ArticlePDF Available

A Novel Geo-opportunistic Routing Algorithm for Adaptive Transmission in Underwater Internet of Things

Authors:
Article

A Novel Geo-opportunistic Routing Algorithm for Adaptive Transmission in Underwater Internet of Things

Abstract and Figures

Localization of sensors in Underwater Internet of Things (UIoTs) is difficult due to the mobility. This changing makes the routing decisions difficult, which results in unreliable communication. This paper proposes Adaptive Transmission based Geographic and Opportunistic Routing (ATGOR) protocol for reliable communication between nodes. ATGOR operates in two parts: election of a small cube to avoid redundant transmissions and selection of reliable nodes which forward data from the selected small cube for optimal transmissions. Furthermore, to guarantee the reliability of the data packets in a harsh acoustic environment, we propose Mobility Aware ATGOR (MA-ATGOR), which predicts the locations of neighboring sensor nodes for successful data delivery. In addition, prediction of tthe locations of the sensor nodes helps in avoiding the void holes along with high packet delivery. The performance of the proposed routing protocols is validated based on the PDR, number of void nodes and energy consumption per packet, through simulations.
Content may be subject to copyright.
266 Int. J. Web and Grid Services, Vol. 18, No. 3, 2022
A novel geo-opportunistic routing algorithm for
adaptive transmission in underwater internet of
things
Turki Ali Alghamdi
Department of Computer Science,
College of Computer and Information Systems,
Umm Al-Qura University,
Makkah, 21955, Saudi Arabia.
Email: taghamdi@uqu.edu.sa
Zahoor Ali Khan
Computer Information Science,
Higher Colleges of Technology,
Fujairah, 4114, UAE
Email: zahoor.Khan@hct.ac.ae
Nadeem Javaid*
School of Computer Science,
University of Technology Sydney,
Ultimo, NSW,
2007, Australia
and
Department of Computer Science,
COMSATS University Islamabad,
Islamabad, 44000, Pakistan
Email: nadeemjavaidqau@gmail.com
*Corresponding author
Abstract: Localisation of sensors in underwater internet of things (UIoTs)
is difficult due to the mobility. This changing makes the routing decisions
difficult, which results in unreliable communication. This paper proposes
adaptive transmission-based geographic and opportunistic routing (ATGOR)
protocol for reliable communication between nodes. ATGOR operates in
two parts: election of a small cube to avoid redundant transmissions and
selection of reliable nodes which forward data from the selected small cube
for optimal transmissions. Furthermore, to guarantee the reliability of the
data packets in a harsh acoustic environment, we propose mobility aware
ATGOR (MA-ATGOR), which predicts the locations of neighbouring sensor
nodes for successful data delivery. In addition, prediction of tthe locations
of the sensor nodes helps in avoiding the void holes along with high packet
delivery. The performance of the proposed routing protocols is validated
based on the PDR, number of void nodes and energy consumption per
packet, through simulations.
Copyright © 2022 Inderscience Enterprises Ltd.
A novel geo-opportunistic routing algorithm for adaptive transmission 267
Keywords: underwater internet of things; UIoTs; adaptive transmission; void
hole; geographic and opportunistic routing; mobility prediction.
Reference to this paper should be made as follows: Alghamdi, T.A.,
Khan, Z.A. and Javaid, N. (2022) ‘A novel geo-opportunistic routing
algorithm for adaptive transmission in underwater internet of things’,
Int. J. Web and Grid Services, Vol. 18, No. 3, pp.266–296.
Biographical notes: Turki Ali Alghamdi received his PhD from the
University of Bradford, UK, in 2010. He is a Professor in Computer Science
Department, Faculty of Computer and Information Systems, University of
Umm Al-Qura in Makkah (UQU) and the Founding Director of UQU Smart
Campus Center (SCC). He has previously been the Vice Dean of Technical
Affairs for IT Deanship in Umm Al-Qura University and the Dean of
E-Learning and IT in Taif University. He has more than 15 years of research
and development experience in IT. His research, focusing on wireless sensor
networks, energy and QoS aware routing protocols, network security and
smart cities.
Zahoor Ali Khan is currently the Division Chair of the Computer Information
Science (CIS) Division and the Applied Media Division, Higher Colleges of
Technology, UAE. He has more than 19 years of research and development.
His current research interests include e-health pervasive wireless applications,
theoretical and practical applications of WSNs, smart grids, and IoT. He is
an editorial board member of several prestigious journals. He also serves
as regular reviewer/organiser of numerous reputed journals, conferences, and
workshops. He is a senior member of IAENG. Several conference articles
have received the Best Paper Awards (BWCCA 2012, IEEE ITT 2017, and
EIDWT-2019).
Nadeem Javaid received his PhD from the University of Paris-Est, France
in 2010. He is currently an Associate Professor and the Director of
Communications Over Sensors (ComSens) Research Laboratory, Department
of Computer Science, COMSATS University Islamabad. He is also working
as a Visiting Professor at the School of Computer Science, University
of Technology, Sydney, Australia. He has supervised 137 master and 24
PhD theses. He has authored over 900 articles in technical journals and
international conferences. He was recipient of the Best University Teacher
Award from the Higher Education Commission of Pakistan in 2016. He is an
Associate Editor of IEEE Access and Editor of Sustainable Cities and Society
journals.
This paper is a revised and expanded version of a paper entitled ‘Adaptive
transmission based geographic and opportunistic routing in UWSNs’
presented at Recent Trends and Advances in Wireless and IoT-Enabled
Networks, Islamabad, Pakistan, 18 October 2017.
268 T.A. Alghamdi et al.
1 Introduction
The need to explore and monitor underwater activities has become researchers’ prime
interest because a big area (71%) of Earth’s surface is comprises of water. Due
to limited storage capacity of sensor nodes, the data of these nodes are stored on
the cloud (Oliveira et al., 2019) for further preprocessing. In underwater internet
of things (UIoTs), all sensor nodes collaborate within their communication range
in an environment provided by grid (You et al., 2020) to ensure good network
performance. These sensor nodes perform collaboration in a network through distributed
computing (Chitharanjan and Senthil Kumar, 2021; Ren et al., 2020). This collaborative
behaviour among sensor nodes has been useful for many marine applications such as
mine reconnaissance, pollution monitoring, minerals extraction, etc. However, several
challenges in underwater wireless senssor networks (UWSNs) make the routing decision
difficult in underwater. One of the major challenges is the localisation of nodes because
their location changes due to their movement with water current. This mobility of
sensor nodes increases the occurrence of a void hole in the neighbourhood of sensor
nodes, which consequently declines the network performance. Therefore, predicting the
exact location of nodes requires a good routing mechanism. Moreover, efficient energy
utilisation is considered to be one of the essential factors in measuring the performance
of an algorithm designed for UWSNs. Therefore, a good arrangement of the sensor
nodes is necessary to monitor, sense, and gather the information from the respective
network.
In this manuscript, we have used the following terms alternatively:
1 UIoTs and UWSNs
2 sensors, sensor nodes, nodes, IoT nodes, and IoTs.
With the advent of advancement in sensing technology, monitoring of reachable and
non-reachable areas in underwater is possible. However, the acoustic environment
has unique features, e.g., limited bandwidth, high propagation delay, high absorption,
attenuation of acoustic signal, dynamicity, etc. The earlier mentioned characteristics
cause imbalanced energy dissipation, which results in low network lifetime.
Many studies are conducted on routing to enhance the lifetime of network. A
routing scheme is proposed for increasing the lifetime of network by balancing the
consumption of energy between network nodes. Yu et al. (2016) divide the network
field via reuleaux triangle to ensure that the duplicate packets are discarded to ensure
the effective neighbour node selection for efficient energy dissipation.
However, the existing neighbour nodes prediction mechanisms are not well grounded
and the transmission ranges are fixed because of which energy wastage is inevitable.
The forwarder node is selected by considering depth, which leads to cyclic selection of
the node. This selection causes the node to deplete its battery very quickly, which results
in creation of void holes. It discontinues the data communication among the network
nodes. Additionally, if the potential neighbour is not found closer to the transmission
threshold, the data signal will still be transmitted with exact allocated power without
knowing that the power of the nodes is wasted. Furthermore, the above discussed routing
strategies use reactive approach because of which delay increases while recovering from
void hole.
A novel geo-opportunistic routing algorithm for adaptive transmission 269
Gul et al. (2019) addressed the above limitations and proposed adaptive
transmission-based geographic and opportunistic routing (ATGOR). Geographic routing
is a paradigm in which location or depth information of the neighbouring nodes
is required to greedily select the next-hop forwarders. It eliminates the need
for establishment and maintenance of complete paths. These characteristics make
geographic routing suitable for energy constrained UWSNs having low data-rate. On
the other hand, opportunistic routing is another paradigm, which provides a subset of
neighbouring nodes instead of a single node as a forwarder. By employing opportunistic
routing with geographic routing, the neighbouring set selection is useful for packet
forwarding. The key challenge is the selection of neighbouring nodes that receives the
packet. In this algorithm, the transmission range is dynamically adjusted based on the
distance from the forwarder node. It avoids the unnecessary dissipation of the node’s
battery, which significantly contributes in the prolongation of the network lifespan.
Moreover, two parameters are used for avoiding selection of forwarder nodes in cyclic
way, which are depth and energy. The reason is that if the depth remains constant,
the energy will change after every transmission of the data packet. In addition to our
conference version, a mobility aware ATGOR (MA-ATGOR) is presented to predict
the locations of mobile sensor nodes in UWSN. The mobility prediction mechanism
minimises the energy dissipation by avoiding the void holes along with the decreased
packet loss and minimum re-transmissions for the lost packets.
The contribution of this paper can be organised in following steps:
our proposed routing strategies perform the adaptive transmission level adjustment
and mobility prediction mechanism for void hole avoidance and efficient energy
utilisation
the network is divided logically into small cubes that provides us the better
selection of next-hop forwarders
the opportunistic routing paradigm is utilised for the selection of forwarders,
consisting of eligible neighbouring nodes.
This work is organised as follows: Section 2 introduces some work related to adaptive
and geospatial routing techniques. Section 3 highlights network model and problem
statement. In Section 4, we have designed our proposed schemes and Section 5 provides
a linear optimisation model. Section 6 describes simulation results. In the last, the paper
ends with conclusion and references.
2 Related work
In this section, some related works on geographic and opportunistic routing protocols
are reviewed. Noh et al. propose a technique in which the data sensed by the underwater
sensors is transmitted to the sonobuoys using geographic and opportunistic routing.
Moreover, the forwarding process towards sonobuoys is continued by searching the
forwarders, this search and selection of forwarder nodes is based upon pressure level. In
this technique, each node is aware of the void incurred from the source to destination
through periodic beaconing. The next-hop forwarder sets are maintained by estimating
the pressure levels of neighbouring nodes with a match of vertical direction towards the
surface sonobuoys (Noh et al., 2012). Noh et al. present a routing algorithm in Noh
270 T.A. Alghamdi et al.
et al. (2015). In this algorithm, the pressure level of sensors is used for the selection
of forwarder nodes. Moreover, void recovery is also performed in this algorithm. A
lower-depth-first recovery method is utilised to resume the greedy forwarding at local
maxima. The local maxima node finds another node at lower pressure than itself to find
a recovery path and resumes the greedy forwarding.
In Nicopolitidis et al. (2010), the process of broadcasting is scheduled in such
a way that clients response within a short time interval. The proposed technique
schedules the broadcast process and also reduces the high latency. Yu et al. (2015)
propose a forwarding technique, which is unlike to vector-based forwarding (VBF),
the transmission radius of virtual pipeline and transmission power levels are adjusted
at each hop in AHH-VBF. It improves network reliability by adaptively adjusting
the transmission radius in case of locally sparse and dense node distributions to
avoid the void holes. The distance between source and reliable forwarder node is
considered to adjust the transmission power, which has a positive impact on network’s
energy consumption. Besides this, for the selection of forwarder, its distance from the
destination node is considered, which results in reduced end2end delay.
In Al-Salti et al. (2014), a multipath grid-based geographic routing (MGGR)
strategy uses three-dimensional grid-based environment with mobile sensor nodes and
utilises some localisation service to find the locations of the nodes. The sensor nodes
communicate in a grid-by-grid manner by exploiting available disjoint paths from
sources to destinations. In MGGR, when there is not a node in any of the cell then
void avoidance is performed. When MGGR encounters a void region, the negative
acknowledgment is sent to the sender node informing the void path. The source node
finds an alternative path from the set of alternative disjoint paths. The grid-by-grid
communication and void bypassing provide high delivery ratio with reduced end2end
delay. Coutinho et al. (2015) propose a depth adjustment-based technique to forward
the sensed data to the surface sinks. Furthermore, the depth adjustment is performed for
void recovery.
Han et al. (2019) design a prediction-based data collection model that is built to
overcome data unbalancing in the underwater environment. In the proposed mechanism,
AUV travels around a predefined trajectory to collect data from sensor nodes. The
proposed mechanism achieves a higher network lifetime: however, time complexity
of the network is compromised in achieving a better packet delivery ratio (PDR). A
stateless opportunistic routing mechanism is designed in Ghoreyshi et al. (2018) that
utilises the information of communication voids received from the neighbouring nodes.
An energy-efficient routing mechanism based on chaotic compressive sensing (Li et al.,
2018) is proposed that uses the random-access method and shortest path selection
to overcome the challenges in the underwater environment. A cross-layer mechanism
proposed in Tran-Dang and Kim (2019) includes two-hop forwarding mechanism for
data transmission. The selection of the forwarder nodes is made through cooperation to
improve the reliability of the relay node. The proposed mechanisms generate a reliable
network model that can last for a longer time; however, the computational overhead of
the network is increased during redundant transmission.
An artificial intelligence (AI)-based network model is designed in Su et al. (2019)
that uses on and off policy to train the network, and adaptively changes the routing
path based on the network condition. Moreover, the routing mechanism introduced in
Sivakumar and Rekha (2020) determines the set of forwarder nodes through memetic
flower pollination (MFP). The mechanism consists of multi-sink architecture and obtains
A novel geo-opportunistic routing algorithm for adaptive transmission 271
optimal route through the best fitness value. Both schemes increase the packet delivery;
however, the computational complexity is the major drawback of these schemes.
A two-way routing mechanism is proposed in Ali et al. (2019) that divides the
network into regions. One-hop and two-hop data forwarding is considered in a multiple
sinks’ environment. Each sink travels around its fixed path and collects data from sensor
nodes in one or two hop fashion to improve the packet delivery. Rahman et al. (2018)
analyse the mobility of the sensor node by proposing a hybrid model of VBF and
spherical division to improve the delivery of data, and consequently decreases the energy
consumption. In an aim to reduce the delay in opportunistic routing, Mazinani et al.
(2018) have designed a mechanism that uses the time of arrival for nodes to find the
node’s location. All these mechanisms aim to reduce the network’s energy consumption
at the expense of increasing the communication overhead.
The authors propose an energy balanced model in Feng et al. (2019) where they
consider the selection of relays based on the depth of the nodes. Moreover, they
balance energy consumption by adaptively changes the energy level of nodes. However,
the end2end delay is compromised in improving network performance. To reduce the
total end2end delay of the network, Zhong et al. (2018) provide a direct transmission
mechanism with the presence of mobile nodes. The decision to send data directly to the
sink or through mobile nodes depends on the network’s condition. These mechanisms
do not provide optimal data delivery due to excess of direct transmission.
Coutinho et al. (2020) propose a mechanism that selects next hop forwarder nodes
based on four parameters, namely energy, link quality, density, and packet advancement.
The transmission power level of nodes is adjusted to reduce the energy dissipation of
nodes. The proposed mechanism improves the PDR in expense of high end2delay. The
introduced mechanism in Javaid et al. (2017) exploits the information of sensor nodes,
and selects forwarder nodes based on the available energy of the nodes in the network.
The nodes with high residual energy are selected as data forwarders thus, reducing the
traffic load on low energy nodes in the network. However, with the passage of time, the
performance of the network deteriorates due to high packet drop. The high packet drop
occurs when the residual energy of each node becomes lower than the residual energy
of the network.
Sher et al. (2018) improve the data delivery by introducing a four-way mechanism
to avoid void holes and reduce collision in the network. For efficient data delivery, the
forwarder nodes are selected based on the residual energy of the nodes. The PDR of
all four schemes is improved, however, the energy consumption and end2end delay are
still compromised. The work proposed in Usman et al. (2020) uses the mechanism of
terrestrial routing in underwater IoT networks. Different cases are considered in which
single sink and multi-sinks are deployed to analyse the performance of the network. The
proposed mechanism enhances the network performance in expense of high deployment
cost.
To increase the performance of the network, Ahmed et al. (2019) introduce
cooperative and non-cooperative mechanisms for CH selection. A hybrid energy
equating game is played cooperatively and non-cooperatively between nodes for the
selection of CHs. The high residual energy nodes with minimum pay off are selected as
CHs for data transmission. The PDR is increased and energy consumption is minimised.
However, the end2end delay of the network is increased due to high computational cost.
A sector-based opportunistic routing is performed in Celik et al. (2020) that searches
for optimal paths for efficient data delivery towards the destination. It exploits the
272 T.A. Alghamdi et al.
topology information locally and globally from the IoT sensor nodes. Optimal routing
is performed with minimum energy consumption and low end2end delay. However, the
total travel distance of the proposed scheme is relatively high that results in increasing
the total delay of the network. The designed protocols in Butt et al. (2019) use
Dijkstra algorithm to reduce the energy consumption of nodes and attain reliable data
transmission. However, the delay is increased while routing the data through greedy
paths towards the destination. Awais et al. (2019) reduce the energy consumption of the
network by providing two schemes. Both schemes reduce the interference and collision
in the network by selecting forwarder nodes based on a greedy approach. Reliable data
delivery is achieved in expense of high end2end delay.
Several authors have discussed malicious and random attacks in the IoT networks.
These attacks result in removing the most important nodes in the network. Therefore,
making the network robust against these attack is the main focus of the researchers. Qiu
et al. (2017, 2019) aim to increase the network resilience using multi-population genetic
algorithm. A similar issue is solved using an adaptive robust evolutionary algorithm in
Qiu et al. (2020). Moreover, the proposed mechanisms based on cooperation and entropy
in Khan et al. (2020) increase the network performance against the malicious attacks.
However, high computational cost is involved in making the network robust against the
attacks.
Due to the importance of IoT networks, Chen et al. (2019) propose a
backpropagation machine learning approach for the construction of robust topology
against the malicious attacks. Similarly, a deep deterministic learning policy model is
proposed in Chen et al. (2020) to improve the network’s stability against the attacks.
Moreover, the research in Parra et al. (2020) detects different types of attacks through
a distributed deep learning approach. However, all aforementioned schemes focus on
making the network resilient against different types of attacks. They fail to focus on
other performance parameters like energy consumption, end2end delay and PDR. The
nodes failure problem is also seen in Bu et al. (2018). Here, a VBF mechanism is
adopted where the selected forwarder nodes close to the sink die rapidly, leaving out
the network’s communication gap.
3 Preliminaries
In this section, the acoustic network model for the ATGOR and MA-ATGOR is
discussed in detail to understand the acoustic communication architecture. Then, the
problem statement is briefly presented, which shows our motivation behind this
proposal. The details are given as follows:
3.1 Network model
The network model is built by assuming that ‘n’ sensor nodes are distributed in 3D
network field, a cube of volume ‘V’ is formed by this field. Moreover, this cube is
divided into ‘M’ small size cube having volume ‘v’, denoted as C1, C2, ..., CM. Each
small cube has its own identification (CI D), which represents the cube coordinates.
If two cubes are adjacent, it means they have a common side or a common corner.
Let suppose, the sensor nodes have uniform deployment with same initial energy and
each sensor generates equal bits of data per second. Where, each sensor measures a
A novel geo-opportunistic routing algorithm for adaptive transmission 273
few spatial factors like temperature and salinity. Such assumptions relate to clustering
methods which have the advantages of scalability and robustness (Harb et al., 2015).
The sensor nodes are assumed to be equipped with the acoustic modems for
underwater communications. Whereas, the sink nodes have both radio and acoustic
modems for terrestrial and acoustic communications, respectively. The assumptions on
which our proposed work is based upon are:
All sensor nodes know their locations at the time of deployment through the
localisation services (Zhou et al., 2010).
The sensor nodes are anchored and can only move randomly in a horizontal
direction because of the water currents (Coutinho et al., 2015).
The sink nodes are special nodes that receive many packets simultaneously
without any collision. In addition, they have no energy constraints.
Every sensor can adjust its transmission power dynamically for avoiding
unnecessary energy dissipation.
The sensor nodes in a small cube directly communicate with the sensor nodes of
the adjacent small cubes.
3.2 System models for sink deployment and mobility analysis
In our work, the energy, computational and memory resources for sink nodes are
unlimited contrary to the ordinary sensor nodes. Therefore, the deployment of multiple
sinks is varied to provide the maximum coverage and guaranteed connectivity among
the network nodes. In order to analyse the effects of sink deployment, we have used
two deployment schemes:
1 geospatial division-based sink deployment (GDSD)
2 GDSD with adaptive mobility (GDSD-AM).
In GDSD, let assume that a set of sensor nodes n={n1, n2, ..., ni}is uniformly
distributed over the 3D network field as shown in Figure 1. The volume of whole
network is uniformly divided into Vregions. There are total Snumber of sinks deployed
in the network. The S/2 sinks are deployed over water surface. The rest S/2 are
uniformly distributed inside the defined dimensions of the acoustic network field. The
sinks, which are deployed inside the water, covers the equal volume and every sensor
node has the ability to compute its distance from every sink. The small cube in Figure 1
shows the communication among the sensor nodes.
Aand Care represented as void nodes, responsible for resuming the communication.
The depths of both nodes are adjusted to Band D, respectively. After the depth
adjustment, the sensor nodes resume the data forwarding by finding a potential node.
The sensor nodes communicate in a multihop manner to forward the sensed data
towards the destination. Furthermore, the collected data is sent to the surface sinks via
mutihopping.
Similar to GDSD, a set (n)of sensor nodes is uniformly deployed over the network
field in GDSD-AM. There are Ssinks deployed in the network in GDSD-AM, which
are uniformly distributed in Vregions. Each region consists of S/nsinks deployed at
274 T.A. Alghamdi et al.
its top surface as shown in Figure 2. Where, nis any real number depending upon the
requirement of regions in the network field. A cube shown at the right side of Figure 2
exhibits the communication within a region from the network field. It can be seen that
the node Ais a void node and informs its nearest sink to adjust its depth to a new
position S2. Unlike to GDSD, the sink nodes adaptively adjust their depths.
Figure 1 GDSD system model
Χ
Α
Β
ψ−αξισ
ξ−αξισ
ζ−αξισ
Σατελλιτε
Μονιτορινγ Χεντερ
Σενσορ Νοδε
Σινκ
Ραδιο Λινκ
Αχουστιχ Λινκ
Figure 2 GDSD-AM system model
Α
Σ−1
Σ−2
Σατελλιτε
Μονιτορινγ Χεντερ
Σενσορ Νοδε
Σινκ
Ραδιο Λινκ
Αχουστιχ Λινκ
ψ−αξισ
ξ−αξισ
ζ−αξισ
A novel geo-opportunistic routing algorithm for adaptive transmission 275
3.3 Problem description
In this section, we explore the void hole problem. The void hole occurs due to
water currents, sparse deployment and depletion of node battery. Therefore, a dynamic
adjustment of transmission power is required for optimal energy dissipation to have
uninterrupted communication among the network nodes. The void regions are illustrated
in Figure 3.
Figure 3 Void hole problem (see online version for colours)
Σινκ
Σενσορ Νοδεσ
Αχουστιχ Λινκ
ςοιδ
Ραδιο Σιγναλ
Σατελλιτε
Μονιτορινγ
Χεντερ
Α
Σ
Although, the adaptive transmission power avoids the void regions, however, the
geographic and opportunistic depth adjustment-based routing (GEDAR) utilises the fixed
transmission power to establish communication among the network nodes (Coutinho
et al., 2015). In GEDAR, a source node Sinitiates the communication via establishing
links with its neighbour nodes that are within its transmission range. To recover the data
packet, the GEDAR performs adjustment of depth using based apparatus to resume the
network operations shown in Figure 3. However, after one hop recovery of void nodes,
the sensor nodes are still prone to transmission failure. Thus, moving the void node at
a new depth does not eliminate the threat of void hole occurrence and also consumes
high energy with increased delay.
4 The proposed schemes
In order to have continuous operations in harsh acoustic environment, the problems of
data loss, signal attenuation, data re-transmissions and void hole occurrence need to be
276 T.A. Alghamdi et al.
addressed very carefully. A few of them are discussed in the subsequent section. For
delivering data packets successfully at the destination in multi-hop manner by avoiding
void holes, an adaptive transmission-based routing algorithm is proposed to address both
void hole and mobility issues in UWSN.
4.1 ATGOR
To describe the proposed transmission strategy clearly, we make some definitions as
follows:
1 Effective transmission region (ETR) is the region in which every sensor node in
another cube is closer to the destination as compared to the source node.
2 Eligible neighbour cube (ENC) is defined as the neighbouring cube of a source
cube, which has the lower depth than the source cube.
3 Eligible neighbour node (ENN) is the neighbouring node of a source node
possessing less distance from the destination than the source node.
Given that the sensor node deployment is uniform over a network field; where, sensor
nodes continuously sense, gather and transmit their data to the nearby sink. The
consumption of energy is dependent on the distance between source and destination.
Now, if we keep the transmission power constant then a lot of energy will be consumed
which results in shorter network lifespan. The detailed discussion of working mechanism
of ATGOR is discussed as follows:
4.1.1 Enhanced periodic beaconing
In ATGOR, the sensor nodes obtain the location information of the reachable sinks and
the neighbouring sensor nodes through the dissemination of periodic beacon messages.
Every sink deployed at the surface of the water have global positioning system to
determine its location. The sink nodes disseminate beacon messages to localise the
underwater nodes. The beacon message, which is sent after specific intervals includes
sequence number, ID, and its Xand Ycoordinates. Recent messages that are sent from
the sink are identified by sequence number. The Zcoordinate is not considered here,
the reason is that the sinks nodes can only move in horizontal direction and cannot
have vertical movement. Moreover, each sensor node has node’s ID, sequence number,
CI D and X,Yand Zcoordinates of nearby sinks. The sequence number is increased
after specific time interval sseconds. When a most recent beacon message is received,
the entry is refreshed on the basis of sequence number. To minimise the collisions, the
periodic beaconing proves to be useful than the flooding mechanism of control messages
(Yu et al., 2015). Every node checks its sequence number when they receive beacon
message and if it is greater than the earlier received beacon then it updates its entries
of the reachable sinks. Similarly, the neighbouring nodes update their entries upon
receiving the beacon message if the entries are more recent than the existing entries.
When a sensor node broadcasts a beacon message, it sets up a new timer. Algorithm 1
shows the mechanism of enhanced periodic beaconing.
A novel geo-opportunistic routing algorithm for adaptive transmission 277
Algorithm 1 Enhanced periodic beaconing
Broadcast periodic beacon (node)
Qa new beacon message
if the timeout expired then
Q.coordinate loc(node)
Q.seq num seq num(node)
Q.CI D CI D(node)
Broadcast Q
Set a new timeout
end if
Receive beacon (node, Q)
if Q is from a sink then
if Q.seq-num >seq-num then
update the reachable sink entry
else
update neigh(Q.seqnum, Q.C ID, Q.I D, Q.loc)
end if
end if
4.1.2 Determine the next-hop small cube
Let us say, a sensor node Sis located in the logical cube C5, as shown in Figure 4. The
solid line circle around Sshows its transmission range. The upper hemisphere of the
circle represents the ETR of the S. It can be seen that there is no sensor node available
in the next small cube to S.
Figure 4 Adaptive transmission in ATGOR
Σ
Χ1
Χ2
Χ3
Χ4
Χ5
Χ6
Τ1
Τ2
Here, C6represents a void cube between source and destination node. Figure 4 shows
that Sadjusts its transmission power dynamically to find small cube nearer to it.
278 T.A. Alghamdi et al.
The big solid line circle shows the adjusted transmission power of node S. Whereas,
the C2is within the ETR, thus, C2can be called as ENC for the source node S. The
sequential steps of small cube selection are given in Algorithm 2.
Algorithm 2 Election of ENC
Initialise all the parameters
Node nireceives packet from node nj
Computes the coordinates of source node nj
Acquires CID
Find ENC in ETR
if An ENC is found by nithen
Acquire its CID
end if
if There exist any void cube then
From Tmax, other level is chosen
Go to 5
end if
4.1.3 Forwarder set selection
In traditional multi-hop routing, a single node is selected as a forwarder where the
probability of re-transmission increases if void hole occurs. The opportunistic routing
is a definitive solution in which the set of forwarder nodes are elected to transmit one
data packet. If one node fails to deliver the data packet then a node with less priority
is nominated to proceed with the transmission process. In our proposed work, the
opportunistic routing is used to select a set of neighbouring nodes from the ENC. The
selected neighbouring nodes are named as ENN and prioritised based on the depth by
assigning a holding time. Whereas, we have also exploited the geographic routing in
combination with opportunistic routing to select a node from ENN set that has highest
residual energy among all. It is necessary to avoid the hidden terminal problem within
a CID in opportunistic routing. If there is comes any situation that the nodes with
highest priority are unable to send data packets, then the nodes with low priority begins
to send data packets. All important step of forwarder node set selections are shown in
Algorithm 3.
Algorithm 3 ENN set selection
begin
ENN forwarder set selection
All nodes within the CID are found
The coordinates of all nodes are acquired in ENC
CID is acquired
The priorities are assigned to ENNs
end
4.1.4 Calculating the holding time
To prioritise the ENNs, the holding time is calculated as follows (Coutinho et al., 2015):
th=tp+
k=1
D(nk, nk+1)
s+i×tproc,(1)
A novel geo-opportunistic routing algorithm for adaptive transmission 279
where tpis the propagation time, tproc denotes the processing time of the packet,
D(nk, nk+1)represents the Euclidean distance and sshows the sound speed of the
signal in the acoustic medium. The propagation time tpshows the delay required for
the complete propagation of the transmitted packet. This time is calculated as: tp=
(RD(ni,nj))
s: here, Rrepresents communication range.
4.2 MA-ATGOR
In the UWSNs, the awareness of the locations of the sensor nodes is very critical.
Furthermore, the pervasive coverage an aqueous environment in the presence of
inevitable node mobility is very challenging in the UWSNs. In this section, we present
a mobility aware routing protocol to predict the location of nodes, which are mobile
due to water currents. This section begins with the network model and followed by the
discussion of mobility prediction mechanism in ATGOR.
Figure 5 MA-ATGOR system model
Σενσορ Νοδε
Σατελλιτε
Μονιτορινγ
Χεντερ
Σινκ
Ανχηορεδ
Νοδεσ
4.2.1 System model of MA-ATGOR
The movement of underwater sensor nodes is not a complete random process. The
movement of one node is closely related to the movement of the nearby nodes because
of the inherent temporal and spatial correlations of sensor nodes. Thus, an ordinary
sensor node deduces its mobility pattern from the mobility pattern of its neighbouring
nodes. On the other hand, the anchor nodes also communicate with other anchor nodes
280 T.A. Alghamdi et al.
to analyse their mobility pattern. Moreover, the spatial correlation is measured for the
ordinary nodes in the neighbouring region. The anchor nodes serve as the intermediate
nodes between the sinks and the ordinary nodes. We assume that there are ‘n’ ordinary
sensor nodes, which are deployed by following random uniform distribution over a 3D
network field. In addition to the network model of ATGOR, there are ‘j’ anchor nodes
deployed at uniform random locations as depicted in Figure 5.
Due to low capabilities of sensor nodes, it is desired to utilise the node energy
very efficiently. The anchor nodes are the powerful sensor nodes having ability to
communicate with sinks. The ordinary nodes communicate with the anchor nodes by
locating them within their communication ranges.
4.2.2 Localisation of anchor nodes and the mobility prediction
The anchor nodes periodically measure their locations for delivering data packet. Let
us assume, the anchor nodes predict their locations after a certain prediction period
T1’. For every anchor node, the mobility speed vector is represented through V=
[v1, v2, ..., vk]. Here, vkdenotes the average speed of the nodes at period ‘k’. We employ
a linear prediction model according to the mobility pattern of the nodes as formulated
in equation (2).
v(k) =
l
p=1
ςlv(kl),1pl, (2)
where pis prediction steps’ length and ςdenotes the coefficient of linear prediction
model. ςcan be computed from the localisation data, based on the previous prediction
step. In localisation period, ‘k’ is an anchor node, which measures its actual location by
disseminating the beacon messages to the surface sinks. At the same time, it measures
its expected new location based on the past measurements as follows:
Locn(j, k) = Loca(j, k) +
k
p=i
T1×v(p),(3)
here, Locn(j, k)is anchor node’s new location ‘k’ and Loca(j, k)is the actual location
of the anchor node ‘k’. We consider the kinematic model to predict the node mobility
due to water currents (Beerens et al., 1994). The kinematic model measures the node
speed in xand ydirections because the depths of the anchor nodes are fixed in the z
dimension as:
vx=k1λv sin(k2x)cos(k3y) + k1λcos(2k1t) + k4,(4)
vy=λv cos(k2x)sin(k3y) + k5,(5)
where vxand vyare the speeds in the xand ydirections, respectively, k1,k2,k3,k4,
k5,vand λshow the environmental factor(s), i.e., water currents and bathymetry.
A novel geo-opportunistic routing algorithm for adaptive transmission 281
4.2.3 Localisation of ordinary nodes and mobility prediction
The spatial and time correlations are utilised to perform mobility prediction for ordinary
sensor nodes. If the velocities of its neighbouring nodes are known, the velocity of node
n’ in the xand ydirections can be predicted as:
vx,y(n, k ) =
n
j=1
φnj vx,y(j, k),(6)
where nis the number of neighbours and φis the related coefficient between the
velocities of the anchor node and the ordinary node. It is computed as: φnj =
1/dnj
n
j=1(1/dnj ).Here, dnj denotes the Euclidean distance between nodes nand j. This
mobility prediction is incorporated into the localisation process and then the location of
an ordinary sensor node can be calculated as:
Locexp(n, k + 1) = Locc(j, k + 1) + T1×v(j),(7)
where Locexp(n, k + 1) and Locc(j, k + 1) represent the expected location of the
ordinary sensor node and the current location of the reference node, respectively. After
performing this localisation, the routing is performed similar to the ATGOR.
5 Linear programming and graphical analysis
Linear programming is widely accepted and used quantitative optimisation technique,
it provides optimal solutions. The optimal solution is obtained by formulating a linear
optimisation problem, which consists of an objective function to be maximised or
minimised, decision variables and the linear or nonlinear constraints. The decision
variables are continuous and can take on any real value, which satisfies the defined
constraints.
In this section, we optimise the energy consumption and the PDR for our proposed
schemes and geometrically represent the feasible regions. Let us say, a set ‘B’ of
values forms a feasible region. This set is represented in the Euclidean plane formed
by decision variables according to the defined constraints. A subset of values from ‘B
is said to be the feasible solution if it is within the feasible region. Among all feasible
solutions, the one that maximises or minimises the objective function is an optimal
solution.
Energy consumption minimisation: It is necessary to minimise the consumption of
energy in data communication for prolonging the network lifespan. For minimal
energy dissipation, we define the following objective function.
min
n
i=1
(Etx(i) + Erx (i)),(8a)
where Etx =l×Ptand Erx =l×Pr.l,Ptand Prare the packet size, power
required for transmission and power used in reception, respectively. The
constraints are:
Eresidual Etx,(8b)
282 T.A. Alghamdi et al.
Dcommunication Rmax,(8c)
Etx 0, Erx 0, Dcommunication 0.(8d)
Figure 6 ATGOR: energy consumption (see online version for colours)
0 0.5 1 1.5 2 2.5 3 3.5
Etx (J)
0
0.5
1
1.5
2
2.5
3
Erx (J)
Etx+Erx = 2.7
P2(0.048, 0.1)
P1(0.048, 0.002)
P3(2.6, 0.002)
P4(2.6, 0.1)
Figure 7 MA-ATGOR: energy consumption (see online version for colours)
0 0.5 1 1.5 2 2.5 3 3.5
Etx (J)
0
0.5
1
1.5
2
2.5
Erx (J)
Etx+Erx = 2.3
P2(0.0185, 0.1)
P1(0.0185, 0.0015)
P3(2.2, 0.0015)
P4(2.2, 0.1)
Graphical analysis for ATGOR: Consider a scenario, where energy consumption
of transmission and reception (with units of joule) can be represented as follows:
A novel geo-opportunistic routing algorithm for adaptive transmission 283
0.05 Etx +Erx 2.7, 0.002 Erx 0.1 and 0.048 Etx 2.6. Figure 6
shows that all the solutions in feasible region are acceptable. Now, each vertex of
the bounded region is tested as: P1: (0.048, 0.002), P2: (0.048, 0.1), P3: (2.6,
0.002) and P4: (2.6, 0.1).
Hence, it is evident that all the solutions in the bounded region are valid. Any
value of the energy consumption lying within this region is feasible for the better
network performance. A point where the energy consumption value is minimum,
is an optimal solution.
Graphical analysis of MA-ATGOR: Similar to ATGOR, we perform the graphical
analysis for MA-ATGOR by following bounds: 0.02 Etx +Erx 2.3, 0.0015
Erx 0.1 and 0.0185 Etx 2.2. Figure 7 shows the feasible region for
energy consumption, which consists of all the feasible solutions. Now, we test
every vertex depicted in Figure 7 at: P1: (0.0185, 0.0015), P2: (0.0185, 0.1), P3:
(2.2, 0.0015) and P4: (2.2, 0.1). Therefore, it can be concluded that all solutions
calculated by MA-ATGOR are valid.
PDR maximisation: We have a scenario in which we have 25 sinks and maximum
number of sensor nodes are 450. Our objective is to minimise the consumption of
energy with maximised PDR.
max(P DR),(9a)
the constraints are given as,
P kttx 0,(9b)
P ktrx 0,(9c)
P DR 0,(9d)
Dcommunication Rmax.(9e)
These are non-negative constraints for the number of packets transmitted and
received.
Graphical analysis for ATGOR: The number of packets successfully received at
the sink can be represented against the node density of network. The PDR of the
network at some specific node densities 150, 200, and 450, are 0.28, 0.65 and
0.89, respectively. Figure 8 shows the bounded region for PDR of the network at
different node densities. These points are within the feasible region and provide
feasible solutions for the network. The optimal solution is at the point where PDR
of the network is maximum.
Graphical analysis for MA-ATGOR: Similar to the above graphical methods, we
perform the graphical analysis of PDR for MA-ATGOR at node densities 150,
200 and 450, the obtained values are 0.33, 0.52, and 0.94, respectively. Thus, it is
proved that above calculated values are within the feasible region, seen from
Figure 9. The optimal solution for MA-ATGOR is the feasible solution with
maximum value.
284 T.A. Alghamdi et al.
Figure 8 ATGOR: PDR (see online version for colours)
0 100 200 300 400 500 600 700 800
Νοδε δενσιτψ
0
0.5
1
1.5
2
2.5
Π∆Ρ
Λ1
Π3(0.87, 450)
Π1(0.28, 150)
Π2(0.46, 200)
Figure 9 MA-ATGOR: PDR (see online version for colours)
0 100 200 300 400 500 600 700 800
Νοδε δενσιτψ
0
0.5
1
1.5
2
2.5
Π∆Ρ
Λ1
Π3(0.94, 450)
Π2(0.52, 200)
Π1(0.33, 150)
6 Simulation results and discussion
To verify the efficiency of ATGOR in terms of void hole avoidance and maximised
PDR, it is compared with a depth adjustment technique named as GEDAR.
A novel geo-opportunistic routing algorithm for adaptive transmission 285
6.1 Parameter settings
For conducting simulations, 150–450 sensor nodes and 25 sinks are deployed in 1,500
m×1,500 m ×1,500 m region. 20% of deployed nodes are anchored nodes. The initial
energy of each node is 10 W. 150 m, 200 m, 250 m, 300 m, 350 m, 400 m and 450 m
are different transmission ranges for sensor nodes. The size of data being sent in the
network is 150. The amount of energy consumed in transmission of data and reception
of data are 2 W and 0.1 W, respectively. Moreover, the energy for idle state is 10 mW.
6.2 Performance metrics
The proposed protocol is evaluated by considering:
1 PDR is the ratio of number of packets received successfully at sink to the packets
transmitted from the source nodes
2 fraction of void nodes, which represents the ratio of average void nodes to the
total number of nodes and the efficiency of the proposed transmission scheme
3 energy consumption per packet of each node is computed through the energy
depletion per received packet from each node and it is measured in joules.
6.3 Results and analysis
6.3.1 Sink deployment and sink mobility analysis
In this section, the proposed schemes is compared with GEDAR. For simulations,
150–450 sensor nodes are deployed over a (1,500 ×1,500 ×1,500) m3region. V
and Vhas a value of 8; each region has dimensions of (750 ×750 ×750) m3and
the transmission range of sensor nodes is 250 m. Sand Spossess the values 16 and
48, respectively. The energy consumption values for transmission, reception, idle state,
depth adjustment, packet size and data rate are taken from Coutinho et al. (2015).
Figure 10 shows the influence of node density on the network’s energy consumption.
This figure shows that energy consumption decreases with high node density. Moreover,
GEDAR has high consumption of energy due to depth adjustment. For low densities,
the energy consumption is high because more sensor nodes perform depth adjustment in
the absence of forwarder nodes. GDSD and GDSD-AM have low energy consumption
due to the distributed sink deployment.
The proposed deployment results in short communication distances because less
transmission power is needed to transmit the data packet towards the destination.
Figure 11 shows the PDR of the network. GDSD-AM has the highest PDR at different
node densities because the displacement of sensor nodes decreases with adaptive
mobility of sinks unlike to GDSD, GEDAR-16 and GEDAR-48. The improved PDR is
a result of the availability of sink nodes at shorter distances.
The energy constrained nodes do not perform the depth adjustment. Thus, the energy
consumed for depth adjustment is preserved in GDSD-AM. The fraction of void nodes
is shown in Figure 12. The proposed schemes have less fraction of void nodes due to the
geospatial division of nodes with uniform distributed sink deployment. In GDSD, the
centered sink deployment leads all the sensor nodes to transmit the sensed data at the
286 T.A. Alghamdi et al.
optimal distances. While, in GDSD-AM, the sinks move adaptively to the new positions
to recover the packet. On the other hand, a packet is transmitted at longer distances
to reach the surface sinks in the GEDAR, so the fraction of voids is high. Hence, the
proposed strategies help in improving the overall network performance with minimum
energy cost among the compared scenarios.
Figure 10 Energy consumption GEDAR, G-GDSD and G-GDSD-AM (see online version
for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
1
2
3
4
5
6
Ενεργψ Χονσυµεδ (ϑ)
ΓΕ∆ΑΡ
Γ−Γ∆Σ∆
Γ−Γ∆Σ∆−ΑΜ
Figure 11 PDR at sink nodes (see online version for colours)
150 200 250 300 350 400 450
Νοδεσ
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Π∆Ρ
ΓΕ∆ΑΡ
Γ−Γ∆Σ∆
Γ−Γ∆Σ∆−ΑΜ
A novel geo-opportunistic routing algorithm for adaptive transmission 287
Figure 12 Fraction of void nodes via GEDAR, G-GDSD and G-GDSD-AM
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.05
0.1
0.15
0.2
0.25
Λοχαλ Νοδεσ
ΓΕ∆ΑΡ
Γ−Γ∆Σ∆
Γ−Γ∆Σ∆−ΑΜ
Figure 13 Energy consumption ATGOR, A-GDSD and AGDSD-AM (see online version
for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.5
1
1.5
2
2.5
3
Χονσυµεδ Ενεργψ (ϑ)
ΑΤΓΟΡ
Α−Γ∆Σ∆
Α−Γ∆Σ∆−ΑΜ
6.3.2 Analysis of the proposed schemes with sink mobility
For better understanding, we divide our simulation analysis of the proposed schemes
into two scenarios as follows:
Scenario 1
Figure 13 shows that the void nodes reduce as the density of nodes keeps on increasing.
The reason is that the sensor nodes adapt transmission range. A trade-off is found
288 T.A. Alghamdi et al.
between finding a reliable forwarder node and energy consumption. When any forwarder
node is not found by sensor nodes, then they adjust their depth. In this depth adjust,
a lot of energy is consumed. However, at the cost of this extra energy, the sensor
nodes can find reliable forwarder nodes. Moreover in ATGOR, the sensor nodes adjust
their transmission range adaptively to find the nearest sink and overcome void holes.
Therefore, our proposed strategy helps in avoiding void holes. Furthermore, we present
the ATGOR scheme with the attributes of GDSD and GDSD-AM. On the other hand,
Figure 14 shows that our proposed strategy has improved PDR. The reason is that
PDR improves with increased node density. Because, when the number of nodes in
a particular region is increased, there would be less chance of existing void holes in
the network. Moreover, when there are many nodes in a particular area then there
are great chances of finding reliable neighbour nodes, which results in enhanced PDR.
Additionally, by adjusting the transmission range, more sinks are found in shorter, which
also results in improved PDR. The energy consumed in sending a packet from each
node is shown in Figure 15. Energy consumption reduces with increased node density
because a lot of neighbours are available for packet forwarding.
Figure 14 Packets received at the sinks with different transmission level (see online version
for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
Π∆Ρ
ΑΤΓΟΡ
Α−Γ∆Σ∆
Α−Γ∆Σ∆−ΑΜ
MA-ATGOR predicts the node mobility pattern based on the neighbours movement,
which results in improved network performance. MA-ATGOR is further tested with the
sink deployment and mobility schemes. Figures 16, 17 and 18 show the fraction of
void nodes, the PDR and the energy consumption, respectively. It can be seen that
MA-ATGOR has less number of void nodes than the ATGOR. The sink deployment and
sink mobility schemes help in improving the network performance. The MA-ATGOR
reduces the communication cost because it minimises the consumption of energy, which
is due to the location prediction mechanism. In MA-ATGOR, high number of nodes
results in decreased consumption of energy, which further improves location coverage.
The MA-ATGOR overcomes the packet drop problem by predicting the locations
of the sensor nodes. With the increase of node density, the MA-ATGOR shows higher
A novel geo-opportunistic routing algorithm for adaptive transmission 289
PDR. The M-GDSD and M-GDSD-AM prove to be efficient in minimising the energy
consumption and fraction of void nodes. Moreover, the packets received at the sinks are
also increased.
Figure 15 Fraction of void nodes via ATGOR, A-GDSD and A-GDSD-AM
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.05
0.1
0.15
0.2
0.25
Λοχαλ Νοδεσ
ΑΤΓΟΡ
Α−Γ∆Σ∆
Α−Γ∆Σ∆−ΑΜ
Figure 16 Energy consumption MA-ATGOR, M-GDSD and M-GDSD-AM
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
Λοχαλ Νοδεσ
ΜΑ−ΑΤΓΟΡ
Μ−Γ∆Σ∆
Μ−Γ∆Σ∆−ΑΜ
Scenario 2
In this scenario, we analyse the impact of transmission range on the network
performance. The impact of different transmission level on PDR are shown in Figure 19.
150 m, 200 m, 250 m, 300 m, 350 m, 400 m and 450 m are transmission ranges, denoted
by T1,T2,T3,T4,T5,T6and T7, respectively.
290 T.A. Alghamdi et al.
Figure 17 Packets received at the sinks using MA-ATGOR, M-GDSD and M-GDSD-AM
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Π∆Ρ
ΜΑ−ΑΤΓΟΡ
Μ−Γ∆Σ∆
Μ−Γ∆Σ∆−ΑΜ
Figure 18 Fraction of void nodes via MA-ATGOR, M-GDSD and M-GDSD-AM
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Ενεργψ Χονσυµεδ (ϑ)
ΜΑ−ΑΤΓΟΡ
Μ−Γ∆Σ∆
Μ−Γ∆Σ∆−ΑΜ
The transmission levels are chosen by considering the suitable distances because too
long transmission range deteriorates the network efficiency. The transmission range
increases, which results in enhanced PDR. Hence, it is concluded that the packets sent
from source are reached to destination successfully.
Figure 20 shows the number of nodes at various transmission level. The increase
in the transmission range has positive impact on reducing the void nodes. As the
transmission level increases, the number of void nodes decreases. The issue of void
area us solved by high transmission level. Figure 21 shows the consumption of energy
in sending each packet. High transmission levels lead to greater transmission distances
A novel geo-opportunistic routing algorithm for adaptive transmission 291
resulting in greater energy consumption. At high node densities, the energy consumption
is less because high node density requires less void areas to overcome.
Figure 19 Packets received at the sinks (see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Π∆Ρ
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
Figure 20 Energy consumption per packet (see online version for colours
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.02
0.04
0.06
0.08
0.1
0.12
Ενεργψ Χονσυµεδ (ϑ)
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
Similarly, we perform the simulation analysis at various transmission ranges for
MA-ATGOR. Figure 22 shows less number of void nodes than ATGOR for same
transmission levels. The MA-ATGOR performs better localisation of sensor nodes,
resulting in improved network performance. It can be seen from Figure 23 that the
MA-ATGOR has minimum energy consumption than the ATGOR. When MA-ATGOR
encounters less number of void nodes, it results in less packet drop and minimum
292 T.A. Alghamdi et al.
re-transmissions. With mobility prediction of neighbouring sensor nodes, the PDR of
the network is increased as shown in Figure 24.
Figure 21 Fraction of void nodes with different transmission for ATGOR (see online version
for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.2
0.4
0.6
0.8
1
Λοχαλ Νοδεσ
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
Figure 22 Fraction of void nodes with different transmission for MA-ATGOR
(see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.2
0.4
0.6
0.8
1
Λοχαλ Νοδεσ
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
A novel geo-opportunistic routing algorithm for adaptive transmission 293
Figure 23 Energy consumption per message (see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.02
0.04
0.06
0.08
0.1
0.12
Ενεργυ Χονσυµεδ (ϑ)
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
Figure 24 Packet delivery ratio (see online version for colours)
150 200 250 300 350 400 450
Νυµβερ οφ νοδεσ
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Π∆Ρ
Τ1=150
Τ2=200
Τ3=250
Τ4=300
Τ5=350
Τ6=400
Τ7=450
7 Conclusions
In this paper, we have kept in mind the harsh environment of UWSN and proposed
two routing mechanisms that continuously report nodes’ activities towards the sinks
in the underwater network. Before performing routing activity, we initially divide the
3D network field into small logical cubes for efficient data transmission. We have
learned that the presence of multiple sinks can efficiently increases the delivery of data
packets. For this purpose, we have deployed multiple sinks in our proposed schemes.
Moreover, it is also concluded that due to adjustment of depth, the consumption of
294 T.A. Alghamdi et al.
energy is high. Therefore, the ATGOR performs limited depth adjustment of nodes. In
ATGOR, distribution of neighbours is considered for the selection of forwarder node.
In an empty adjacent cube, void regions are avoided by using ATGOR, because it has
ability to adjust its communication range. In this way, it selects most reliable node
within its new adjusted communication range. To further circumvent the effect of void
nodes and depth adjustment, the proposed MA-ATGOR performs the mobility prediction
for neighbouring nodes. The MA-ATGOR gives precise locations of the neighbouring
nodes, which reduces energy consumption and provides better network coverage. To
find out how we can maximise and minimise our objective function, we have considered
linear optimisation for the evaluation of our proposed model. It shows that the adaptive
transmission power, the geo-opportunistic routing and mobility prediction results in
improved PDR.
References
Ahmed, G., Zhao, X. and Fareed, M.M.S. (2019) ‘A hybrid energy equating game for energy
management in the internet of underwater things’, Sensors, Vol. 19, No. 10, p.2351.
Ali, M., Khan, A., Aurangzeb, K., Ali, I., Mahmood, H., Haider, S.I. and Bhatti, N. (2019)
‘CoSiM-RPO: cooperative routing with sink mobility for reliable and persistent operation in
underwater acoustic wireless sensor networks’, Sensors, Vol. 19, No. 5, p.1101.
Al-Salti, F., Alzeidi, N. and Arafeh, B. (2014) ‘A new multipath grid-based geographic routing
protocol for underwater wireless sensor networks’, in 2014 International Conference on
Cyber-Enabled Distributed Computing and Knowledge Discovery, IEEE, October, pp.331–336.
Awais, M., Javaid, N., Rehman, A., Qasim, U., Alhussein, M. and Aurangzeb, K. (2019) ‘Towards
void hole alleviation by exploiting the energy efficient path and by providing the interference-free
proactive routing protocols in IoT enabled underwater WSNs’, Sensors, Vol. 19, No. 6, p.1313.
Beerens, S.P., Ridderinkhof, H. and Zimmerman, J.T.F. (1994) ‘An analytical study of chaotic stirring
in tidal areas’, Chaos, Solitons & Fractals, Vol. 4, No. 6, pp.1011–1029.
Bu, R., Wang, S. and Wang, H. (2018) ‘Fuzzy logic vector-based forwarding routing protocol
for underwater acoustic sensor networks’, Transactions on Emerging Telecommunications
Technologies, Vol. 29, No. 3, p.e3252.
Butt, S.A., Bakar, K.A., Javaid, N., Gharaei, N., Ishmanov, F., Afzal, M.K., Mehmood, M.K. and
Mujahid, M.A. (2019) ‘Exploiting layered multi-path routing protocols to avoid void hole regions
for reliable data delivery and efficient energy management for IoT-enabled underwater WSNs’,
Sensors, Vol. 19, No. 3, p.510.
Celik, A., Saeed, N., Shihada, B., Al-Naffouri, T.Y. and Alouini, M.S. (2020) Opportunistic Routing
for Opto-Acoustic Internet of Underwater Things, arXiv preprint arXiv:2002.08420.
Chen, N., Qiu, T., Zhou, X., Li, K. and Atiquzzaman, M. (2019) ‘An intelligent robust networking
mechanism for the internet of things’, IEEE Communications Magazine, Vol. 57, No. 11,
pp.91–95.
Chen, N., Qiu, T., Mu, C., Han, M. and Zhou, P. (2020) ‘Deep actor-critic learning-based
robustness enhancement of internet of things’, IEEE Internet of Things Journal, Vol. 7, No. 7,
pp.6191–6200.
Chitharanjan, K. and Senthil Kumar, R. (2021) ‘A study of resource allocation techniques in cloud
computing’, International Journal of Business Information Systems, Vol. 36, No. 2, pp.254–269.
Coutinho, R.W., Boukerche, A., Vieira, L.F. and Loureiro, A.A. (2015) ‘Modeling and analysis of
opportunistic routing in low duty-cycle underwater sensor networks’, in Proceedings of the 18th
ACM International Conference on Modeling, Analysis and Simulation of Wireless and Mobile
Systems, November, pp.125–132.
A novel geo-opportunistic routing algorithm for adaptive transmission 295
Coutinho, R.W., Boukerche, A., Vieira, L.F. and Loureiro, A.A. (2015) ‘Geographic and opportunistic
routing for underwater sensor networks’, IEEE Transactions on Computers, Vol. 65, No. 2,
pp.548–561.
Coutinho, R.W., Boukerche, A. and Loureiro, A.A. (2020) ‘A novel opportunistic power controlled
routing protocol for internet of underwater things’, Computer Communications, Vol. 150,
pp.72–82.
Feng, P., Qin, D., Ji, P., Zhao, M., Guo, R. and Berhane, T.M. (2019) ‘Improved energy-balanced
algorithm for underwater wireless sensor network based on depth threshold and energy level
partition’, EURASIP Journal on Wireless Communications and Networking, Vol. 2019, No. 1,
pp.1–15.
Ghoreyshi, S.M., Shahrabi, A. and Boutaleb, T. (2018) ‘A stateless opportunistic routing protocol
for underwater sensor networks’, Wireless Communications and Mobile Computing, Vol. 2018,
Article ID 8237351, pp.1–18.
Gul, S., Javaid, N., Wadud, Z., Sher, A. and Ahmed, S. (2019) ‘Adaptive transmission based
geographic and opportunistic routing in UWSNs’, in Recent Trends and Advances in Wireless
and IoT-Enabled Networks, pp.283–290, Springer, Charm.
Han, G., Long, X., Zhu, C., Guizani, M., Bi, Y. and Zhang, W. (2019) ‘An AUV location
prediction-based data collection scheme for underwater wireless sensor networks’, IEEE
Transactions on Vehicular Technology, Vol. 68, No. 6, pp.6037–6049.
Harb, H., Makhoul, A. and Couturier, R. (2015) ‘An enhanced K-means and ANOVA-based clustering
approach for similarity aggregation in underwater wireless sensor networks’, IEEE Sensors
Journal, Vol. 15, No. 10, pp.5483–5493.
Javaid, N., Cheema, S., Akbar, M., Alrajeh, N., Alabed, M.S. and Guizani, N. (2017) ‘Balanced
energy consumption based adaptive routing for IoT enabling underwater WSNs’, IEEE Access,
Vol. 5, pp.10040–10051, DOI: 10.1109/ACCESS.2017.2706741.
Khan, M.A., Javaid, N., Javaid, S., Khalid, A., Nasser, N. and Imran, M. (2020) ‘A novel
cooperative link selection mechanism for enhancing the robustness in scale-free IoT networks’,
in 2020 International Wireless Communications and Mobile Computing (IWCMC), IEEE, June,
pp.2222–2227.
Li, X., Wang, C., Yang, Z., Yan, L. and Han, S. (2018) ‘Energy-efficient and secure transmission
scheme based on chaotic compressive sensing in underwater wireless sensor networks’, Digital
Signal Processing, Vol. 81, pp.129–137, ISSN 1051-2004, DOI: 10.1016/j.dsp.2018.07.006.
Mazinani, S.M., Yousefi, H. and Mirzaie, M. (2018) ‘A vector-based routing protocol in underwater
wireless sensor networks’, Wireless Personal Communications, Vol. 100, No. 4, pp.1569–1583.
Nicopolitidis, P., Papadimitriou, G.I. and Pomportsis, A.S. (2010) ‘Adaptive data broadcasting
in underwater wireless networks’, IEEE Journal of Oceanic Engineering, Vol. 35, No. 3,
pp.623–634.
Noh, Y., Lee, U., Wang, P., Choi, B.S.C. and Gerla, M. (2012) ‘VAPR: void-aware pressure routing
for underwater sensor networks’, IEEE Transactions on Mobile Computing, Vol. 12, No. 5,
pp.895–908.
Noh, Y., Lee, U., Lee, S., Wang, P., Vieira, L.F., Cui, J.H., Gerla, M. and Kim, K. (2015) ‘Hydrocast:
pressure routing for underwater sensor networks’, IEEE Transactions on Vehicular Technology,
Vol. 65, No. 1, pp.333–347.
Oliveira, D., Dantas, J., Rosa, N., Maciel, P., Matos, R. and Brinkmann, A. (2019) ‘A dependability
and cost optimisation method for private cloud infrastructures’, International Journal of Web and
Grid Services, Vol. 15, No. 4, pp.367–393.
Parra, G.D.L.T., Rad, P., Choo, K.K.R. and Beebe, N. (2020) ‘Detecting internet of things attacks
using distributed deep learning’, Journal of Network and Computer Applications, Vol. 163,
p.102662, ISSN: 1084-8045, DOI: 10.1016/j.jnca.2020.102662.
296 T.A. Alghamdi et al.
Qiu, T., Liu, J., Si, W., Han, M., Ning, H. and Atiquzzaman, M. (2017) ‘A data-driven robustness
algorithm for the internet of things in smart cities’, IEEE Communications Magazine, Vol. 55,
No. 12, pp.18–23.
Qiu, T., Liu, J., Si, W. and Wu, D.O. (2019) ‘Robustness optimization scheme with multi-population
co-evolution for scale-free wireless sensor networks’, IEEE/ACM Transactions on Networking,
Vol. 27, No. 3, pp.1028–1042.
Qiu, T., Lu, Z., Li, K., Xue, G. and Wu, D.O. (2020) ‘An adaptive robustness evolution algorithm with
self-competition for scale-free internet of things’, in IEEE INFOCOM 2020-IEEE Conference on
Computer Communications, IEEE, July, pp.2106–2115.
Rahman, Z., Hashim, F., Rasid, M.F.A. and Othman, M., (2018) ‘Totally opportunistic routing
algorithm (TORA) for underwater wireless sensor network’, PLoS One, Vol. 13, No. 6,
p.e0197087.
Ren, J., Chen, Z. and Zhang, Z. (2020) ‘Distributed data mining in grid computing environment’,
International Journal of Web and Grid Services, Vol. 16, No. 3, pp.305–320.
Sher, A., Khan, A., Javaid, N., Ahmed, S.H., Aalsalem, M.Y. and Khan, W.Z. (2018) ‘Void hole
avoidance for reliable data delivery in IoT enabled underwater wireless sensor networks’,
Sensors, Vol. 18, No. 10, p.3271.
Sivakumar, V. and Rekha, D. (2020) ‘A QoS‐aware energy‐efficient memetic flower pollination routing
protocol for underwater acoustic sensor network’, Concurrency and Computation: Practice and
Experience, Vol. 32, No. 4, p.e5166.
Su, Y., Fan, R., Fu, X. and Jin, Z. (2019) ‘DQELR: an adaptive deep Q-network-based energy-and
latency-aware routing protocol design for underwater acoustic sensor networks’, IEEE Access,
Vol. 7, pp.9091–9104, DOI: 10.1109/ACCESS.2019.2891590.
Tran-Dang, H. and Kim, D.S. (2019) ‘Channel-aware energy-efficient two-hop cooperative routing
protocol for underwater acoustic sensor networks’, IEEE Access, Vol. 7, pp.63181–63194,
DOI: 10.1109/ACCESS.2019.2916185.
Usman, N., Alfandi, O., Usman, S., Khattak, A.M., Awais, M., Hayat, B. and Sajid, A. (2020) ‘An
energy efficient routing approach for IoT enabled underwater wsns in smart cities’, Sensors,
Vol. 20, No. 15, p.4116.
You, M., Luo, W. and He, M. (2020) ‘Resource scheduling of information platform for general
grid computing framework’, International Journal of Web and Grid Services, Vol. 16, No. 3,
pp.254–272.
Yu, H., Yao, N. and Liu, J. (2015) ‘An adaptive routing protocol in underwater sparse
acoustic sensor networks’, Ad Hoc Networks, Vol. 34, pp.121–143, ISSN: 1570-8705,
DOI: 10.1016/j.adhoc.2014.09.016.
Yu, H., Yao, N., Wang, T., Li, G., Gao, Z. and Tan, G. (2016) ‘WDFAD-DBR: weighting depth
and forwarding area division DBR routing protocol for UASNs’, Ad Hoc Networks, Vol. 37,
pp.256–282, ISSN: 1570-8705, DOI: 10.1016/j.adhoc.2015.08.023.
Zhong, X., Chen, F., Fan, J., Guan, Q., Ji, F. and Yu, H. (2018) ‘Throughput analysis on 3-dimensional
underwater acoustic network with one-hop mobile relay’, Sensors, Vol. 18, No. 1, p.252.
Zhou, Z., Peng, Z., Cui, J.H., Shi, Z. and Bagtzoglou, A. (2010) ‘Scalable localization with mobility
prediction for underwater sensor networks’, IEEE Transactions on Mobile Computing, Vol. 10,
No. 3, pp.335–348.
... The article [25] proposes Adaptive Transmission-based Geographic and Opportunistic Routing (ATGOR) protocol for reliable communication between nodes. This is an enhanced version of Grid-based routing in which, to avoid redundant transmission and optimize the forwarder selection, it uses the small cube-based clustering approach. ...
Article
There is a remarkable need of advancement in data communication for application of sensor and ad-hoc networks for reliable and secure delivery of data. Due to the dynamic behavior of the network, it is an emerging demand to achieve reliability in data communication. The existence of different network environment features and diverse dynamic topology leads to the need of protocol design for efficient and reliable delivery of data in underwater acoustic sensor networks. The paper proposes energy efficient and reliable protocol design for routing data packets in Underwater Wireless Sensor Networks environment using sector based forwarding mechanism. The sector based forwarding mechanism helps in routing the reliable data in an efficient and secure manner The proposal suggests to model the network in such a way that the entire network is divided into sectors to optimize the hop count that further leads towards the energy efficient delivery of data. The mechanism of data forwarding is done vertically by maintaining sector wise sequence through pivot node which is considered as sector head. The proposed sector-based network topology mechanism is useful in identifying the number of hops used in communication along with enhancement in energy for effective data communication. The experiment performed and its consecutive results suggests that the proposed protocol enriches the energy utilization of the network. The enhancement of data communication reliability in Underwater Wireless Sensor Networks environment is managed by the mechanism of reduced hop count.
Article
Full-text available
Nowadays, there is a growing trend in smart cities. Therefore, Terrestrial and Internet of Things (IoT) enabled Underwater Wireless Sensor Networks (TWSNs and IoT-UWSNs) are mostly used for observing and communicating via smart technologies. For the sake of collecting the desired information from the underwater environment, multiple acoustic sensors are deployed with limited resources such as memory, battery, processing power, transmission range, etc. As the replacement of resources for a particular node is not feasible due to the harsh underwater environment. Thus, the resources held by the node needs to be used efficiently to improve the lifetime of a network. In this paper, to support smart city vision, a terrestrial based "Away Cluster Head with Adaptive Clustering Habit" (ACH) 2 is examined in the specified three dimensional (3-D) region inside the water. Three different cases are considered, which are: single sink at the water surface, multiple sinks at water surface and sinks at both water surface and inside water. "Underwater (ACH) 2 " (U-(ACH) 2) is evaluated in each case. We have used depth in our proposed U-(ACH) 2 to examine the performance of (ACH) 2 in the ocean environment. Moreover, a comparative analysis is performed with state of the art routing protocols including: Depth-based Routing (DBR) and Energy Efficient Depth-based Routing (EEDBR) protocol. Among all scenarios followed by case 1 and case 3 , the number of packets sent and received at sink node are maximum using DEEC-(ACH) 2 protocol. The packets drop ratio using TEEN-(ACH) 2 protocol is less as compared to other algorithms in all scenarios. Whereas, for dead nodes DEEC-(ACH) 2 , LEACH-(ACH) 2 and SEP-(ACH) 2 protocols' performance is different for every considered scenario. The simulation results shows that the proposed protocols outperform the existing ones.
Article
The reliability of Internet of Things (IoT) connected devices is heavily dependent on the security model employed to protect user data and preventing devices from engaging in malicious activity. Existing approaches for detecting Phishing, DDoS, and Botnet attacks often focus on either the device or on the back-end. Such independent approaches don't exploit the opportunity for analyzing attacks in a distributed fashion by utilizing the massive low-computational capabilities of IoT devices and the high-computational capabilities of back-end servers. In this paper, we propose a cloud-based distributed deep learning framework for Phishing and Botnet attacks. The model is comprised of two key security mechanisms working cooperatively; (1) a Distributed Convolutional Neural Network (DCNN) model embedded as an IoT device micro-security add-on for detecting phishing and application layer distributed denial of service (DDoS); and (2) a cloud-based temporal Long-Short Term Memory (LSTM) network model hosted on the back-end for detecting Botnet attacks and ingest CNN embeddings to detect distributed phishing attacks across multiple IoT devices. The distributed CNN model, embedded into a ML engine in the client's IoT device, allows us to detect and defend the IoT device from phishing attacks at the point of origin. We created a dataset consisting of phishing and non-phishing URLs for training the proposed CNN add-on security model and selected the N_BaIoT dataset for training the back-end LSTM model. The joint training method minimizes communication and resource requirements for attack detection and maximizes the usefulness of extracted features. In addition, an aggregation of schemes allows the automatic fusion of multiple requests to improve the overall performance of the system. Our experiments show that the IoT micro-security add-on running the proposed CNN model is capable of detecting phishing attacks with an accuracy of 94.3% and a F-1 score of 93.58%. Using the back-end LSTM model, the model detects Botnet attacks with an accuracy of 94.80% and a F-1 score of 0.6666% using all malicious data points in the used dataset. Thus, the findings demonstrate that the proposed approach is capable of detecting attacks, both at device and at the back-end level in a distributed fashion.
Article
The extensive applications in the Internet of Things (IoT) have inspired a growing network scale. However, due to the resources-limited IoT devices and the numerous cyber-attacks against applications, maintaining the robustness and communication capabilities for the applications is increasingly challenging. In this paper, we consider IoT network topologies that provide robust communication for heterogeneous networks and study the networking stability of IoT devices and the intelligent evolution computing in network architectures. We explicate the network robustness problem both for the network architecture and the resistance to cyber-attacks. For the network architecture, we optimize the robustness of IoT network topology with a scale-free network model which has good performance in random attacks. In the case with the resistance to cyber-attacks, a deep deterministic policy learning (DDLP) algorithm is proposed to improve the stability for large scale IoT applications. Simulations show that the proposed algorithms greatly advance the robustness of IoT network topology compared to other algorithms, with a less computational cost.
Article
In smart cities, the Internet of Things (IoT) consists of many low-power smart nodes. Its robustness is essential for protection of communication in data science against node failures caused by energy shortage or cyber-attacks. Scale-free networking topology, widely applied in IoT, is effectively resilient to random attacks but is vulnerable to malicious ones in which high-degree nodes are made to fail. The prohibitively high computational cost of existing robustness optimization algorithms is an obstacle to efficient topology self-optimization. To solve this problem, a novel robust networking model based on artificial intelligence is proposed to improve IoT topology robustness to protect its communication. Using the Back-Propagation neural network learning algorithm, the model extracts topology features from a dataset by supervised training. The experimental results show that the model achieves better prediction accuracy, thereby optimizing the topology with minimal computation overhead.