the presented HMSFO-EARS technique involves the integration of the MSFO algorithm with the ABHC concept. Moreover, the presented HMSFO-EARS technique derives a fitness function for maximizing the lifespan and minimizing energy consumption. To demonstrate the enhanced performance of the HMSFO-EARS technique, a series of experiments was performed. The simulation results indicate the better performance of the HMSFO-EARS technique over other recent approaches. The proposed model improves the reliability of the network and decreases the failure rates. In addition, the proposed model balances the tradeoff between energy consumption, data transmission reliability, scalability, network lifetime, and latency when selecting a routing technique for WSN. In the future, the presented HMSFO-EARS algorithm can be extended to the design of a data aggregation protocol to boost network efficiency. Additionally, the performance of the proposed model can be tested on a large-scale real-time environment.