Sage Journals: Discover world-class research

Abstract

In this article, to monitor the fields with square and circular geometries, three energy-efficient routing protocols are proposed for underwater wireless sensor networks. First one is sparsity-aware energy-efficient clustering, second one is circular sparsity-aware energy-efficient clustering, and the third one is circular depth–based sparsity-aware energy-efficient clustering routing protocol. All three protocols are proposed to minimize the energy consumption of sparse regions, whereas sparsity search algorithm is proposed to find sparse regions and density search algorithm is used to find dense regions of the network field. Moreover, clustering is performed in dense regions to minimize redundant transmissions of a data packet, while sink mobility is exploited to collect data from sensor nodes with an objective of minimum energy consumption. A depth threshold $(d_{th})$ value is also used to minimize number of hops between source and destination for less energy consumption. Simulation results show that our schemes perform better than their counter-part schemes (depth-based routing and energy-efficient depth-based routing) in terms of energy efficiency.

Keywords

Energy efficient network stability clustering underwater wireless sensor networks sparse and dense networks maximum coverage minimum distance

Introduction

Underwater wireless sensor network (UWSN) has attracted industrial and research community due to its detrimental nature which provides many unique applications like under-sea exploration, aquatic environment monitoring, underwater pollution monitoring, coastal surveillance for defense strategies, mineral extraction, and so on.^1–3 Typical UWSN architecture consists of sink(s) and sensor nodes to gather useful information. However, communication is a challenging task in dynamic underwater environment due to limited battery capacity of a sensor node, path loss, multi-path, noise, and so on.⁴

Due to dynamic nature of aquatic environment, battery replacement of a node is difficult in case of battery failure. To prolong the network lifetime, an energy-efficient routing protocol is proposed for UWSNs by Umar et al.⁵ It uses mobile sink(s) (MS) to gather data from sensor nodes to increase throughput and minimize packet drop ratio, however, at the cost of more end-to-end delay. Similarly, Wahid and Kim⁶ select the forwarder node with high residual energy and low depth to avoid energy hole creation in the network. It balanced the energy consumption in dense regions of the network, although nodes near the sink(s) die quickly as compared to nodes far from the destination due to imbalanced traffic load. In the work by Khan and Cho,⁷ MS is used to gather data packets from sensor nodes. All sensor nodes form local clusters and choose a cluster head (CH). Instead of visiting each node separately, the MS visits only CHs and gathers data from all the local clusters and achieves considerable energy efficiency.

The proposed schemes^5–7 perform better in dense regions, whereas in later rounds, distance between the sender and receiver nodes increases resulting in more energy consumption. However, these schemes do not prevent extra traffic load on sensor nodes close to the sink. Due to large data load, sensor nodes near the sink consume their energy quickly compared to nodes those are deployed far from the sink. This quick depletion of energy increases the packet drop ratio. Therefore, it is required to devise new techniques for conscious utilization of nodes battery to improve the network lifespan.⁸

This article (an extension of Azam et al.⁹) introduces three energy-efficient routing protocols, sparsity-aware energy-efficient clustering (SEEC),⁹ circular sparsity-aware energy-efficient clustering (CSEEC), and circular depth–based sparsity-aware energy-efficient clustering (CDSEEC), to minimize the data load in the dense regions and to avoid energy hole creation in sparse regions of the network field. A $d_{th}$ value is used to minimize the number of hops between the source and the destination to conserve energy of the node. Two algorithms are also proposed: sparsity search algorithm (SSA) to find dense region and density search algorithm (DSA) to find sparse regions in the network field. In $SEEC$ , two MSs are deployed in the middle of a sparse region to avoid energy hole creation. Moreover, one remains in the middle of sparse region until the death of all sensor nodes in that region. The second MS moves in sparse regions to gather data from sensor nodes, whereas $CSEEC$ is proposed to avoid energy hole creation in sparse regions. Both MSs change their regions after some interval to gather more data from sensor nodes with less energy consumption. Similarly, $CDSEEC$ is presented to minimize energy consumption at the cost of network throughput.

The rest of the article is organized as follows. Existing literature is discussed in section “Related work.” In section “Proposed protocols,” details of proposed schemes are given. Similarly, performance evaluation is presented in section “Simulation discussion,” and, finally, the conclusion of our work is given in section “Conclusion and future work.”

Related work

In this section, we discuss some of the latest energy-efficient routing protocols. The routing protocols are divided into two categories: energy-efficient routing protocols in UWSNs and energy-efficient routing protocols in terrestrial wireless sensor networks (WSNs). The protocols of each category are discussed in upcoming subsections.

Energy-efficient routing protocols in UWSNs

Yan et al.¹⁰ propose a localization-free depth-based routing (DBR) protocol for UWSNs. DBR uses depth metric for data forwarding via multi-hopping. This scheme is suitable for dense networks, while in sparse networks, packet delivery ratio (PDR) gradually decreases due to more distance between the source and the destination. Yu et al.¹¹ propose weighting depth and forwarding area division depth-based routing (WDFAD-DBR) protocol to avoid void nodes in the network field. WDFAD-DBR finds the neighbors up to two hops between the sender and the receiver for reliable data packet delivery. It enhances the network lifetime with the division of forwarding area according to node density and channel condition. It performs unequal clustering for handling different amount of data load in the network field which results in less energy consumption.

Yoon et al.¹² present an autonomous underwater vehicle (AUV)-aided underwater routing protocol (AURP) for UWSNs. In AURP, AUVs and gateway nodes are used to collect maximum data from the network nodes with less energy consumption of the network. Similarly, in the work by Khan and Cho,⁷ a distributed data gathering (DDG) routing protocol is proposed to gather aggregated data packet from CH directly via AUVs. In this way, transmission distance between the sender and the receiver is reduced and the network lifespan is increased.

Clustering schemes^13–17 are proposed for UWSNs, in which CHs are selected on the basis of high residual energy and low depth. Data packets are aggregated at the nominated CH; then, a composite (single compressed data packet) data packet is transmitted to the sink(s) via multi-hopping if it is not in the transmission range. Jiang et al.¹⁸ propose node non-uniform deployment based on clustering (NNDBC) algorithm. NNDBC improves network connectivity by determining the heterogeneous communication range instead of homogeneous communication range of the sensor nodes. The network lifetime is prolonged due to the contribution of lower degree sensor nodes to prevent the early death of higher degree sensor nodes.

The balance transmission mechanism (BTM) is proposed for underwater acoustic sensor networks (UASNs).¹⁹ In BTM, data transmission is divided into two phases: In phase-I, a path is set up for energy optimization, and in phase-II, a data balance transmission algorithm is implemented for stable data transmission. In the work by Akbar et al.,²⁰ three-dimensional sink mobility (3D-SM) scheme has been proposed to improve network lifetime. The 3D network field is divided into four rectangular cuboids (RCs), whereas MS and courier nodes (CNs) are used to collect the data from sensor nodes in the network. In 3D-SM, MS is deployed in one cuboid, and in the remaining three cuboids, CNs are deployed at minimum distance from sensor nodes to reduce the energy consumption of the network. In 3D-SM, energy efficiency is achieved at the cost of end-to-end delay.

Chen and Lin²¹ propose a mobile geocast routing protocol (3D zone of reference) to overcome energy hole creation and to minimize energy consumption of sensor nodes while maximizing the data collection from the network field. In the work by Javaid et al.,²² the authors propose three chain-based routing schemes to find global optimal path for data transmission. First one is 4-chain-based routing scheme in which sensor nodes are divided into four groups to form inter-connected chains for data routing, where all the sensor nodes select a chain head in their respective chain to forward a data at the respective destination. Second contribution is 2-chain-based routing scheme in which sensor nodes are divided into two groups and both groups are inter-connected with each other to find global optimum paths from local optimum paths to route data between the source and the destination. The last one is the single-chain-based routing scheme, in which sensor nodes form a single-chain of sensor nodes in a cylindrical network for the selection of global optimum path between the sender and the receiver. The 4-chain-based routing scheme performs better than the other two schemes because it selects optimum number of neighbors for data transmission in the network field.

Authors propose an ACOA-AFSA (ant colony optimization algorithm–artificial fish swarm algorithm) fusion routing algorithm for UWSNs to reduce the energy consumption and the transmission delay²³ of the network field. It possesses the advantages of both AFSA and ACOA. The use of fusion algorithm reduces the transmission delay in the existing routing protocols and energy consumption that improves the robustness of the routing protocols theoretically. In Table 1, we have given the comparison of discussed existing UWSN routing protocols.

Table 1.

Comparison of existing UWSN routing protocols.

Protocol	Technique	Parameters achieved	Parameters compromised
DBR¹⁰	Depth-based routing	Maximum throughput and low energy consumption	Stability period
EEDBR⁶	Depth- and energy-based routing	Low end-to-end delay and maximum throughput	High energy consumption
WDFAD-DBR¹¹	Depth- and energy-based routing	High energy efficiency	Less throughput
DEADS⁵	Cooperative routing	Data reliability, maximum throughput, and Less packet drop	High energy consumption
BTM¹⁹	Multi-hop routing	Balanced energy consumption	Minimum throughput
NNDBC¹⁸	Clustering	High connectivity and network coverage rate	High energy consumption
DDG⁷	Multi-hop routing	Maximum throughput and efficient energy consumption	Large overhead
AURP¹²	AUVs for distant transmissions	High packet delivery ratio and less energy consumption	High end-to-end delay
3D-SM²⁰	MS and CNs for data transmission	Minimum energy consumption	High end-to-end delay
Mobicast²¹	AUVs	Improved throughput and minimized end-to-end delay	Increased energy consumption and message overhead
Chain based²²	Application-oriented routing protocol for cylindrical network	Increased network lifetime, data sending rate, decreased path loss, transmission loss, and end-to-end delay	Increased routing overhead
ACOA-AFSA²³	Fusion routing algorithm	Reduced transmission delay and energy consumption	High routing overhead

UWSN: underwater wireless sensor network; DBR: depth-based routing; EEDBR: energy-efficient depth-based routing; WDFAD-DBR: weighting depth and forwarding area division depth-based routing; DEADS: depth and energy aware dominating set–based algorithm; BTM: balance transmission mechanism; NNDBC: node non-uniform deployment based on clustering; DDG: distributed data gathering; AURP: AUV-aided underwater routing protocol; 3D-SM: three-dimensional sink mobility; ACOA: ant colony optimization algorithm; AFSA: artificial fish swarm algorithm; AUVs: autonomous underwater vehicles; CNs: courier nodes; MS: mobile sink.

Energy-efficient routing protocols in terrestrial WSNs

Zhou et al. propose an energy-efficient routing protocol for solving combinatorial problems in WSNs. Authors introduce multi-objective free search algorithm (MOFS) to optimize the network coverage and ensure the efficient energy consumption. This scheme has not only considered the individual node energy but also has focused on the entire network node energy for the maximization of network lifespan. However, end-to-end delay is increased due to active and sleep mode switching mechanism. Jing et al.²⁴ propose a boundary detection method (BDM) for large-scale coverage holes in terrestrial WSNs which is based on minimum critical threshold constraint. BDM has low computational complexity when detecting large-scale coverage holes in terrestrial WSNs.

A first clustering-based routing protocol for terrestrial WSNs was low-energy adaptive clustering hierarchy (LEACH).²⁵ In LEACH, local clusters of nodes are formed on the basis of minimum distance. In each cluster, CH collects data from nodes and then forward it to a base station (BS). It achieves energy efficiency due to the random rotation of CHs through a predefined mechanism in order to balance the energy consumption of the nodes. However, the CH selection is not optimal which results in unequal cluster formation which causes imbalance energy consumption of the network. threshold sensitive energy efficient sensor network protocol²⁶ is an improved form of LEACH by introducing clustering scheme in reactive WSNs. Stable election protocol (SEP)²⁷ and distributed energy-efficient clustering (DEEC)²⁸ are clustering routing protocols for heterogeneous WSNs. In SEP and DEEC, two different energy levels are defined and nodes are categorized into normal nodes and advanced nodes on the basis of these energy levels for balanced energy consumption.

Ahmad et al.²⁹ propose an adaptive clustering for terrestrial WSNs, in which nodes associate with CHs to reduce the communication distance and avoid back transmissions, thus resulting in minimum energy consumption and high network lifetime. First, CHs are elected on the basis of predefined threshold value. Then, through natural selection phase, optimal number of CHs is selected on the basis of optimal distance due to which CH achieves balanced load during data receiving and forwarding phase. It achieves maximum throughput and prolonged network lifetime in homogeneous, heterogeneous, and reactive and proactive conditions. In the work by Baranidharan and Santhi,³⁰ a genetic algorithm–based energy-efficient clustering hierarchy (GAECH) technique is proposed for load balancing between sensor nodes. GAECH uses fitness function to form well-balanced clusters in the network to increase the stability period and lifespan of the network.

For collecting information from the network field, algorithms like artificial bee colony (ABC) and ant colony optimization (ACO) techniques are used to achieve better network lifetime. In the work by Kumar and Kumar,³¹ ABC and ACO are combined to design a hybrid algorithm called ABC-ACO; it works in three phases: in first phase, optimal number of sub-regions is selected; in second phase, CHs are selected using ABC algorithm from the selected sub-regions of phase one; in final phase, data are transmitted by CHs using ACO algorithm. ACO algorithm chooses best route to the BS for minimum energy consumption of sensor nodes.

In the work by Wang et al.,³² the authors propose an energy-aware sink relocation (EASR) strategy for MS in terrestrial WSNs. Sensor nodes near the sink consume their energy quickly due to imbalanced data traffic resulting in short lifespan of the network. EASR uses residual energy of a sensor node to adjust its transmission range adaptively. Based on this information, a sink is relocated accordingly for uniform load on all sensor nodes. Thus, EASR achieves maximum network lifetime at the cost of high end-to-end delay. Comparison of all discussed terrestrial WSN routing protocols is given in Table 2.

Table 2.

Comparison of existing terrestrial WSN routing protocols.

Protocol	Technique	Parameters achieved	Parameters compromised
MOFS³³	Node sleep/active mechanism in WSNs	Reduce energy consumption of the whole network nodes	High end-to-end delay and network coverage
LEACH²⁵	Reducing global communication in WSNs	Maximum throughput	Maximum energy consumption and short stability period
ACH2²⁹	Residual energy–based routing	Energy efficiency, maximum throughput, and stability period	Transmission delay
GAECH³⁰	Balanced clustering in terrestrial WSNs	Energy efficiency and network stability period	Maximum delay and minimum throughput
ABC-ACO³¹	Residual energy and distance from BS	Less energy consumption	Maximum end-to-end delay

MOFS: multi-objective free search algorithm; LEACH: low-energy adaptive clustering hierarchy; ACH2: Adaptive clustering habit; GAECH: genetic algorithm–based energy-efficient clustering hierarchy; ABC: artificial bee colony; ACO: ant colony optimization; BS: base station; WSN: wireless sensor network.

Proposed protocols

In our work, we have focused on energy efficiency of sensor nodes in sparse regions of the network field. $SEEC$ , $CSEEC$ , and $CDSEEC$ are proposed to minimize the data load in the dense regions and avoid energy hole creation in sparse regions. In $SEEC$ , two MSs are deployed in the middle of a sparse region to avoid energy hole creation, whereas $CSEEC$ avoids energy hole creation by constant mobility of both MSs in sparse regions. Similarly, $CDSEEC$ achieves high network lifetime at the cost of low network throughput.

SEEC

At network setup, network field is divided into $N_{regions}$ regions, and sensor nodes are randomly deployed in the network field as shown in Figure 1. A static sink is deployed at the top center of the network field and two MSs are deployed in the network field and then a hello message packet is broadcasted by all sinks across the network. A hello message carries information on its source. Each sensor node receives a hello packet and updates its hop count from a sink. In this way, every sensor node acquires its distance in terms of hop count from a sink. Each sensor node broadcasts a hello message in its transmission range with its hop count from the sink. If a sensor node has less hop count, it discards hello message, else it updates its number of hops and rebroadcasts the hello message in its transmission range. This process continues until every sensor node acquires its hop count from a sink.

Figure 1.

The network model of SEEC.

Now, dense and sparse regions are calculated to start the data transmission in the network. In dense regions, a CH is nominated and then a hello message is sent to inform its neighbor nodes about their selection in the cluster. Similarly, after every round, a hello message is broadcast by only MSs, so every CH acquires hop count for multi-hopping. In this way, dynamic changes of the MS location are updated in order to avoid data packet loss and conserve energy of the network nodes.

The logical division of the network field is shown in Figure 2. The network field is divided to identify sparse and dense regions to restrict the mobility of sinks only in sparse regions of the network. Where $R_{L}$ denotes left-side regions, $R_{R}$ shows right-side regions, and $O (x_{0}, y_{0})$ used to denote the origin of the coordinates in the network. Equation (1) shows the logical division of left side of the network field

$R_{Ln} (x_{0}, y_{0}) = (x_{0} + W, y_{0} + nB)$ (1)

Figure 2.

Division of network field into the left and right n regions.

The value of $W$ denotes the $width$ of the network field. $B$ shows breadth of the network field, which is formulated as follows

$B = \frac{breadth}{n}, \forall breath, n \in R$ (2)

The following equation (3) divides the right side of the network field

$R_{Rn} (x_{n}, y_{n}) = (x_{n} + W, y_{n} + nB), \forall W, B, n \in R$ (3)

Searching sparse and dense regions

The number of sensor nodes is searched in each region to find sparse or dense regions of the network field according to algorithms 1 and 2. The number of regions is sorted in ascending order on the basis of number of sensor nodes. The sparse and dense regions are decided based on the density of the nodes in each logical region. Let us suppose, to be selected as a dense region, there must be $N_{th}$ number of nodes to meet the predefined threshold. Equation (4) is used to calculate the sparse and dense regions of the network. Sparse and dense region search mechanism is shown in Figure 3

$N_{th} = {\begin{matrix} 1 & if N \geq N_{th} \\ 0 & if N < N_{th} \end{matrix}$ (4)

Algorithm 1. Density search algorithm (DSA).
1: procedure Density Search2: $N \leftarrow Number of nodes$ 3: $R \leftarrow Number of regions$ 4: for each node $n$ $N$ do5: Find coordinates of node n6: for each region $r \in R$ do7: if $n \in r$ then8: Nodes(r) = n9: end if10: $R (r) \leftarrow r$ 11: end for12: end for13: R = Descending sort(R)14: $dense = {r \| r \in R \land r \leq R / 2}$ 15: for each region $r \in R$ do16: if $r = dense$ then17: dense regions = r18: end if19: end for20: end procedure

Algorithm 1. Density search algorithm (DSA).

1: procedure Density Search2:

N \leftarrow Number of nodes

R \leftarrow Number of regions

4: for each node

n

N

do5: Find coordinates of node n6: for each region

r \in R

do7: if

n \in r

then8: Nodes(r) = n9: end if10:

R (r) \leftarrow r

11: end for12: end for13: R = Descending sort(R)14:

dense = {r | r \in R \land r \leq R / 2}

15: for each region

r \in R

do16: if

r = dense

then17: dense regions = r18: end if19: end for20: end procedure

Algorithm 2. Sparsity search algorithm (SSA).
1: procedure Sparsity Search2: $N \leftarrow Number of nodes$ 3: $R \leftarrow Number of regions$ 4: for each node $n \in N$ do5: Find coordinates of node n6: for each region $r \in R$ do7: if $n \in r$ then8: Nodes(r) = n9: end if10: $R (r) \leftarrow r$ 11: end for12: end for13: R = Descending sort(R)14: $sparse = {r \| r \in R \land r > R / 2}$ 15: for each region $r \in R$ do16: if $r = sparse$ then17: sparse regions= r18: end if19: end for20: end procedure

Algorithm 2. Sparsity search algorithm (SSA).

1: procedure Sparsity Search2:

N \leftarrow Number of nodes

R \leftarrow Number of regions

4: for each node

n \in N

do5: Find coordinates of node n6: for each region

r \in R

do7: if

n \in r

then8: Nodes(r) = n9: end if10:

R (r) \leftarrow r

11: end for12: end for13: R = Descending sort(R)14:

sparse = {r | r \in R \land r > R / 2}

15: for each region

r \in R

do16: if

r = sparse

then17: sparse regions= r18: end if19: end for20: end procedure

Figure 3.

Flow chart of network.

When $N$ is greater or equal to the $N_{th}$ , then the region is named as a dense region; otherwise, it is marked as a sparse region.

Clustering in dense region

We used clustering technique in $N_{regions}$ dense regions to minimize the energy consumption due to duplicate packets transmission. The cluster is formed after the selection of CH on the basis of high residual energy and low depth. The following conditions given in equations (5)–(7) are necessary to become a CH

$C H_{d} > R_{d}$ (5)

where $R_{d}$ shows depth of the destination node to any of the sink from a source node

$E_{CH} > E_{ave}$ (6)

Similarly, $E_{CH}$ is the residual energy of a CH and $E_{ave}$ is the average residual energy of an individual node in the cluster

where

$\begin{matrix} E_{ave} = \frac{Total Residual Energy}{Number of Alive Nodes} \\ rand \leq Th \end{matrix}$ (7)

where

$Th (i) = \frac{p}{1 - p (\mod (r, \frac{1}{p}))}$

First, for the selection of CH, condition given in equation (5) is checked. If depth of a node is higher than the other neighboring nodes, then the node cannot be a CH in the current round. However, when depth of a node is lower than the depth of other nodes of the cluster, then equation (6) is verified. The node is selected as a CH if residual energy of the node is greater than the average residual energy of an individual node in the cluster. Equation (7) finds the probability p of a node to be nominated as a CH and it is calculated at the start of each round. Where $r$ denotes the current round. The node is selected as a CH if it has not been CH for the last 1/p rounds. The node generates a random number (rand) and then compares the rand number with threshold value $Th (i)$ .²⁵ The node is considered a normal node if rand value of a node is greater than the $Th$ . In case rand value of the node is less than $Th$ value, then the node is finally nominated as CH. In SEEC, only one node is selected as a CH in each dense region per round.

We illustrate an example of CH selection as shown in Figure 4(a). We have two nodes N1 and N2 with different energy levels and at the same depth. The node with high residual energy is selected as a CH, that is, N2, while in Figure 4(b), N1 is selected as a CH because both nodes have same energy level; however, they are at different depths from the destination. Thus, low depth node is selected as a CH. After the selection of CH, it broadcasts hello message packet in its transmission range to inform sensor nodes about their selection in the cluster. After the selection, all sensor nodes in a cluster send their data to the nominated CH, where all data packets are compressed and transmitted to a nearest sink directly if in transmission range, else transmitted via multi-hopping.

Figure 4.

Scenario of the CH selection in SEEC.

SM in sparse region

After the identification of sparse regions using SSA algorithm, two MSs are deployed to gather data from the sensor nodes. The mobility pattern of both sinks varies to avoid void hole creation in the sparse regions. Let $M S_{1}$ is deployed in sparse regions and changes its position in each round from most sparse to the least sparse region. In every round, $M S_{1}$ positioned itself at the center of a sparse region to collect more number of data packets from the sensor nodes in that region, while $M S_{2}$ gathers data packets from a least sparse region until the death of all sensor nodes in that region. Both MSs positioned themselves at the center of a sparse region to collect maximum data from all sensor nodes and avoid energy hole creation. Instead, all the sensor nodes send data to a sink, sensor nodes send data packets to their CH. From there, data packets are aggregated and forwarded to the near by sink in its transmission range by CH, else it transmits data packet to a sink via multi-hopping through neighboring CHs. The center of the left and right sparse regions ( $L S_{region}$ , $R S_{region}$ ) is computed to deploy MS at the middle of the region as shown in Figure 1

$L S_{region} (x_{0}, y_{0}) = (x_{0} + \frac{W}{2}, y_{0} + \frac{nB}{2})$ (8)

$R S_{region} (x_{n}, y_{n}) = (x_{n} + \frac{W}{2}, y_{n} + \frac{nB}{2})$ (9)

where $n$ is the total number of left-side or right-side sparse regions.

CSEEC

The network model of CSEEC is shown in Figure 5. We have divided the network field into $N_{circles}$ concentric circles. Then, each concentric circle is divided into equal parts. Network setup is same as discussed in section “SEEC.” A static sink is deployed at the top center of the circular network field, whereas both $M S_{1}$ and $M S_{2}$ move in sparse regions constantly to gather data from sensor nodes of the network. Rest of the process is same as mentioned in section “SEEC.”

Figure 5.

Network model of CSEEC.

The network model is shown in Figure 5, where $C_{circles}$ denotes number of concentric circles of the network field, $R$ represents radius of the network field, and $r$ is the radius of the inner concentric circle. The following equation shows the number of concentric circles

$C_{circles} = \frac{R}{r}$ (10)

where coordinates of MSs are calculated as follows

$X = r_{i} \cos θ_{i}$ (11)

$Y = r_{i} \sin θ_{i}$ (12)

The process of finding sparse and dense regions and cluster formation is same as discussed in SEEC.

Data communication in dense regions

Once the dense regions are specified according to equation (4), then sensor nodes transmit their sensed data to the respective CH. Each CH verifies before transmitting its composite data packet to the neighbor CH that none of the sink is in its transmission range. If CH is not within the transmission of any sink, then data packets are forwarded via multi-hopping through neighbor CHs; otherwise, directly transmit to the sink within its vicinity. At CH it is checked; if any of the MSs is in communication range, then data are passed to it, else multi-hopping continues till a composite packet reached its destination.

Mobility pattern in sparse regions

After the identification of sparse regions in the network field, $M S_{1}$ is deployed at a most sparse region and $M S_{2}$ at a least sparse region as shown in Figure 5. Both MSs change their regions after each round from most sparse to least sparse region and vice versa. Both MSs positioned themselves at the center of a sparse region in order to gather more number of data packets from the sensor nodes in that region. Both sink(s) always visit different regions in every round to cover more regions and to avoid energy hole creation resulting in more PDR. Sensor nodes send data packets directly if any of the sink is in their transmission range, else data packets are transmitted in multi-hop fashion.

CDSEEC

Let the network field be divided into two semi-circles as shown in Figure 6. The lower semi-circle consists of $N$ concentric half circles. Each half circle is divided into $N_{sectors}$ sectors. We divided network field into small regions and then further into sectors because it is easy to identify sparse and dense regions based on the number of nodes. Two MSs are deployed in sparse regions of lower half-coronas as shown in Figure 7. Both MSs move from a least sparse to a most sparse region and vice versa. Sensor nodes are randomly deployed in both upper and lower half-coronas. In the upper half circle, depth threshold value is used to minimize number of hops which decreases the data load on sensor nodes deployed close to the static sink as shown in Figure 8. The logical division of the network field is shown in Figure 6. The process of finding sparse and dense regions, cluster formation, and SM is same as discussed in SEEC.

Figure 6.

Network model of CDSEEC.

Figure 7.

Network operation of lower semi-circle.

Figure 8.

Network operation of upper semi-circle.

Data transmission in upper semi-circle

We used DBR technique¹⁰ in the upper semi-circle of the network field as shown in Figure 8. DBR is based on greedy algorithm, in which packets are delivered from a source node to the sink using depth information of the neighboring nodes having minimum distance to the destination. A sensor node collects data packet and transmits to the BS via multi-hopping. The sensor nodes send their depth information along with the data packet to other nodes. The decision of forwarding data packet depends on the depth of both current node and previous node (PN). On receiving a packet, the current node compares its depth with the depth of the PN, which is stored in the header of the received packet. If current node depth is smaller than the depth of PN, then the current node is selected as a next forwarding node (FN) for transmitting data packet. Otherwise, the current node discards the data packet immediately. It is possible that multiple neighbor nodes of an FN have same depth for forwarding data packet to the next hop. If all neighboring nodes broadcast their data packet, high collision and high energy consumption will occur. In order to reduce collision and energy consumption, a depth threshold $d_{th}$ is used to reduce the redundant transmissions of the data packet. The packet is forwarded only if the difference between the depth of PN and the depth of current node is larger than the depth threshold $d_{th}$ . In CDSEEC, we used a large positive threshold value to achieve energy efficiency by restricting the neighboring nodes from data forwarding. The restriction of $d_{th}$ prevents the large number of sensor nodes to take part in data transmission which results in less energy consumption of the network nodes.

Data transmission in lower semi-circle

Clustering technique is used in the lower semi-circle of the network, where $N$ number of nodes is randomly deployed. Lower semi-circle is shown in Figure 7, which is divided into $n$ sectors, those are further divided into sub-sectors. The division is performed in order to avoid collision due to high density of nodes; packet drop ratio increases resulting in wastage of nodes’ energy and degrades the network performance. Let us suppose $M S_{1}$ gathers composite data packet from high node density cluster (high node density region is calculated by DSA algorithm), whereas $M S_{2}$ gathers data from low node density cluster (low node density region is computed according to SSA). In each round, CHs are nominated and nodes those in transmission range are associated with the respective CH. Both MSs always gather data from different regions even when two clusters have same node density; in that case, it depends upon the total energy of the cluster. Suppose that the energy of both clusters is same, then randomly both MSs are deployed, however, in different regions. The data packets are transmitted to MSs directly if in range, else multi-hop approach is adopted to deliver composite data packet to any of the MS; lower semi-circle CHs never transmit their data packet to the static sink. In this way, it preserves the energy of the nodes because communication over long distances is avoided and also number of hops is minimized ultimately leading to efficient energy consumption and higher network lifetime.

Simulation discussion

For performance evaluation, we compare our routing protocol with two depth-based localization-free routing protocols, DBR and energy-efficient depth-based routing (EEDBR), of UWSNs. In DBR, forwarder sensor nodes are selected on the basis of depth, while in EEDBR, the selection of a forwarder node is based on the high residual energy and low depth from the sink. Therefore, low depth and high residual energy sensor nodes consume their energy at much higher rate than those sensor nodes deployed away from the sink due to imbalanced traffic load. The imbalanced data load results in energy hole creation that maximizes the energy consumption and decrease the network lifetime. Due to these consequences, we have proposed SEEC routing protocol for UWSNs to counter these challenges which are evaluated via simulations.

SEEC

In order to get fair results, we have run simulations for 2000 rounds with same parameters. All the sensor nodes are randomly deployed in the network field of $100 m \times 100 m$ , the number of nodes is 100, the initial residual energy of each sensor node is 5 J with the transmission range of 30 m, and the predefined depth threshold of each sensor node is 25 m. Results of our protocol are compared with DBR and EEDBR on the basis of network lifetime, network residual energy, packets received at sink, stability period, and instability period of the network.

The network lifetime of our proposed routing protocol is better than the DBR and EEDBR routing protocols because in DBR low depth and in EEDBR higher residual energy and low depth nodes are selected for data forwarding. Extra load on sensor nodes near the sink consumes their complete energy in data forwarding which results in energy hole creation which reduces the network lifetime as shown in Figure 9, while in SEEC, network lifetime is enhanced due to mobility in sparse regions and clustering in dense regions where CH is selected to gather data from its neighboring nodes. Instead of gathering data from each node, MSs directly receive aggregated data packet from the CH resulting in less energy utilization of the network nodes. It is clearly shown in Figure 9 that our scheme outperforms the counter-part techniques in terms of network lifetime.

Figure 9.

Network lifetime.

In EEDBR, nodes with high residual energy and low depth are selected for forwarding data to the sink, which causes more energy consumption of the network nodes, while in DBR, low depth nodes are selected as data forwarders. However, DBR performs better in dense regions, while in sparse regions, long propagational distances result in more energy consumption as shown in Figure 10. Therefore, two MSs are deployed in sparse regions to minimize the distance between the source and the destination. Moreover, in dense regions, clustering is performed which causes less energy utilization of all sensor nodes. Thus, average network residual energy is consumed less than the DBR and EEDBR.

Figure 10.

Total energy consumption of the network.

In EEDBR, packet delivery in each round is higher than DBR and SEEC due to the selection of high residual energy and low depth sensor nodes to forward data at the cost of high energy consumption as shown in Figure 11, while in our proposed routing protocol, the ratio of packets sent in each round is less because CHs transmit compressed data packet of all neighbors to the sink(s) instead of multiple packets. However, in DBR and EEDBR, the total number of sent packets decreases as network sparsity increases, whereas in our proposed scheme, packet received ratio is more in later rounds due to sinks mobility in sparse regions as shown in Figure 12.

Figure 11.

Network throughput.

Figure 12.

Number of packets received in each round.

CSEEC and CDSEEC

CSEEC and CDSEEC are enhanced forms of SEEC. In these schemes, we have changed the mobility pattern because in SEEC, one MS moves in sparse regions, while the other one remains in a sparse region until the death of all nodes in that region. This leads to more energy consumption due to less network volume coverage. However, we have kept the periodic mobility of both sink(s) to ensure the maximum coverage of the network field.

Simulation parameters are shown in Table 3. The radius of the network field is kept 57 m, number of nodes is 100, and each node initially has 5 J of energy with transmission range of 50 m. The size of the data packet is 50 bytes and simulations are run for 3500 rounds.

Table 3.

Simulation parameters.

Parameter	Value
Network field radius	57 m
Number of nodes	100
Initial energy of nodes	5 J
Data packet size	50 bytes
Transmission range	50 m
Number of rounds	3500

The packets received at the sink per round are shown in Figure 13. Initially, SEEC has better results as compared to its counter-part schemes. However, in the sparse network the sink mobility is inefficient due to which packet drop ratio is high. Due to infeasible sink movement a lot of energy is wasted that degrades the network performance. Due to consistent mobility of both MSs in sparse regions of the network field, CSEEC covers more regions which avoids energy hole creation in the network resulting in more packets at sinks in each round. Similarly, in CDSEEC, MS only gathers data from sparse regions of the lower semi-circle. Moreover, clustering is performed in dense regions of the lower semi-circle. Due to multi-hopping in upper semi-circle with depth threshold value, number of hops is minimized between the source and the destination which results in less energy consumption and less packets in each round as shown in Figure 13.

Figure 13.

Packets received at sink per round.

Figure 14 shows that CDSEEC outperforms its compared schemes SEEC and CSEEC in terms of energy efficiency. CDSEEC makes clusters in only lower semi-circle and uses MSs in low node density clusters and high node density clusters of lower semi-circle which balanced the energy consumption of the network. Due to multi-hopping in upper semi-circle with predefined depth threshold value, energy utilization is also minimized, while in CSEEC, clustering is performed in dense regions of the complete network field and MSs gather data packets of all sparse regions from all over the network field which increases energy consumption as compared to CDSEEC. However, it performs better than SEEC due to its periodic movement of both MSs in sparse regions of the network field as shown in Figure 15.

Figure 14.

Network lifetime.

Figure 15.

Network residual energy.

The PDR of SEEC and CSEEC in each round is comparatively better than CDSEEC as shown in Figure 16. The sensor nodes are restricted from forwarding data packets with predefined depth threshold and a composite data packet from CHs results in less PDR. In later rounds, PDR decreases in SEEC and CSEEC due to more energy consumption in earlier rounds, while CDSEEC utilizes energy efficiently which results in better network lifetime in later rounds of the network as shown in Figure 14.

Figure 16.

Packet delivery ratio (PDR).

Performance trade-offs

Trade-offs are given in Table 4 of our proposed schemes, SEEC, CSEEC, and CDSEEC, and counter-part schemes, DBR and EEDBR. SEEC achieves stability period and better energy consumption at the cost of low throughput, while DBR and EEDBR achieved higher throughput by compromising network stability period due to the imbalanced load on the nodes near the sink, whereas CSEEC and CDSEEC achieve better network lifetime and stability period at the cost of network throughput.

Table 4.

Performance trade-offs.

Protocols	Technique	Parameters achieved	Trade-offs
DBR	Depth-based routing	Low energy consumption and maximum throughput	Maximum end-to-end delay
EEDBR	Depth- and energy-based routing	High energy efficiency and maximum throughput	Stability period
SEEC	Cluster-based routing	Low energy consumption and high stability period	Low throughput
CSEEC	Cluster-based routing	High stability period	Low throughput
CDSEEC	Cluster-based routing	High network lifetime and stability period	Low throughput

DBR: depth-based routing; EEDBR: energy-efficient depth-based routing; SEEC: sparsity-aware energy-efficient clustering; CSEEC: circular sparsity-aware energy-efficient clustering; CDSEEC: circular depth–based sparsity-aware energy-efficient clustering.

Conclusion and future work

In this article, we proposed three routing protocols for UWSNs to improve the network lifetime and to avoid energy hole creation. Sparse regions are identified to deploy MSs and data are gathered more efficiently from sensor nodes. A depth threshold was defined to minimize the number of hops between the source and destination which resulted in less energy consumption. The deployment of MSs in sparse regions avoided the creation of energy holes and clustering minimized the data load on the sensor nodes deployed near the sink. The simulation results show that SEEC achieved high network lifetime and comparatively better stability period than existing compared schemes (DBR and EEDBR). However, CSEEC performed better than SEEC in terms of network lifetime and PDR at the cost of network throughput. Similarly, CDSEEC achieved higher network lifespan than SEEC and CSEEC at the cost of low network throughput.

In future, we plan to work on sparsity control of a network to get better network lifetime with uniform load distribution on all sensor nodes near the sink. Also, interference will be considered in dense regions to minimize collision and increase the reliability of data packet delivery in a network.

Footnotes

Academic Editor: Miltiadis Lytras

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research,authorship,and/or publication of this article: This work was supported by a grant from the “Research Center of the Female Scientific and Medical Colleges,” Deanship of Scientific Research,King Saud University.

References

Akyildiz

Pompili

Melodia

. Underwater acoustic sensor networks: research challenges. Ad Hoc Netw 2005; 3(3): 257–279.

Akyildiz

Pompili

Melodia

. State-of-the-art in protocol research for underwater acoustic sensor networks. In: Proceedings of the 1st ACM international workshop on underwater networks, Los Angeles, CA, 25 September 2006, pp.7–16. New York: ACM.

Partan

Kurose

Levine

. A survey of practical issues in underwater networks. ACM SIGMOBILE Mob Comput Commun Rev 2007; 11(4): 23–33.

Ayaz

Abdullah

. Underwater wireless sensor networks: routing issues and future challenges. In: Proceedings of the 7th international conference on advances in mobile computing and multimedia, Kuala Lumpur, Malaysia, 14–16 December 2009, pp.370–375. New York: ACM.

Umar

Javaid

Ahmad

. DEADS: depth and energy aware dominating set based algorithm for cooperative routing along with sink mobility in underwater WSNs. Sensors 2015; 15(6): 14458–14486.

Wahid

Kim

. An energy efficient localization-free routing protocol for underwater wireless sensor networks. Int J Distrib Sens N 2012; 12: 1–12.

Khan

Cho

. A distributed data-gathering protocol using AUV in underwater sensor networks. Sensors 2015; 15(8): 19331–19350.

Cui

J-H

Kong

Gerla

. Challenges: building scalable and distributed Underwater Wireless Sensor Networks (UWSNs) for aquatic applications. Channels 2005; 1–17.

Azam

Majid

Ahmad

. SEEC: sparsity-aware energy efficient clustering protocol for underwater wireless sensor networks. In: Proceedings of the 30th international conference on advanced information networking and applications, Crans-Montana, 23–25 March 2016, pp.352–361. New York: IEEE.

10.

Yan

Shi

Cui

. DBR: depth-based routing for underwater sensor networks. In: International conference on research in networking, Singapore, 5–9 May 2008, pp.72–86. Berlin: Springer.

11.

Yao

Wang

. WDFAD-DBR: weighting depth and forwarding area division DBR routing protocol for UASNs. Ad Hoc Netw 2016; 37: 256–282.

12.

Yoon

Azad

. AURP: an AUV-aided underwater routing protocol for underwater acoustic sensor networks. Sensors 2012; 2: 1827–1845.

13.

Domingo

. A distributed energy-aware routing protocol for underwater wireless sensor networks. Wireless Pers Commun 2011; 57(4): 607–627.

14.

Wang

Zheng

. Distributed minimum-cost clustering protocol for underwater sensor networks (UWSNs). In: Proceedings of the IEEE international conference on communications, Glasgow, 24–28 June 2007, pp.3510–3515. New York: IEEE.

15.

Anupama

Sasidharan

Vadlamani

. A location-based clustering algorithm for data gathering in 3D underwater wireless sensor networks. In: International symposium on telecommunications (IST), Tehran, Iran, 27–28 August 2008, pp.343–348. New York: IEEE.

16.

Wang

Liu

. LUM-HEED: a location unaware, multi-hop routing protocol for underwater acoustic sensor networks. In: Proceedings of the 2011 international conference on computer science and network technology (ICCSNT), Harbin, China, 24–26 December 2011, pp.2336–2340. New York: IEEE.

17.

Liu

Wei

. A new multi-path routing protocol based on cluster for underwater acoustic sensor networks. In: Proceedings of the 2011 international conference on multimedia technology (ICMT), Hangzhou, China, 26–28 July 2011, pp.91–94. New York: IEEE.

18.

Jiang

Liu

. Node non-uniform deployment based on clustering algorithm for underwater sensor networks. Sensors 2015; 15(12): 29997–30010.

19.

Cao

Dou

Dong

. Balance transmission mechanism in underwater acoustic sensor networks. Int J Distrib Sens N 2015; 11: 2.

20.

Akbar

Javaid

Hussain Khan

. Efficient data gathering in 3D linear underwater wireless sensor networks using sink mobility. Sensors 2016; 16(3): 404.

21.

Chen

Lin

. Mobicast routing protocol for underwater sensor networks. IEEE Sens J 2013; 13(2): 737–749.

22.

Javaid

Jafri

Khan

. Chain-based communication in cylindrical underwater wireless sensor networks. Sensors 2015; 15(2): 3625–3649.

23.

Chen

Shi

. An ACOA-AFSA fusion routing algorithm for underwater wireless sensor network. Int J Distrib Sens N 2012; 2012: 1–9.

24.

Jing

Kong

. Boundary detection method for large-scale coverage holes in wireless sensor network based on minimum critical threshold constraint. J Sensors 2014; 2014: 1–13.

25.

Heinzelman

Chandrakasan

Balakrishnan

. Energy-efficient communication protocol for wireless microsensor networks. In: Proceedings of the 33rd annual Hawaii international conference on system sciences, New York, 4–7 January 2000, pp.1–10. New York: IEEE.

26.

Manjeshwar

Agrawal

. TEEN: A routing protocol for enhanced efficiency in wireless sensor networks. In: Proceedings of the 15th international parallel and distributed processing symposium (IPDPS), San Francisco, CA, 23–27 April 2000, p.189. New York: IEEE.

27.

Smaragdakis

Matta

Bestavros

. SEP: a stable election protocol for clustered heterogeneous wireless sensor networks. Technical Report BUCS-TR-2004–022, pp.1–12. Boston, MA: Department Computer Science, Boston University.

28.

Qing

Zhu

Wang

. Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks. Comput Commun 2006; 29(12): 2230–2237.

29.

Ahmad

Javaid

Khan

. Routing scheme to maximize lifetime and throughput of wireless sensor networks. IEEE Sens J 2014; 14(10): 3516–3532.

30.

Baranidharan

Santhi

. GAECH: genetic algorithm based energy efficient clustering hierarchy in wireless sensor networks. J Sensors 2015; 2015: 1–8.

31.

Kumar

. Hybrid swarm intelligence energy efficient clustered routing algorithm for wireless sensor networks. J Sensors 2015; 10(5): 1155–1174.

32.

Wang

Shih

Pan

. A network lifetime enhancement method for sink relocation and its analysis in wireless sensor networks. IEEE Sens J 2014; 14(6): 1932–1943.

33.

Zhou

Liang

. Multiobjective coverage control strategy for energy-efficient wireless sensor networks. Int J Distrib Sens N 2012; 8: 720734.

Monitoring square and circular fields with sensors using energy-efficient cluster-based routing for underwater wireless sensor networks

Abstract

Keywords

Introduction

Related work

Energy-efficient routing protocols in UWSNs

Energy-efficient routing protocols in terrestrial WSNs

Proposed protocols

SEEC

Searching sparse and dense regions

Clustering in dense region

SM in sparse region

CSEEC

Data communication in dense regions

Mobility pattern in sparse regions

CDSEEC

Data transmission in upper semi-circle

Data transmission in lower semi-circle

Simulation discussion

SEEC

CSEEC and CDSEEC

Performance trade-offs

Conclusion and future work

Footnotes

Declaration of conflicting interests

Funding

References