Underwater clustering based hybrid routing protocol using fuzzy ELM and hybrid ABC techniques

Abstract

Underwater wireless sensor networks (UWSNs) are designed to perform cooperative monitoring and data collection tasks by combining several elements, such as automobiles and sensors located in a particular acoustic area. Several studies have been carried out to improve energy efficiency and routing reliability. However, UWSN faces several challenges, such as high ocean interference and noise, long transmission delays, limited bandwidth, and low sensor node battery energy. In this work, a novel underwater clustering-based hybrid routing protocol (UC-HRP) has been proposed to address these issues. The overall process is carried out in three phases. In the first phase, the fuzzy-ELM approach is used to initialize the cluster based on parameters such as Doppler spread, path loss, noise, and multipath. In the second phase, the cluster head is selected using Cluster Centre Cluster Head Selection (C3HS) based on Link quality, distance, node degree, and residual energy. In the third phase, Hybrid Artificial Bee Colony (HABC) algorithm is used for selecting an optimal route based on the parameters such as reliability, bandwidth effectiveness, average path loss, and average transmission latency. The performance of the proposed UC-HRP method is evaluated using a variety of parameters, including the network lifetime, packet delivery ratio, alive nodes, and energy consumption. The proposed technique improves the network lifetime by 14.03%, 16.25%, and 18.34% better than ACUN, ANC-UWSNS, and MERP respectively.

Keywords

Underwater wireless sensor networks fuzzy extreme learning machine cluster centre cluster head selection hybrid artificial bee colony algorithm

1 Introduction

The surface of a planet is made up of radiances collected by the Nimbus-7 total ozone mapping spectrometer between latitudes 70° for 14.5 years. (November 1978 to May 1993). [1]. Animals and other creatures abound in the oceans [2]. They are the primary basis of food, transportation, and further resources for several beings on Earth [3]. Water makes up almost 70% of the surface area of the world, which spans 140 million square miles [4]. The ocean is approximately four kilometers deep, with the majority of that depth yet undiscovered [5]. The majority of the water on Earth—roughly 96% —is found in the oceans, which are crucial for human life since they offer natural resources, marine defense, and other advantages [6]. Because underwater wire-less communications have only recently been developed, underwater communication is still primarily accomplished today via communication cables [7].

Underwater wireless sensor networks (UWSNs) re-main becoming more than more popular in both academia [8] and business because of the wide number of application areas they may be employed in, including environment monitoring, disaster risk reduction, derived steering, energy exploration, and monitoring [9]. Strong, distinctive technologies in sub-merged monitoring, ocean inspection, marine super-vision, and convention expansion for sensing flood-ed benchmarks have recently garnered a lot of attention in the field of UWSNs [10]. A typical UWSN is made up of a huge quantity of sensor nodes that are moored toward the ocean floor along with wirelessly connected to one or more underwater gateways [11]. All of the UWSN’s sensor nodes are scattered around the stream, generally from the edge to the bottom [12]. The base station (BS), other nodes, and most locations where data is gathered stored, and analyzed require that the wireless sensor nodes be positioned at a distance from them [13].

In underwater environments, underwater communication is subject to several restrictions [14]. The link superiority of this node is bumped by supplementary fundamentals, comprising the Doppler occurrence transfer and ecological blare interfering due to the complexity of underwater grooves. The spectacular velocity of records broadcast, the loyalty of data announcement, meshwork bandwidth, and energy convention of sub-aqua grids are the entire frankly influenced by thou interventions [15]. Therefore, the clustering technique is utilized to minimize the energy utilization of the antenna appliance node and to increase the effectiveness of data transmission and the lifetime of the network. In the transmission of network data, the route is crucial [16].

Data transfer from the network’s resource node to the target node is definite by a routing mechanism [17]. The submarine atmosphere is difficult and unsound with submerged sensor nodes including moderately partial dispensation, storage, and announcement funds Contrasted to terrestrial wireless sensor networks [18], the routing compact for UWSNs is further complex and constrained [19]. As the network is divided into nodes, more energy is consumed, resulting in transmission delays [20]. To overcome these challenges Underwater Clustering based Hybrid Routing Protocol (UC-HRP) has been proposed. The main contribution of the proposed method is as follows:

The primary goal of the proposed method is to extend the life cycle and balances the energy consumption of the network.

The proposed UC-HRP approach involves three phases namely cluster formation, cluster head selection, and optimal routing.

The fuzzy-ELM approach is used to initialize the cluster based on parameters such as Doppler spread, path loss, noise, and multipath in the first phase.

The Cluster Centre Cluster Head Se-lection (C3HS) approach is used to select the cluster head in the second phase based on Link quality, distant node degree, and residual energy.

The Hybrid Artificial Bee Colony (HABC) algorithm is employed in the third phase to select the best route based on reliability, bandwidth effectiveness, average path loss, and average transmission delay.

The proposed method produces high net-work lifetime, throughput, and less energy consumption during data transmission.

The remaining section of this research is demonstrated as follows. The literature review is explained in Section 2. The proposed UC-HRP was shown in Section 3, along with an explanation and the related algorithm. The performance outcome and their analysis are provided in Section 4. Section 5encloses with Conclusions and future work.

2 Literature survey

In recent years, many academics have focused on UWSN, which has drawbacks in terms of data collection, energy utilization of individual nodes, node failure, data packet latency, packet delivery, and packet dropping. Numerous clustering and routing techniques have recently been created for WSNs to increase network longevity and energy efficiency. In this section, these techniques are briefly illustrated.

In 2019 Faheem, M., et al. [21] proposed a multi-objective evolutionary guiding method inspired by biology for UWSN-based requests. It is exciting to construct a very constant clustering-based routing device used for a well-organized data congregation with low energy consumption due to its distinctive immersed properties. The widespread replication’s findings show that the planned organization is successful in achieving its intended goals, despite using well-known UWSN-based routing methods for observing and investigating marine environments.

In 2021Khan, M.F., et al. [22] proposed a technique for adaptive node groups for sensor node collection in UWSNs (ANC-UWSNs). The provided organization is contrasted by extra procedures, such as the ant colony optimizer (ACO), a complete knowledge tool used to assess the efficiency of the rendering algorithm being difficult and more physical. DFO outperforms the other algorithms, according to the results. When compared to other algorithms, it creates more optimal clusters, which improves general transmitting formerly lengthens the lifetime of a network

In 2021 Luo, J., et al. [23] recently, a protocol for underwater routing was presented. The proposed protocols are detailed, provide benefits and drawbacks of sensor node energy consumption, and identify the best transmission path. They also present research challenges and potential future directions for underwater routing protocols, which can aid future exploration by the researcher.

In 2020 Coutinho, R.W., et al. [24] proposed an innovative power control-based intelligent (PCR) routing technology for the Internet of Things. The suggested method chooses the appropriate broadcast power rating at each sensor node by taking into account the region density, link quality, distance, packet development, and energy excess. Due to hidden mortal issues, the appearance of the underwater acoustic channel, and IoT scenarios, no applied program of data packets is inspirational. The effectiveness of the PCR process was shown by the numerical results obtained by tracking the broadcast power of the underwater sensor nodes.

In 2020 Faheem, M., et al. [25] proposed a company that sends data traffic across a broad network to reduce high energy usage and associated difficulties. Trendy terms of low potential, EC, native finest problematic, extraordinary quantity, and PDR for underwater observing submissions, the FFRP routing system routine is discernible. The optimal routine of our planned arrangement, in contrast to all other routing organizations in UWEs, is validated by the replication scores.

In 2019Wan, Z., et al. [26] proposed an extremely creative approach to making inter-cluster claims. The ACUN procedure extends the effective time of the network and conserves network life, but practical applications were unable to suggest the ACUN procedure and contrast it to the ATP modus operandi and DEBCR computation in terms of network energy utilization, node endurance rate, and network running existence.

In 2020 Yao Sun., et al. [27] proposed a technique that turns the web into a multi-agent scheme and permits the knots to decide on the world’s best path together via support learning. To moderate the likelihood of hotspot formation, an adaptive cluster head range approach without any additional communication overhead is adopted. According to replication consequences, the proposed adaptive clustering routing protocol outperforms existing methods in terms of routing efficiency, energy consumption, and network longevity.

In 2019 Jouhari, M, et al. [28] proposed a determination to increase network size and statement series. To overcome the sink area troubles for network lifetime extension. Acoustic communication between nodes is implemented in this protocol for low data rates and long communication ranges. In this protocol for its low information frequency and long communiqué collection MI, on the other hand, is used for high data rates and short communication ranges.

In 2020 Fattah, S., et al. [29] offered a technique that sought to cut down on localization time while allowing for collisions. Because of their limitations and the demands of the applications, UWSNs highlight the divide between applications and technologies. A mixture gathering energy sensors method for ocean-atmosphere monitoring must exist carefully to gather and also deploy extra dependable renewable energies in the challenging maritime atmosphere.

In 2021, Wei, X., et al. [30] proposed reliable data collection strategies in UWSNs, as well as issues specific to UWSNs and their effects on reliable information groups. This paper primarily discusses the glitches unique to UWSNs and their influence on information group reliability. This paper classifies them based on their capability to improve consistency at all key points of the data group. Finally, several potential research directions are identified and discussed.

In 2021 Al-Andoli, M., et al. [31] proposed a unique deep autoencoder combining continuation algorithms and identifying community patterns in complicated networks using particle swarm optimization (PSO). The frequency of local minima and the inefficiency of BP algorithms were two constraints in GD optimization using BP algorithms that were solved by PSO. Based on the data, the authors conclude that the proposed strategy performed best when compared to other methods.

In 2022 Al-Andoli, M.N et al. [32] proposed an approach for detecting the community in a huge complex network based on parallel deep learning (CNs). To reduce the amount of space and temporal complexity, the CN is divided into numerous pieces using a partitioning approach. The findings show how well the suggested deep learning approach with hybrid BP-PSO optimization is in identifying communities in large convolutional networks. while using the least amount of CPU and GPU device processing time.

However, several related studies have been conducted to reduce energy consumption to ensure a reliable network in UWNS. This paper proposed a UC-HRP method to reduce transmission latency and thereby improve network lifetime by minimizing energy consumption. The three phases of the proposed methodology are cluster formation, cluster head selection, and routing.

3 Proposed methodology

In this Research, Underwater Clustering-Based Hybrid Routing Protocol (UC-HRP) has been proposed to improve network lifetime and energy consumption in the UWSN. The overall process is carried out in three phases. In the first phase, the fuzzy-ELM approach is used to initialize the cluster based on parameters such as Doppler spread, path loss, noise, and multipath. In the second phase, the cluster head is selected using Cluster Centre Cluster Head Selection (C3HS) based on Link quality, distance, node degree, and residual energy. In the third phase, Hybrid Artificial Bee Colony (HABC) algorithm is used for selecting an optimal route based on the parameters such as reliability, bandwidth effectiveness, average path loss, and average transmission latency. The overall framework for the proposed method is depicted in Fig. 1.

Fig. 1

Overall framework for proposed UC-HRP Architectures.

Fig. 2

Flow diagram of proposed Hybrid ABC method.

3.1 Cluster head formation

In this proposed method the fuzzy-ELM approach is used to initialize the cluster based on parameters such as Doppler spread, path loss, noise, and multipath. In factual humanity formation struggle, harmonize to the altered weights; the influences of the guidance spears must be changed. A deposit M of labeled guidance spears with allied fuzzy membership. $(z_{1}, k_{1}, m_{1}), \dots \dots, (zF, kF, mF)$

Every guidance spear z_d is specified as a marquee k_d and a fuzzy membership m ≤m_d≤1 by adequately by small σσ > 0. This fuzzy membership m_d is the approach of the equivalent spears z_d about solitary cluster form because 1/2||ɛ_d|| is a measure of error, thus 1/2||ɛ_d|| is evaluated of fault through similar weighting. This formulation problem for the restricted peerless established fuzzy ELM preserve be phrased as

Subject to: $p_{(z_{d})} β = K_{d}^{K} - ɛ_{d}^{K}, d = 1 \dots \dots \dots F$

Fuzzy Extreme Learning Machine (FELM) is equal to cracking the next twin optimization problem, according to the Karush–Kuhn–Tucker (KKT) Formula:

$\begin{matrix} L_{FELM} = \frac{1}{2} {| | β | |}^{2} + G \frac{1}{2} \sum_{d = 1}^{N} m_{d} | | ɛ_{d} | |^{2} \\ - \sum_{d = 1}^{F} \sum_{l = 1}^{M} \propto_{d, l} (p (z_{d}) β_{l} - k_{d, l} + ɛ_{d, l}) \end{matrix}$ (1)

The Karush–Kuhn–Tucker (KKT) theorem’s corresponding optimality conditions are as follows: $\frac{\partial L_{FELM}}{\partial β_{l}} = 0 \to β_{l} = \sum_{d = 1}^{F} \propto_{d, l} p (z_{d})^{K} \to β = P^{K} α$ (2)

Where K=[k₁, k₂, … . , k_F] ^K, and P^K is the Moore-Penrose widespread contrary to environment H. When P^KP remains non-singular

P^K = (P^KP) ^-1P^K, or when PP^K is non-singular $\frac{\partial L_{FELM}}{\partial ɛ_{d}} = 0 \to α_{d} = {Gm}_{d} ɛ_{d,} d = 1, \dots . ., F$ (3) $\frac{\partial L_{FELM}}{\partial α_{d}} = 0 \to p (z_{d}) β - K_{d}^{k} + ɛ_{d}^{K} = 0, d = 1, \dots \dots ., F$ (4)

By substituting (2) and (3) into (4), the equation can be equivalently written as $(\frac{M}{G} + {PP}^{K}) \propto = K$ (5)

Where K= $[\begin{matrix} k_{1}^{K} \\ . \\ . \\ k_{F}^{K} \end{matrix}]$ and the fuzzy matrix $M = {\begin{matrix} [\begin{matrix} \frac{1}{m_{1}} & 0 & 0 & \dot{} & \dot{} 0 \\ 0 & \frac{1}{m_{2}} & 0 & \dot{} & \dot{} 0 \\ . & \dot{} & \dot{} & \dot{} & \dot{} \\ \dot{.} & \dot{} & \dot{} & \dot{} & \dot{} \\ 0 & 0 & 0 & \dot{} & \dot{} \frac{1}{m_{F}} \end{matrix}] \end{matrix}}_{F \times F}$ $β = P^{K} {(\frac{M}{G} + {PP}^{K})}^{- 1} K$ (6)

Thus, the inputs with the different fuzzy matrices can make different contributions to the learning to the output weights b. Then, the output function of FELM

$F (z) = p$

$(z) β = p (z) P^{K} {(\frac{M}{G} + {PP}^{K})}^{- 1} K$ (7)

For binary cluster formation problems, FELM wants simply a unique productivity node, and the output function is $F (z) = sign (p (z) p^{K} {(\frac{M}{G} + {PP}^{K})}^{- 1} K)$ (8)

For m-cluster form belongings, the predicted cluster form marquee of tough spears is the index digit of the output node which has the peak output cost. $Label (Z) = argmax f_{l} (z)$ (9)

Similarly, concordant to altered appeals, the fuzzy matrix M can be located agreeably to solve different problems.

3.2 Cluster Centre Cluster Head Selection (C³HS)

Cluster Centre Cluster Head Selection (C³HS) means the clustering technique selects optimal cluster heads. The C3HS design was centered on clustered sensors. The fuzzy-ELM algorithm was used to identify the network’s clusters. The suggested C3HS is a different CH selection (CHS) procedure designed to increase network lifetime while minimizing energy consumption. Unlike other procedures in the literature, this algorithm necessitates two-step clustering. That is, a distinct separation of the nodes in a cluster that is initially identified for each cluster remains after the network has been clustered using fuzzy-ELM methods. The nodes of this particular subgroup, termed “CC” (cluster-centered), are given precedence during CH selection. The group of nodes outside of the cluster known as OCC (Out of Cluster Centred) may only become CH when every node in the CC has terminated. Below, we’ll go over the specifics of the CC and OCC resolution courses. The new approach introduced by

C³HS will improve the efficiency of CH selection by increasing the prominence of the central nodes. Let us precisely define the proposed method. To better understand the following equations, let us first describe the representations:

F: The number of clusters used in the C³HS algorithm.

G_f: C³Hs algorithm makes use of the Centre of the F^th cluster.

$(z_{g}^{f}, n_{g}^{f}) :$ C³Hs algorithm makes use of the location of the F^th cluster center.

$(z_{j}^{f}, J)$ : C³Hs algorithm makes use of the location of the M^thcenter of the F^th cluster.

n_f: C³HS algorithm makes use of the entire number of sensors in the F^th cluster.

$J_{j}^{f}$ : C³HS algorithm makes use of the space among theM^th the node of cluster F and the F^th cluster center.

k_f: C³HS algorithm uses the mean expanse to the F^th cluster center, which is the range for CC_K .

GG_f: TheF^th cluster’s cluster-centered nodes are utilized by the C³HS algorithm.

QGG_f: C³HS algorithm uses cluster-centered nodes for the F^th cluster.

P_f: The F^ththe universal set is the same as the F^th cluster defined by fuzzy-ELM used in the C³HS algorithm. Now that the calculations are established, $G_{f} = \frac{1}{h_{j}} \sum_{k = 1}^{h_{j}} z_{j}$ (10) $J_{j}^{f} = \sqrt{(z_{,}^{f} - z_{g}^{f}) + {(N_{,}^{f} - N_{g}^{f})}^{2}}$ (11) $k_{f} = \frac{1}{h} \sum_{j = 1}^{h_{f}} J_{j}$ (12) ${GG}_{f}^{,} = (z_{j}, N_{j}) \in P_{f} | J_{j} ⩽ k_{f}$ (13) ${QGG}_{f} = {GG}_{f}^{,} = (z_{j}, N_{j}) \in P_{f} | (z_{j}, N_{j}) ∄$ (14)

The CC boundary is a spherical zone with the area k_j which corresponds near the regular Euclidean reserve of altogether the devices to the F^th cluster center as shown in Equation (13) OCC_j remains the complement of the standard CC_j. It contains the set of sensors of the F^th a cluster that is not in CC_j as specified in Equation (14).

3.3 Optimal Routing based on Hybrid ABC algorithm

The hybrid optimization method also categorizes the most valuable barriers to the creation of UWSNs. The optional procedure’s role is to regulate which approaches take the superlative risks of success. Uniform the requirement used for straight reaction software design spirit represent the solution of the upgrade process, then the determination remains fair for the routes chosen aimed at its steering in UWSN. The number of energy ports in the nodes, the route’s quantity, and the route’s convenience determine the route’s fitness. For optimal route selection, the hybrid ABC-PA algorithm is used.

The ABC algorithm’s employed bee routinely expenditures earlier route data to find a new route and stakes out the route evidence with the observer moth. Established on the info collective by the employed bees, onlooker bees will enter the node and find a new route. The task of scout bees is to find innovative and treasured routes, then they will pursue casually near the node.

Step 1: Change the procedure’s constraints and people’s sizes-V equals the scope of PA, and PA matches one-third of V. Conclude the number of employed bees (N c_d), observer bees (N a_c), and detective bees (Ne_d), as well as the number of duplications L, switch amount bound, region search HL, and other parameters.

Step 2: Encrypt and compute the suitability worth of each distinct. Exhausting a firm non-dominated organization process, examine the most recent people’s Pareto optimal path set. Permitting the Pareto position and crowding expanse of the separate gene in the residents, the following two values must be calculated: The Pareto abundant I rank of each solution, as well as its crowding distance Zj If the Pareto level of the two genes is the same, the gene with the lower gathering reserve is chosen.

Step 3: Check whether the next merging criteria are met; if any of them are uncertain, the optimization process comes to an end; otherwise, move on to Step 4. The sum of the iterations reaches a specified higher destined, which is 20×h×f in this paper. There is no new best path for producing continuous P.

Step 4: (Employed bee phase): The employed bees’ goal is to locate a superior route supply close to the current route. The efficiency of the algorithm is increased by combining the traits to introduce the subsequent crossover and mutation.

Step 4.1: Crossover. Border operations are divided into two stages: operation sequence (OS) and machine allocation (MA).

Step 4.2: Mutation. The execution process for the MA mutation, which uses a method of choosing the shortest processing time, is as follows:

Step 4.2.1: Create k random numbers at random, where D is the whole amount of actions.

Step 4.2.2: Choose k positions at random from the gene’s MA;

Step 4.2.3: For the process of agreeing to collectively site, the machine with the thorough handling period is chosen for replacement after the regular of existing machines. Doubt the device with the direct machining time is the unique MA, choose the device with the second-shortest machining time.

Step 5: (Onlooker bee phase): The primary responsibility of the observer bee is near collect the route material mutually by the employed bees. The hybrid Ant Bee Colony algorithm’s onlooker bee searches in a casual manner, besides there is no assurance that a recovering route determination is found later in the search.

Step 5.1. Neighbourhood Structure design: Using an adaptive variable neighborhood search strategy, the computer can choose a neighborhood with a high search efficiency effectiveness of the search method suggested in this study.

Step 5.2. Adaptive neighbourhood search: VNS represents four buildings in the neighbourhood. The neighbourhood search process’s spectator bee phase is the pseudo-code used for the best solution F* in PA.

Step 6: (Scout bee phase): If the route consumes not remained efficient once the pursuit space edge group, it is superfluous, the detective bee approach is triggered to produce a new route, and the optimization is constant. In this study, we employ the following techniques to build a new route during the scout bee phase if the route is inefficient once the edge group is, in order to increase the variety of the population: effectiveness of the search method suggested in this study.

Step 6.1: A non-dominated clarification s₀ from PA and an optimal clarification in PA result from complete crossover and mutation, resulting in a new clarification s₀.

Step 6.2: If s_k is better than s_′, s_k replaced s_′. The examine performance of the perfect ant colony process can be developed through this process, and the previous conjunction of the system can be avoided to a large extent.

Step 7: Finally, take the Pareto optimal path. The ABC algorithm’s overall flow chart with PA.

4 Results and discussion

This section discusses the proposed Underwater Clustering-based Hybrid Routing Protocol (UC-HRP). For implementation, the Network Simulator (NS2) with 4 GB of RAM and an Intel Core processor was chosen. The performance of our proposed UCH-RP protocol is evaluated in this segment. The default model parameters are shown in Table 1. By changing the number of rounds to 150, 300, 450, 600, 750, and 900, the simulation is run with 1000 nodes. are spread out at random in a 500×500-meter area. Here, the initial energy was set to 100 J for execution, and the transmission range was set to 15 m. Additionally, an underwater MAC frequency called MAC 950.15-DYNAV is used. In our Aqua-Sim-based simulations, the MAC layer interference is successfully managed because each node keeps a local copy of incoming packets, and collision of packets is detected using the local difference in received power levels. The data was collected from 1.8 km of ocean depth, and the water’s temperature ranges from 30°C to 500°C. The simulation duration is lastly set to 900 s.

Table 1
Simulation Parameters

Parameters Values

Data-aggregation factor 0.9

Distance between the layer of nodes 250 m

High Transmission power 1.9 Watts (W)

Total no. of nodes 1000

Initial energy 100 (J)

Channel types Wireless

Simulation time 900 s

Average ocean depth 1.8 km

Standard MAC 950.15-DYNAV

Cluster diameter 350m

Frequency 50 Kilohertz

Water Salinity 58

Water temperature 3⁰c to 50⁰c

Transmission Range 15m

Packets size 84 bytes

Network size 500 m×500 m

Parameters	Values
Data-aggregation factor	0.9
Distance between the layer of nodes	250 m
High Transmission power	1.9 Watts (W)
Total no. of nodes	1000
Initial energy	100 (J)
Channel types	Wireless
Simulation time	900 s
Average ocean depth	1.8 km
Standard	MAC 950.15-DYNAV
Cluster diameter	350m
Frequency	50 Kilohertz
Water Salinity	58
Water temperature	3⁰c to 50⁰c
Transmission Range	15m
Packets size	84 bytes
Network size	500 m×500 m

4.1 Performance metrics

The following variables for the proposed model were used in a performance evaluation:

Throughput: The collective information frequency achieved in the destination nodes for the entire network is defined as “network throughput.”

Packet Delivery Ratio: The packet delivery ratio is the proportion of envelopes successfully established by every drop on the water’s surface, as determined by each sensor node in the system.

Energy Consumption: The total amount of energy consumed by nodes as a result of effective information envelope broadcasting

End-to-End Delay: Time taken for data to arrive late it takes a data envelope to spread any terminus or sink node from a basis node.

Residual Energy: While transmitting, receiving, and aggregating data, a sensor node uses energy. The energy that remains is referred to as “residual energy."

Alive nodes: After a certain number of data-gathering rounds, the number of alive nodes remains. When there are more alive nodes, perform better.

Network Lifetime: when the first node in the network expires due to exhaustion of its energy.

4.2 Comparative Analysis

In this section, the simulations to see how well the proposed UCHRP methodology performs. The proposed protocol is compared to the MERP [21], ACUN [26], and ANC-UWSNS [22] protocols. The UCHRP protocol is evaluated using a variety of metrics, including lifetime, network lifetime, energy consumption, average delay, network performance, and package delivery ratio.

Figure 3 shows the performance analysis of the proposed UCHRP network using other methods. In 900 nodes, the suggested technique outperformed other approaches with a Network Lifetime value of 1400 rounds. Following research, MERP demonstrates subpar Lifetime Performance. When compared to the MERP model, the average network lifetime is 17.5%, while the ACUN model has a typical network lifetime of 20.0% and the ANC-UWSNS model has a typical network lifetime of 39.0%, with the proposed model having an overall higher network lifetime of 45.70%.

Fig. 3

Network lifetime Vs Number of nodes.

Figure 4 depicts the difference in energy consumption between ACUN and MERP, ANC. As previously stated, terminated packet broadcasts in ACUN consume a lot of energy. Figure 4 illustrates how much less energy UCHRP uses in comparison to conventional practices. By equally distributing energy throughout the network and eliminating hotspots through unequal clustering, the network’s energy usage may be decreased. The proposed UCHRP bases its routing choices on a trust model that takes into account a node’s remaining energy. As shown in Fig. 4, the energy consumption of ACUN increases in lockstep with the number of nodes. As the number of nodes increases, more nodes participate in furthering, increasing global energy consumption. Because MERP allows a limited number of nodes to advance the data envelopes, its energy consumption is lower than that of ACUN.

Fig. 4

Energy Consumption Vs Number of nodes.

The residual energy is compared to existing and proposed models in Fig. 5. The total residual energy of the network is depicted in the figure, and it is clear that the UCHRP had a higher total residual energy, meaning that the network could live for a longer time. As seen in Figure, when compared to other algorithms, the UCHRP significantly decreased the residual energy deviation. The leftover energy deviation was more evenly distributed in UCHRP compared to other protocols because the nodes with more remaining energy were chosen as the CH more frequently than the nodes with less remaining energy. In comparison to the MERP model, the average residual energy is 18.4%. The ACUN model has an average residual energy of 18.32%, the ANC-UWSNS model has an average residual energy of 19.15%, and the proposed UCHRP model has an overall residual energy of 35.86%.

Fig. 5

Residual Energy Vs Number of nodes.

In Fig. 6, the routines of all three procedures are compared in terms of end-to-end delay. Delay in networks is a common scenario however efficient the network. It is impossible to predict how it would behave in later stages. The way the network reacts to changes in the topology to minimize the delay is used to measure its performance. According to the figures, ACUN has a long end-to-end delay because each node in ACUN rules the set for a specific amount of time based on the complexity change in its earlier stage. Furthermore, the ACUN delay increases as the number of nodes increases. Because each node grips the packet, adding more nodes causes additional delay. The MERP, ANC, and UCH-RP end-to-end delays are comparable.

Fig. 6

End-to-End delay Vs number of nodes.

Figure 7 compares the performance of the UCHRP model to other approaches such as ANC-UWSNS, MERP, and ACUN percentage of packet delivery. The proposed approach outperforms current protocols in terms of PDR. Based on the graph, the proposed scheme displayed the highest PDR for all networks up to 88.5%. The figure shows that the packet delivery ratio of all methods increases with the number of nodes. Due to optimal cluster head selection and the formation of more stable clusters, we can justify the proposal’s superior performance in PDR.

Fig. 7

Packet delivery ratio Vs number of nodes.

Figure 8 shows the throughput for different node counts. It provides a measure of how quickly data can be transmitted through a network. WSN aims to increase throughput while extending network lifetime and dependability. The fundamental reason for the increase in throughput is the shortage of energy resources. Increased energy results in a significant rise in throughput. The suggested Underwater Clustering-based Hybrid Routing Protocol (UC-HRP) has a higher average throughput than the current ANC-UWSNS [12], ACUN [14], and MERP [11] approaches. The proposed method outperforms the current methodologies in terms of average throughput (25.34%, 23.7%, and 21.6%). The proposed strategy outperforms the existing methods by 22.43%, 16.4%, and 13.5%, respectively, as shown in Fig. 6.

Fig. 8

Throughput Vs Number of nodes.

Figure 9 compares the proposed UC-HRP method to the existing ANC-UWSNS [12], ACUN [14], and MERP [11] methods. Figure 9 shows the performance of the suggested optimal CHS in the WSN system about the number of alive nodes using the proposed technique. As shown in Fig. 9, The suggested UC-HRP approach produces a lot of live nodes. It is discovered that the UC-HRP strategy increases the percentage of network alive nodes when compared to previous approaches. In comparison to earlier strategies, it is seen that the UC-HRP scheme increases the percentage of alive nodes in the network.

Fig. 9

Alive nodes Vs Number of nodes.

Figure 10 depicts a high system energy consumption between rounds 0–900 in all transmitting organizations due to the considerable number of faulty networked sensor nodes. Uneven links between sensor nodes are the cause of a sizeable fraction of the ACUN network’s sensor nodes quickly expiring, which led to excessive energy consumption. There is no acceptable, robust backup tool available to ACUN to deal with or recover from this specific issue. As evidenced by the imitations, ANC’s energy consumption organization outperforms ACUN’s. However, the UC-HRP concert outperforms ACUN and ANC due to very steady connections between sensor nodes and round-to-round node failure behavior problems.

Fig. 10

Energy Consumption between Numbers of Rounds.

Figure 11 shows that when rounds are created, the ACUN in each transmitting scheme increases collectively with how many sensor nodes were used overall and how many rounds there were. It should be noted that every routing scheme strives to reach its stable state to produce data packets at the highest rate feasible between the range of round numbers 0–900. To achieve the highest level of network constancy, UC-HRP outperforms ACUN, ANC-UWSNS, and MERP. Figure 12 shows this at the start of each round. The data package is temporarily provided by each acceptance node. It is used in UC-HRP to select forwarder nodes for re-transmission, i.e., to increase or decrease a node’s communiqué radius.

Fig. 11

Packet delivery ratio between the Number of Rounds.

Fig. 12

Network lifetime between the Number of Rounds.

A superior limit means that the most distant lower broadcast delay and minimizes the number of nodes engaged in re-transmission by choosing the node as the forward node Moreover, the UC-HRP pattern-matching data-aggregation device moderates package scope, so combined machines exclude total network energy and spread network lifetime.

5 Conclusion

In this work, a novel underwater clustering-based hybrid routing protocol (UC-HRP) has been proposed to address these issues. The overall three phases make up the process. First, the fuzzy-ELM technique is employed to initialize the cluster based on parameters such as Doppler spread, path loss, noise, and multipath. Using Cluster Centre Cluster Head Selection (C³HS), the cluster head is selected in the second phase based on the link quality, distance, node degree, and residual energy. The third phase is when Hybrid Artificial Bee Colony (HABC) algorithm is used for selecting an optimal route based on the parameters such as reliability, bandwidth effectiveness, average path loss, and average transmission latency. The effectiveness of the suggested UC-HRP approach is assessed using a variety of parameters, including the network lifetime, packet delivery ratio, alive nodes, and energy consumption. Simulation results demonstrate that UC-HRP is an effective network routing strategy. By using the suggested method, the network lifetime is increased by 14.03%, 16.25%, and 18.34% better than ACUN, ANC-UWSNS, and MERP respectively. Future research will concentrate on reducing transmission uncertainty by using effective transmission overhearing techniques in a wireless open medium. Moreover, there is an intrusion detection feature and fuzzy handling of constraints partial information more efficiently might be added to this system.

References

Herman

J.R.

and Celarier

E.A.

, Earth surface reflectivity climatology at 340–380nm from TOMS data, Journal of Geophysical Research: Atmospheres 102(D23) (1997), pp. 28003–28011.

Wallace

A.R.

, The Geographical Distribution of Animals: With a Study of the Relations of Living and Extinct Faunas as Elucidating the Past Changes. e-artnow. 2020.

Khan

, Packet Reliability Using Cooperative Routing in Underwater Wireless Sensor Networks (Doctoral dissertation, Department of Electrical Engineering A thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Information Technology (MS IT) In School of Electrical Engineering and Computer Science, National University of Sciences and Technology). 2021.

Bradley

J.T.

, comprising about 54,100 square miles of land area and is surrounded on three sides by the waters of the Atlantic Ocean and the Gulf of Mexico, Count Climates of the States: Eastern States, plus Puerto Rico and the US Virgin Islands 1 (1974), 45.

Yearsley

J.M.

and Sigwart

J.D.

, Larval transport modeling of deep-sea invertebrates can aid the search for undiscovered populations, PLoS One 6(8) (2011), e23063.

Blasiak

, Wynberg

, Grorud-Colvert

, Thambisetty

, Bandarra

N.M.

, Canario

A.V.

, da Silva

, Duarte

C.M.

, Jaspars

, Rogers

and Sink

, The ocean genome and future prospects for conservation and equity, Nature Sustainability 3(8) (2020), 588–596.

Akyildiz

I.F.

, Pompili

and Melodia

, Underwater acoustic sensor networks: research challenges, Ad hoc networks 3(3) (2005), pp. 257–279.

Khasawneh

A.M.

, Kaiwartya

, Abualigah

L.M.

and Lloret

, Green computing in underwater wireless sensor networks pressure centric energy modeling, IEEE Systems Journal 14(4) (2020), 4735–4745.

Abiodun

O.I.

, Jantan

, Omolara

A.E.

, Dada

K.V.

, Mohamed

N.A.

and Arshad

, State-of-the-art in artificial neural network applications: A survey, Heliyon 4(11) (2018), pp. e00938.

10.

Darehshoorzadeh

and Boukerche

, Underwater sensor networks: A new challenge for opportunistic routing protocols, IEEE Communications Magazine 53(11) (2015), pp. 98–107.

11.

Sirvent

J.A.L.

, Realistic acoustic prediction models to efficiently design higher layer protocols in underwater wireless sensor networks. Diss. universidad miguel hernandez de elche. 2012.

12.

Faheem

, Tuna

and Gungor

V.C.

, LRP: Link quality-aware queue-based spectral clustering routing protocol for underwater acoustic sensor networks, International Journal of Communication Systems 30(12) (2017), e3257.

13.

Gola

K.K.

and Gupta

, Underwater sensor networks:‘Comparative analysis on applications, deployment and routing techniques’, IET Communications 14(17) (2020), 2859–2870.

14.

Jaruwatanadilok

, Underwater wireless optical communicationchannel modeling and performance evaluation using vector radiativetransfer theory, IEEE Journal on Selected Areas inCommunications 26(9) (2008), 1620–1627.

15.

Sandifer

P.A.

and Walker

A.H.

, Enhancing disaster resilience by reducing stress-associated health impacts, Frontiers in public health 6 (2018), 373.

16.

Ullah

, Khan

, Zareei

, Ali

, Khattak

H.A.

and Din

I.U.

, Energy-effective cooperative and reliable delivery routing protocols for underwater wireless sensor networks, Energies 12(13) (2019), 2630.

17.

Ashraf

, Saleem

and Ahmed

, Sagacious communication link selection mechanism for underwater wireless sensors network, Int J Wirel Microw Technol 10(4) (2020), 22–33.

18.

Yang

, Dai

and Wei

, Challenges, threats, security issues and new trends of underwater wireless sensor networks, Sensors 18(11) (2018), 3907.

19.

Ghanem

, Mansoor

A.M.

and Ahmad

, A systematic literature review on mobility in terrestrial and underwater wireless sensor networks, International Journal of Communication Systems 34(10) (2021), e4799.

20.

Yue

Y.G.

and He

, A comprehensive survey on the reliability of mobile wireless sensor networks: Taxonomy, challenges, and future directions, Information Fusion 44 (2018), 188–204.

21.

Faheem

, Ngadi

M.A.

and Gungor

V.C.

, Energy efficient multi-objective evolutionary routing scheme for reliable data gathering in Internet of underwater acoustic sensor networks, Ad Hoc Networks 93 (2019), 101912.

22.

Khan

M.F.

, Bibi

, Aadil

and Lee

J.W.

, Adaptive node clustering for underwater sensor networks, Sensors 21(13) (2021), 4514.

23.

Luo

, Chen

, Wu

and Yang

, A survey of routing protocols for underwater wireless sensor networks, IEEE Communications Surveys & Tutorials 23(1) (2021), 137–160.

24.

Coutinho

R.W.

and Boukerche

, June. PCon: A Novel Opportunistic Routing Protocol for Duty-Cycled Internet of Underwater Things. In 2019 IEEE Symposium on Computers and Communications (ISCC) (2019), (pp. 1-6). IEEE.

25.

Faheem

, Butt

R.A.

, Raza

, Alquhayz

, Ashraf

M.W.

, Shah

S.B.

, Ngadi

M.A.

and Gungor

V.C.

, QoSRP: A cross-layer QoS channel-aware routing protocol for the Internet of underwater acoustic sensor networks, Sensors 19(21) (2019), 4762.

26.

Wan

, Liu

, Ni

and Xu

, An energy-efficient multi-level adaptive clustering routing algorithm for underwater wireless sensor networks, Cluster Computing 22(6) (2019), 14651–14660.

27.

Sun

, Zheng

, Han

, Li

and Yin

, Adaptive clustering routing protocol for underwater sensor networks, Ad Hoc Networks 136 (2022), 102953.

28.

Jouhari

, Ibrahimi

, Tembine

and Ben-Othman

, Underwater wireless sensor networks: A survey on enabling technologies, localization protocols, and internet of underwater things, IEEE Access 7 (2019), 96879–96899.

29.

Fattah

, Gani

, Ahmedy

, Idris

M.Y.I.

and Targio Hashem

I.A.

, A survey on underwater wireless sensor networks: Requirements, taxonomy, recent advances, and open research challenges, Sensors 20(18) (2020), 5393.

30.

Wei

, Guo

, Wang

and Qiu

, Reliable data collection techniques in underwater wireless sensor networks: A survey, IEEE Communications Surveys & Tutorials 24(1) (2021), 404–431.

31.

Al-Andoli

, Cheah

W.P.

and Tan

S.C.

, Deep autoencoder-based community detection in complex networks with particle swarm optimization and continuation algorithms, Journal of Intelligent & Fuzzy Systems 40(3) (2021), 4517–4533.

32.

Al-Andoli

M.N.

, Tan

S.C.

and Cheah

W.P.

, Distributed parallel deep learning with a hybrid backpropagation-particle swarm optimization for community detection in large complex networks, Information Sciences 600 (2022), 94–117.