IEDA-HGEO: Improved Energy Efficient with Clustering-Based Data Aggregation and Transmission Protocol for Underwater Wireless Sensor Networks

Abstract

With the emerging technology in underwater wireless sensor networks (UWSN), many researchers are undergoing this field since it cannot maintain the batteries and recharge them manually. Network duration should be taken into account because they can easily be recharged by a non-conventional resource like solar energy. When coming to the data collection process, clustering is an effective method to construct vitality effective UWSNs. The clustering properties of UWSNs differ from those of terrestrial wireless sensor networks (TWSNs) due to the sparse deployment of nodes as well as the dynamic nature of the channel. This paper proposes improved efficient data aggregation in a Hexagonal grid with energy optimization (IEDA-HGEO) protocol for effective data transmission with an optimal clustering process. It is further compared with ERP²R n energy-efficient routing protocol and EGRC (Energy-efficiency Grid Routing based on 3D Cubes). The three techniques mentioned above are specifically examined for their applicability to underwater communication, and their performance is compared in terms of energy consumption, efficiency, throughput, packet delivery ratio, and delay. The proposed method achieved the following metrics: delay 41%, energy consumption 48%, efficiency 95%, throughput 95%, and PDR 92%.

1. Introduction

UWSNs have become a key piece of method for underwater monitoring and exploration, including scientific, commercial, and military applications, over the past ten years [1,2]. UWSNs have various advantages over their remote sensing competitors that can deliver localized and more accurate data collecting. They can also use a wider range of sensors, such as chemical, temperature, light, and motion sensors, among others. Traditional underwater instrumentation equipment is being replaced by UWSN technology. In the past, large sensor nodes with data-storage capabilities have been physically placed in the target space below the water. For the duration of the operation, each node runs autonomously to collect readings in accordance with a predetermined program [3]. SNs are picked up at the conclusion of the operation, and the information gathered is recovered and processed. To relay real-time data to an offshore or even on-shore control station for immediate analysis, UWSN technology gives underwater sensor nodes networking capabilities. The underwater sensor network deployment can be controlled interactively by sending control signals from the control station to underwater SNs via a communication channel. In comparison to conventional instrumentation methods, UWSNs provide significant benefits [4].

Because UWSN-SN are powered by batteries, which are challenging to replace or recharge in aquatic environments [5], energy conservation is a serious challenge. A basic research problem is the creation of routing protocols that are reliable, scalable, and energy-efficient in these networks. Due to the fact that the majority of data forwarding protocols currently in use were created for stationary networks, they cannot be used directly with ground-based sensor networks [6].

Figure 1 denoted the basic clustering formation in the UWSN network, whereas two types of sinks, namely onshore and surface sinks, are there to transfer the collected information to the satellite. Moreover surface node, cluster head and underwater sink node have interconnection with each other to collect the data [7]. An autonomous underwater vehicle is used here for moving purposes from one cluster to another cluster or contact with the surface node in case of any emergency during the transmission of packets [8].

The contribution of this research is as follows:

• To take into account the energy consumption and travel time of AUVs, we propose a data-gathering protocol based on AUV path planning.
• To propose improved efficient data aggregation in IEDA-HGEO protocol for effective data transmission with optimal clustering process.
• It is further compared with ERP²R n energy-efficient routing protocol and EGRC.

The paper is organized as follows: Section 1 gives a brief explanation of UWSN and its applications, difficulties during the transmission process, existing energy-efficient protocols, and the clustering process in the network. Section 2 explained some surveys about existing protocols in the clustering process of UWSN. Section 3 explains the proposed methodology for efficient clustering and data transmission process. Section 4 gives the detailed structure of results by comparing them with existing techniques. Finally, Section 5 ends with a conclusion and future work.

2. Literature Review

A new distributed energy-aware routing protocol is DUCS [9]. It is specifically made for long-term, non-time-critical aquatic monitoring applications utilizing UWSNs without GPS support and random node mobility. A suggested Underwater Positioning Scheme (UPS) in [10] has ordinary nodes listening exclusively to the signals from beacon nodes, and after receiving four beacon messages, the time difference is converted into a range distance. The authors of [11] employ Cayley–Menger to use a mobile beacon node to find the coordinates of sensor nodes. The combined radio and acoustic signals, which are immune to multipath fading, are used to calculate the separation between nodes. In [12], scientists put out a number of mobile node-based strategies, like AUVs, to lessen the impact of unequal energy use. These techniques involve an AUV travelling around the network on a tour path and stopping at a designated location known as a tour point to obtain the gathered data from static nodes in a neighbourhood. For WSNs, Selvi et al. introduced the UCAPN method, which lengthens network lifespan [13]. To balance node energy consumption as well as extend network lifetime, UCAPN groups SNs into clusters of varying sizes. Information from non-cluster-head nodes is first transmitted directly to the closest cluster-head, and then it is transmitted to the sink node. With the help of a control method wherein BS regulates the number of CHs and CHs regulate cluster members, Gulnaz Ahmed et al. [14] presented a MOCHs selection for WSNs. This technique provides robust clustering while addressing the issue of backward transmission. By including energy-collecting techniques, this protocol can be made much better [15]. According to the node density, Khan et al. established a protocol in [16] that can adapt to three different types of networks. Underwater sensors are allocated cubic areas in [17] and placed regionally inside such a framework. To avoid void nodes, another work in [18] is introduced. Every second hop of the packet in this work is examined to see if the node’s status is void or not. A void-aware pressure routing (VAPR), which addresses the void problem in this class of greedy routings, is proposed in [19]. [20] makes a proposal for Clustered Vector-Based Forwarding (CVBF) routing protocol. CVBF is made for areas of seawater that are both sparse and dense. According to the authors, CVBF enhances the data delivery ratio as well as decreases end-to-end delay. However, in these protocols, cluster reconstruction for switching CHs by AUVs is repeated until their missions are accomplished to balance the energy consumption among sensor nodes. AUVs can, therefore, run out of energy more quickly before completing their missions as a result of the foregoing constraints’ significant energy consumption. What is more, the CHN determination conspires, and the information transmission process is further developed in the PE-Filter convention. Nonetheless, the conventions in [21,22] are intended for TWSNs, and they ought to be altered for UWSNs. Wang et al. took on an energy-proficient lattice directing in light of 3D solid shapes (EGRCs) for UWSNs, where the organization is separated into heaps of little blocks, and each 3D square is viewed as a group [23]. In addition, the EGRC convention streamlines the CHN choice and further develops the quest cycle for the following bounce hub. Be that as it may, the EGRC does not present the detail of the information combination instrument as the overt information repetitiveness might exist and ought to be diminished. In [24], a submerged bunching convention based on the fluffy c means and the moth-fire advancement (FCMMFO) was proposed to upgrade the presentation of UWSNs. Be that as it may, the multi-jump steering way has not been advanced in [25]. Ahmed et al. presented a submerged grouping convention as indicated by repetitive transmission control (RTC), which takes out the overt information repetitiveness at the CHN level and at the area head level [26].

3. Proposed Methodology

Improved efficient data aggregation in Hexagonal grids with energy optimization (IEDA-HGEO) in UWSNs is proposed. The features of our suggested protocol are as follows: The first is to cover as much ground as possible. To cover most nodes, we divided our WSN into hexagonal areas. The best cell shape for clustering in a network is hexagonal. One node in each cell serves as the cell’s CH, and the cluster head is selected using a unique method that involves picking the node in smaller cells with the highest residual energy and closest location to the base station (BS) to handle data aggregation as well as transmission.

3.1. Deployment of Gateway Nodes

The Surface Gateways (SG) and Cluster head make up the underwater network (CH). SGs are static nodes fastened to surface-based buoys. They have electromagnetic and acoustic interfaces, respectively. Through an electromagnetic interface, SGs link the underwater network to the Internet. SG uses an acoustic interface to transmit and receive packets to the underwater network. One or more CHs may be connected to each SG. To relay packets from SGs to the active AUVs at the ocean floor and vice versa, CHs are positioned underwater at various depths is shown in Figure 2.

Due to the characteristics of an underwater communication network, the energy consumption scheme of UWSNs sets it apart from the energy-consuming strategy of WSNs. Equation (1) illustrates the numerical solution.

$$E(\mathrm{distance}, fc)= \mathrm{EnergyTh} (\mathrm{distance}, fa)$$

(1)

I here stands for a node’s frequency. It is equivalent to Equations (2) and (3)

$$fa={10}^{fa\left(fc\right)/10}$$

(2)

$$fa\left(fc\right)=\left[0.11 * \left({(fc)}^2*{\left(\frac{1}{\left(1+fc\right)}\right)}^2\right)+0.22\right.\left({(fc)}^2*{\left(\frac{1}{\left(1+fc\right)}\right)}^2\right)\cdots +n\left({(fc)}^n * {\left(\frac{1}{\left(1+fc\right)}\right)}^n\right)$$

(3)

Since the exposure zone of every sensing device is boundary-based, Equation (4) may be used to describe private network radial distance NR, node density ND, sector width RingW, hop limit MaxH, and the total number of participants Np.

$$Np ={{\displaystyle \sum }}_{i=0}^{n}NR * ND * MaxH * \mathrm{Ring} * W$$

(4)

From inside to outside, the UWSN’s connectivity is separated into various little ring sections or Ar1 to Arn. RingW/OT, where RingW is the network radius and OT is the ideal connection path length threshold for sensing devices, can be used to identify the maximum number of ring regions. Dlmax is a representation of the route’s maximum delay. The frequency of Yij is equal to 5 whenever nodes i and j are connected. If not, Yij’s total value is 0. We view the search for multi-hop routes as a multi-objective optimization problem, with the sole objective of locating the best route at the most reasonable price. Following Equations (5)–(7) shows how an objective formula (Fobj) is obtained. The proposed system architecture is shown in Figure 3.

$$\mathrm{Min} \mathrm{Fobj}={{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{Y}_{ij*Dij}$$

(5)

$$Dij = Dtij \mathrm{*} \mathrm{MinPower} \mathrm{*} Kij \mathrm{+} Drj$$

(6)

$${{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^{n}Dij Yij<\mathrm{Dlmax}$$

(7)

3.2. Cluster Head Selection

In this study, two processes of TCH and FCH selection are used to determine cluster heads. Using a tentative CH selection method based on EBT and Trust Value, the cluster head is chosen. To choose TCH, the node is given a timer, and trust values are calculated using the node’s total Trust value. The TCH is determined by the node with the highest trust value and energy. Additionally, the planned head count, node degree, and competition range are used to determine the final cluster head is shown in Figure 4.

3.2.1. TCH Selection Based on Energy-Based Timer (EBT)

According to the energy of each sensor node, a timer is allocated to the nodes. The nodes’ allotted waiting times are determined by energy. High energy nodes are supported in this phase as the potential new cluster head. If not, CH is the same node with maximum transmission energy. The model description for this timer with an energy basis is as follows. If each node can determine the average energy value of its neighbours, then let us say that node I has k neighbours. $S_i=\left\{i_{1, }{i}_{2, }{i}_{3, }\dots i_{n }\dots .i_{k, }\right\}$, where $i_n$ is the nth neighbour node. The average energy of node I is given by following Equation (8):

$$Average Energy\left(i\right)=\left\{\begin{array}{c}\frac{1}{k}{\displaystyle \sum }_{n=1}^{k}Energy\left(i_n\right) k>0\\ 0 k=0\end{array}\right.$$

(8)

TCH is chosen using an energy-based timer from the SNs. The following equation is utilized to evaluate the energy-based waiting time value for any sensor node ID$S_i$ by Equation (9).

$$Wait Time \left(s_i\right)=\frac{Avg Energy of s_i Neighbor node}{Energy of S_i}$$

(9)

According to the equation above, the waiting time gets shorter as node energy rises. Less waiting time will be allotted to the node with more energy. The potential Cluster Head is chosen to be this node. Other SNs leave CH selection when they receive this message before the start of their waiting time. Selected tentative CH broadcasts a tentative CH message within its broadcast range.

In a hierarchical fashion, Fn gathers data. Fn acts as a parent node and gathers information from its offspring. Distance (d) between parent and child nodes is equal to 2r_dt < d < 2R_f and rdt are communication and dth ranges of Fnby Equation (10).

$$S=\left\{C\in S\mid 2r_{dt}<d_{F\to C}<2R_f\right\}$$

(10)

The total amount of data gathered at Fnis if each C transmits a packet of l bits by Equation (11).

$$F_{\mathrm{data}}= {{\displaystyle \sum }}_i^{\left|S\right|}{\ell }_i$$

(11)

Following data gathering, we compute Fn’s energy consumption using Equation (12) as follows:

$$\left.E_{F_n}=e_s+e_t\times R\left({{\displaystyle \sum }}_i^{\left|S\right|}{\ell }_i+\ell \right)\varphi \right)+EDA+SNR$$

(12)

where R is the radius, φ is the data aggregation factor, es is sensing energy, et is electronic energy per bit during transmission, and EDA is data aggregation energy.

The volume of S1 is evaluated as follows by Equation (13),

$$\begin{array}{cc}{V}_{S1}& ={{\displaystyle \int }}_0^r{{\displaystyle \int }}_0^{\frac{\pi }{2}}{{\displaystyle \int }}_0^{2\pi }{\varrho }^2 \cos \varphi d\theta d\varrho \\ & =\pi {{\displaystyle \int }}_0^r(R^2-Z^2)dz\hfill \\ & =\frac{2}{3}\pi R^3\hfill \end{array}$$

(13)

Similarly, the volume of S2 is evaluated as follows by Equations (14) and (15),

$$V_{S2}=\frac{2}{3}\pi r^3$$

(14)

$$V_{ss}=V_{S2}-V_{S1}=\frac{2}{3}\pi \left(R^3-r^3\right)$$

(15)

The number of nodes in the bounded spherical segment is given by the following equation if the network’s node density is ψ by Equations (16) and (17).

$$N_{ss}=\frac{2}{3}\pi \psi \left(R^3-r^3\right)$$

(16)

$$E_{F_n-all}^{rcv}=e_r+\frac{2}{3}\pi \psi \left(R^3-r^3\right)\times \ell$$

(17)

3.2.2. TCH Selection Based on Trust Value

To identify node behaviour, node quality, and node services, Trust Value (TV) is utilized. Additionally, it is utilized for sensor node routing, reconfiguration, and data aggregation. It offers a method for calculating the reliability of SNs. In this study, the trust value is employed to gather information and keep track of various node activities. Trust value is utilized to locate potential CH along with Energy based Timer (EBT). To maximize the effectiveness of optimal CH selection, tentative CH selection uses the EBT and TV techniques. The trust value of nodes is determined using Equation (18) below.

$$Trust Value{\left(TV\right)}_{nodes}=\frac{N_{FD}}{N_{REC}}$$

(18)

where NFD stands for the quantity of packets forwarded, and NREC for the quantity of packets received. The node with the greatest trust value is chosen as the temporary cluster leader after the trust values of each individual node are determined. The final CH process is then carried out. The outcome of the TCH selection is finally returned by the EBT and the TV.

3.3. Data Transmission

All nodes handshake with one another and communicate their attributes with the predecessor node in the route to start the network’s functioning and route discovery. By calculating the distance between the two nodes using their position attribute obtained during handshaking, the neighbour node is selected. To find the route, each node needs to be handshaking with other nodes. When a route is discovered, node N1 is taken for the source node, and the other nodes are taken into account when selecting the optimal neighbour node. Currently, N1 is sending “HELLO” messages to every other node and is receiving “REPLY” messages in return. Figure 5 depicts the proposed data transmission flowchart.

Now, node N1 sends a message to node N2, who then begins sending “HELLO” packets to all of the other nodes and receiving “REPLY” messages from them. It is possible to obtain multipath in a similar way between the source and destination nodes.

The forwarder determines the candidate node’s distance from both the forwarder and the sink for each transmission. The priority rises with decreasing value. The highest priority node in each forwarder set is the first to be enlarged, followed by nodes whose distances are less than the node transmission radius (Cr), and finally, nodes in the Ci whose distances are less than the communication radius for all nodes already present in the cluster. The forwarder chooses the highest priority set as the next-hop forwarder set after performing the calculation. If the highest priority set is unable to transmit, the remaining forwarder set will only transmit the packets in order.

We introduce an EL to balance energy consumption; to do this, we divide the node’s original energy into m equal parts. If the current energy of a node is greater than i−1 parts and less than or equal to I parts, the node’s EL is 𝑖 (1 ≤ 𝑖 ≤ 𝑚). Each uw-sensor has 2 transmission options during the steady data transmission phase: sending the packets via MT or sending them straight to the uw-sink (DT). The Algorithm 1 represents the proposed IEDA_HEGO system.

Algorithm 1: Algorithm of IEDA-HGEO.

Require: $- \mathrm{Set}\_\mathrm{rn}(\mathrm{SN}, \left.S{N}_x\right)\to +{ \mathrm{USet} }_{-}rn\left(SN\right)$ $\mathrm{Initially}: HC\left(r{n}_x\right)=0$ ${ \mathrm{hop} }_{-}\mathrm{set}\left(r{n}_y\right)=0$ Ensure: $-S{N}_{Ry}:$$$$\mathrm{elected} \mathrm{as} RN$ $1: \mathrm{if} \mathrm{USet} rn\left(SN\right)=$null then $2: \mathrm{Elect} RN$ $(\mathrm{USet}\_\mathrm{rn}(\mathrm{SN}$$) \mathrm{is} \mathrm{a} \mathrm{set} \mathrm{of} \mathrm{SN} \mathrm{responding} \mathrm{to} SPK$$\mathrm{packets} \mathrm{to} NN$) $3: \mathrm{USet}\_\mathrm{rn} \left(SN\right)=$$\mathrm{USet} rn\left(SN\right)+{\mathrm{Set}}_{-}rn\left(S{N}_y,S{N}_x\right)$ $\mathrm{where} S{N}_y\in S{N}_{Ry}$$, \mathrm{with} \mathrm{a} \mathrm{value} \mathrm{of} Se{t}_{-}rn\left(S{N}_y,S{N}_x\right)$$\mathrm{in} RPK$packet $4: HC\left(r{n}_x\right)=HC\left(r{n}_x\right)+1$ $5: \mathrm{Repeat} \mathrm{Steps} 1 \mathrm{to} 5 \mathrm{to} \mathrm{elect} \mathrm{the} \mathrm{next} \mathrm{neighbour} \mathrm{node} \mathrm{for} S{N}_y$$$Dest_Node $6: \mathrm{hop}-\mathrm{set}\left(r{n}_y\right)=$$\mathrm{hop} \mathrm{set}\left(m_y\right)+1$ Return $5:$ end if 8: if no SPK control packets are returned, then 10: End if $11: \mathrm{for} \mathrm{every}SPK$$\mathrm{is} \mathrm{generated} \mathrm{from} S{N}_z$ do 15: compute Confidence_level using Equation (4) $16: \mathrm{USet} rn\left(SN\right)=$$\mathrm{USet} rn\left(SN\right)+$$\mathrm{Set} rn\left(S{N}_z,S{N}_x\right)$ $17: \mathrm{if} \mathrm{Confidence}\_\mathrm{level}\left(y\right)<$$\mathrm{Confidence}\_\mathrm{level}\left(z\right)$ ) then $18: \mathrm{Remove} S{N}_{Ry}$$\mathrm{from} \mathrm{USet} rn\left(SN\right)$ $19: \mathrm{Add} \mathrm{the} S{N}_{Rz}$$\mathrm{as} \mathrm{the} \mathrm{next} \mathrm{node} \mathrm{for} S{N}_x$ 20: end if $21: \mathrm{drop} \mathrm{this} SPK$ control packet 22: end for 23: $HC\left(S{N}_x\right)=HC\left(S{N}_x\right)+1$

4. Experimental Analysis

Settings for many parameters can be seen in

Table 1. Simulation parameters.

Parameters	Value
Network area	1000 × 2000 × 250 m3
Transmission radius	260 m
Number of nodes	1055
Size of data packets	1500, 2000, 3000, and 4000 bytes
Transmission power for data packet	2.8 (W)
RPK and SPK control packets size	13 bytes and 9 bytes
Transmission power for a control packet	1.5(W)

. Unlike earlier studies, the network is deployed in three dimensions. The region in three dimensions is 1000 × 2000 × 250 m³. Where BS is located at the ocean’s surface, there are 1055 sensor nodes. Nodes are spread out at random between 1000 and 250 metres. The 260-m transmission radius of SN allows for packet delivery in three to four hops. There are four different packet sizes: 1500, 2000, 3000, and 4000 bytes. RNs, CHs, and CCOs aid in achieving network connectivity. The RPK and SPK control packet sizes are set to 13B and 9B, respectively. In IEDA-HGEO, a levelling parameter of 0.85 has been selected for link quality. For data packets, the transmission power is set at 2.8 W, while for control packets, it is 1.5 W. The SNs can produce random packets, and the chain made up of the RNs, CHs, and CCOs should be used to forward these packets to the BS. In the simulations, 4300 packets are created with various payload counts, and energy usage is noted. The MATLAB simulator (R2021a) was used to carry out the experiments. It is a shorter version of “matrix and laboratory” and was invented by MathWorks. For research investigations, structural engineering, and a wide range of scientific issues needing precise numerical estimates, Matplotlib offers a complete answer.

This section evaluates and compares the performance of the proposed improved efficient data aggregation in IEDA-HGEO with that of two existing widely used routing protocols: ERP2R and EGRC. These two algorithms were chosen because they are well-known in the literature and share the same objectives as the method under consideration. The following measures are utilized to assess the performance of the EE-DHS: Energy consumption, energy efficiency, throughput, network lifetime, and delay are all factors is represented in

Table 2. Comparison of various parameters for the proposed protocol.

Parameters	ERP2R	EGRC	IEDA-HGEO [Proposed]
Energy consumption	55	51	48
Efficiency	88	92	95
Throughput	91	93	95
Packet delivery ratio	85	87	92
Delay	45	43	41

.

Energy consumption is the term used to describe how much energy each individual node uses to process information. The Equation Eelec* k (1) is used to describe energy consumption while data transmission through the packet to connected element ‘j’. The K bit of the data packet is receiving sensing element I while using energy which is represented by Equation (19).

Tx(x,y) = Eene*M + Eamp * d2(x,y) * M

(19)

Weight between connected nodes I and j are denoted by dij.

When one bit of energy is transmitted, Equation (20) is

etx(d)= pd1 + ptd*dn

(20)

where pd1 is the power dissipated by sending 1 bit of data, ptd is the energy required to transfer the nodes across a long distance

Figure 6 shows the energy consumption of our suggested EE-DHS, which results in 30% more resource conservation. The current methods, ERP2R and EGRC, use 50% and 65% of the resources, respectively. Data PDR, as shown in Figure 6, is calculated by dividing the total number of data packets generated by the source by the number of data packets that were successfully delivered. It is common for some nodes to have low energy levels while other nodes have high energy levels while the system is operating. Nodes with very little remaining energy must cut back on energy use because their operational lifetime is almost up. To determine the delivery ratio of data packets, packet delivery ratio trace files are post-processed. Specifically, the relationship between sent as well as received packets. The average rate of successfully delivered messages via a communication connection is known as throughput. This information may travel over a logical or physical link or go through a specific network node. Typically, the throughput is expressed in bits per second, though it can also be expressed in data packets per second.

Discussion: This work centres around reproduction analysis instead of real execution. In the real execution, loads of submerged sensor hubs and a boat on the ocean surface are required. The hubs are furnished with sensors to detect and procure data, a battery to give energy, a memory gadget to store information, a processor to accomplish controlling and handling capabilities, an acoustic modem to accomplish submerged remote acoustic correspondences, a power enhancer, the waterproof gadget, etc. As far as handling, the hubs ought to be fast, stable, and energy-saving. In memory, they need to have an enormous stockpiling limit and guarantee that no information is lost after the passing of hubs. With respect to the submerged remote correspondence innovation, we are attempting to accomplish low dormancy, low blunder rate, and significant distance interchanges.

5. Conclusions

In the future networking field of underwater acoustic sensor networks, network coverage and energy consumption are the key concerns. In this paper, we introduced the IEDA-HGEO algorithm, an improved energy-efficient data collection method for hexagonal grid structures in wireless sensor networks. Here, we use a grid structure with multi-hop routing as a data transmission mechanism and use energy, distance, and end-to-end delay as factors to find the next hop. With reference to energy consumption, throughput, PDR, and delay, our suggested method was thoroughly compared to two other well-known routing methods, the ERP2R and EGRC. Simulation results demonstrated that our technique performed better than previous algorithms as well as successfully extended network lifetime. Proposed method achieved the following metrics: delay 41%, energy consumption 48%, efficiency 95%, throughput 95%, and PDR 92%.