Enhanced Adaptive Geographic Opportunistic
Routing with Interference Avoidance Assisted with
Mobile Sinks for Underwater Wireless Sensor
Talha Naeem Qureshi1, Nadeem Javaid1,∗
1COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
∗Corresponding author: www.njavaid.com, firstname.lastname@example.org
Abstract—Acoustic communication in Underwater Wireless
Sensor Networks (WSNs) is limited due to distinctive attributes
including communication channel high latency, multi-path fading
and exponential degrading on signal due to dynamic noise
characteristics. Inappropriate selection of forwarder node leads
to dramatic death due to inefﬁcient, unbalanced energy depletion
that results in creation of void hole for neighboring nodes.
In most of scenarios, forwarder nodes are over penalized by
selecting same node all the time. Enhanced adaptive mechanism
is proposed to cater over penalization of forwarder node by
adaptively lowering the priority of node gradually along with
residual energy. Proposed routing protocols are based on geo-
opportunistic routing paradigm based on interference avoidance
which is assisted by mobile sinks. More speciﬁc and uniﬁed
routing decisions are formed by dividing the whole network ﬁeld
into cubes. Forwarder nodes are elected on geographic location
of neighboring cubes depending on packet delivery probability.
Void node recovery mechanism is also proposed and evaluated by
deploying mobile sink to directly gather data from void nodes.
Extensive simulations are performed to evaluate the proposed
work. Simulations prove that proposed work signiﬁcantly in-
creases packet delivery and decrease fraction of void nodes.
Index Terms—Geographic routing; opportunistic routing; void
region communication; underwater sensor networks.
In Underwater Wireless Sensor Network (UWSN), group
of sensor nodes are deployed inside desired area that commu-
nicate through acoustic channel. Monitoring of physical envi-
ronment, temperature, humidity, battle ﬁeld, oceans, volcanoes
and many more are carried out by these sensors  . Sen-
sors are randomly deployed over the area of interest to gather,
sense, monitor and transmit the data of interest to desired
location. Deployed nodes are having limited energy source,
so energy is considered as key element in designing routing
protocol. Routing protocols are desired to prolong the network
lifetime by efﬁciently using the limited energy residing with
sensor node. Necessary factors required to be considered
for designing routing protocols for acoustic protocol are of
keen importance for successful communication among the
nodes. Most dominant factor over underwater channel is high
delay associated with acoustic signal propagation due to low
propagation speed 1500 m/s of sound signal and high bit error
rate due to unintended noise, acoustic variable medium nature,
very low bandwidth, large scale and multi-path fading.
Geographic routing protocols allows sender and reciever to
communicate without mainting complete data route due to
stateless nature. The algorithm only search for one optimal,
eligible and potential neighbor which act as a potential data
forwarder to handover data packet. Moreover, these routing
algorithms are highly optimal under high node density because
it follows greedy forwarding criteria to route data in multi-hop
scenario . There is an intense need for designing a routing
strategy which balances energy consumption and enhance net-
work lifespan are desired . Geographic routing protocols are
highly appreciated due to low energy consumption, scalability,
simple implementation and prolonging lifespan  .
Void holes are created due to sudden death of sensor node
in desired routing path that creates breakage in data route and
down stream nodes cannot forward sensed data to surfaced
sonobuoy or sinks. The issue of failure of data delivery can be
effectively controlled by opportunistic routing. In opportunistic
routing, a set of more than one forwarder nodes are elected
as a next hop instead of one. If any node from the set fails
even then the data is transfered by other nodes in set to the
next hop . However due to forwarding of multiple copies
of data by complete set results in delivery of multiple copies
at receiver and degrades performance. To avoid reception of
multiple copies, holding time and control messages system can
be added. Void nodes are also created due to death by over
penalizing of sensor node by electing it as a forwarder node.
Most of energy consumption is done during data transmission
and reception operation. Forwarder node has to transmit and
receive data of other nodes as well. By electing same node as
a forwarder node all the time results in leave the node as dead
due to all energy dissipation.
To overcome the over penalizing of nodes, enhanced adap-
tive criteria is proposed and evaluated. Mechanism suggested
dynamically lower the priority of node to the elected as
forwarder node with the decrease in residual energy of node.
Lower energy nodes are not elected as potential forwarder
node from complete set of forwarder nodes and life span of
network is enhances to signiﬁcant extend.
The rest of the paper is organized as follows. A com-
prehensive overview of underwater routing protocols are ex-
plained in section 2. Section 3 explains the requirements
of network model including architecture and probability of
2018 International Conference on Frontiers of Information Technology (FIT)
978-1-5386-9355-1/18/$31.00 ©2018 IEEE
packet degradation over data channel. Section 4, 5 and 6
are the proposed geographic opportunistic routing schemes
with enhanced adaptive probability. Section 7 is discussion
over results of proposed evaluated algorithms against already
present algorithms. Section 8 concludes the whole paper.
II. RELATED WORK
Many existing energy efﬁcient, void hole recovery position
based routing protocols are proposed for USNs. Mainly rout-
ing protocls are of two categories,geographic location based
routing protocols and opportunistic routing protocols.
Geographic location coordinates of sender and receiver is
used by protocol to establish data connection. Data is trans-
fered closed to the destination on basis of physical location
of nodes in between the path. These protocols are efﬁcient
in terms of scalability. Oppertunistic routing protocls are
designed to avoid retransmission problem. In Relative Distance
Based Forwarding (RDBF) , location information is used
to start ﬁnding routing path towards destination. Sender node
ﬁnds the optimal node to farword data based on distance and
ﬁtness function with respect to destination. Node having less
distance in comparison to sink have more chances to be elected
as next hop.
Routing and Multi-cast Tree-based Geo-Casting (RMTG)
 based on multi-factor mechanism including distance,
neighbor state, route discovery for identiﬁcation of node near
to destination and routing path maintenance procedure. ARP
protocol addressed void hole and data path link breakage
issues. Adaptive Routing Protocol (ARP)  assign different
priorities to data packets depending on application. Trade-off
between throughput and latency exists. Higher priority packets
are delay sensitive and are routed on priority.
In Depth Based Routing , authors used greedy forward-
ing to ﬁnd the high depth nodes. The approach leads to the
issue of energy hole creation. Depth information is broadcast
instead of location coordinates that results in unnecessary
energy dissipation. Same packet is broadcast by multiple nodes
of same depth that also results in collection at receiver end.
A Void Aware Pressure based Routing technique (VAPR)
 is proposed by Noh et al. VAPR uses geographic and
opportunistic routing for transmitting the sensed data from
sensor nodes to water surfaced sonobuoy. Scheme select the
forwarder nodes in upwards vertical direction. Noh et al. pre-
sented pressure based anycast routing algorithm (HydroCast)
for UWSNs . Different pressure level of nodes are used
for next-hop selection. Authors proposed adaptive push system
for disseminating the sensed data in .
W L. Rodolfo et al. proposed GEDAR  which used
depth adjustment for void hole recovery. Depth adjustment in
vertical direction of void nodes are done through winch based
apparatus. Greedy forwarding approach is used to forward data
and next hop selection. Recovery procedure is triggered to
search for alternate route in case void node. If no secondary
route is present then depth adjust is done to resume the
Different from the studies discussed above, our proposed
schemes Enhanced Adaptive Geo-spatial Division based Geo
Opportunistic routing scheme for Interference Avoidance
(EAGDGOR-IA), Enhanced Adaptive Geo-spatial Division
based Geo Opportunistic routing scheme (EAGDGOR-SM)
and Enhanced Adaptive Geographic Routing for Maximum
Coverage (EAGRMC-SM) use cube based geographic routing
decisions. Next hop potential forwarder node is elected based
on enhanced adaptive criteria that incorporates energy fraction.
Nodes priorities are adaptively changed and lowered by our
criteria along with decrease in residual energy. In order to save
potential forwarder nodes from over penalizing, threshold is
incorporated to exempt nodes with very low energy out of
potential forwarder selection criteria. Our schemes effectively
control energy consumption along with void hole reduction to
high extent. Packet delivery is also improved due to very low
number of retransmissions by incorporating packet delivery
A. Network Architecture
Three dimensional sensor equipped aquatic swarm archi-
tecture is considered for random deployment of under water
sensor nodes as shown in ﬁgure 1. Logical ﬁeld is further
divided into small cubes where each cube is equal to the com-
munication coverage diameter of the sensor node to perform
geographic routing based on cube identity. The whole ﬁeld is
divided into Cn=[c1,c
3, ..........cn]total number of cubes.
Nodes can communicate through the nodes of neighboring
cubes over acoustic link. We have considered two operational
modes of under water sensor nodes. First mode is monitoring
or sensing mode. Node sense the desired attributes from
environment in ﬁrst mode. Second is data transfer mode. In
this mode, sensor send the sensed data towards destination or
data collection point through acoustic link.
Our sensor equipped aquatic swarm architectonic consists
3, ..........nn]under water sensor nodes
3, ..........sn]surfaced sonobuoys. All sen-
sor nodes are equipped with limited battery, computational
resources and memory. Sonobuoys have both acoustic and
radio communication capability along with global positioning
system. It uses acoustic link to gather data from underwater
sensor nodes and use radio link to transmit over to data
collection point  . Sonobuoys can move horizontally
due to water currents . Underwater sensor nodes can
adjust vertical depth through winch integrated mechanism or
by pressure release apparatus.
Network topology is considered as a graph G(t)=(V,E),
where tis the speciﬁc time interval, Vis the vertices of graph
relating under water sensor nodes and Eis the edges be-
tween vertices, directing acoustic communication link between
Figure 2 shows the mobile sink architecture. In some scenar-
ios under water sensor node fails to ﬁnd any neighbor within
communication range towards data collection point. This node
is termed as void node. Mobile sink can be affectively used
to eliminate void node issue. Mobile sink can move into the
water itself to directly collect sensed data from void node and
sent to the surface sonobuoy over acoustic channel.
Fig. 1: Static Network Architecture
Fig. 2: Mobile Network Architecture
B. Packet Delivery Probability
We estimate the packet delivery probability in this section
for further use in our proposed protocols. In geographic oppor-
tunistic routing we select a set of nodes as next hop forwarding
set on the basis of under water packet delivery probability
p(m, d), where mis the number of bits to be transfered over
distance d. For calculation we have used underwater acoustic
channel model  . Every acoustic data path has a path
loss factor that is deﬁned as signal fading or attenuation over
a distance dfor a data signal having ffrequency and also
refereed as large scale fading. Path loss can be obtained by
the below mathematical expression.
A(d, f )=dka(f)d(1)
Where signal absorption factor is a(f)dmeasured in dB/km
and signal broadcasting coefﬁcient is k. Average Signal to
Noise Ratio (SNR) can be obtained for signal over distance d
In above equation Ebis average transmission energy per bit
and noise power density constant is N(o). We have considered
Binary Phase Shift Keying (BPSK) modulation as used by
maximum acoustic transmitters  . Probability error
Pe(d)is as per mathematical expression below .
Packet delivery probability p(m, d)can be calculated by
using below equation.
EAGDGOR-IA utilized the idea od cubes as shown in
ﬁgure 1 for data transmission. All nodes of one cube can
directly communicate with all nodes in neighboring cubes
in three dimension. The cube having source node that need
to transmit data towards data collection point is considered
as Source Cube (SC). Algorithm ﬁrst ﬁnds the appropriate
optimal Neighboring Cube (NC) from all NCs to transmit data
and termed it as Destination Cube (DC). EAGDGOR-IA works
in two phases, in phase 1 it ﬁrst ﬁnds the DC to transmit data
then in phase 2 it ﬁnds the appropriate forwarder node in DC
to handover desired data.
In ﬁrst phase each sensor node obtain its location coor-
dinates through localization mechanism as per . All the
nodes propagate beacon message within communication radius
Rcto establish acoustic connection within range. Beacon
message contains cube ID, node ID, sequence number, location
coordinates and sonobuoys in communication range of under-
water sensor node as shown in ﬁgure 3. All the nodes maintain
neighborhood table for nodes and sonobuoys in range based
on received beacon messages. Periodic broadcast of beacon
message results in unnecessary energy dissipation. To control
energy consumption, only the speciﬁc change in neighbor list
or sonobuoys in range is transmitted in beacon message.
Fig. 3: Beacon message
SC containing sender node measures its euclidean distance
from nearby sonobuoys. Then all the NCs measure their
euclidean distance from their known sonobuoys. The NC
having shortest distance from sonobuoy is selected as DC if
no sonobuoy is directly in range of SC. All other NCs are also
prioritized according to distance to ensure backup in case of
any failure with DC.
After section of DC, we ﬁnd potential forwarder node from
all nodes in DC in next phase. In opportunistic forwarding,
multiple nodes at a time are selected as next hop forwarder to
avoid packet loss by any node in case of bad link quality
and high path loss to eliminate hidden terminal problem.
In geographic opportunistic routing, packet is broadcast over
multiple node set in the DC. It is very effective in reducing
packet retransmission due to failure of packet delivery to
destination and also efﬁciently save energy required for packet
retransmission. However, it will lead to increase in end to end
latency due to neighboring nodes should wait for packet to
deliver the end last destination in the forwarding set .Selection
of next hop forwarder set is done through advancement to-
wards destination (ADV) as proposed in . In EAGDGOR-
IA we calculate ADV for all nodes in DC as per mathematical
ADV (nd)=D(nd, sd)−D(ns, ss)(5)
Where D(ns, ss)is the euclidean distance between source
node and nearest sonobuoy and D(nd, sd)is the euclidean
distance between destination node and sonobuoy in range
of destination node. Whether node ndbelong to potential
neighbor set Fset or not, we use Normalized ADV (NADV) to
calculate ﬁtness of node as per mathematical equation below.
dis the euclidean distance between source node
sto destination node dand p(ds
d,m)is the packet delivery
probability over distance ds
dfor transferring mbits. The
objective to opportunistic routing is to calculate the subset
of potential forwarder nodes from set of all neighbors.
NADV for selection of potential forwarder nodes does not
consider energy level of nodes resulting in dead nodes due
to high magnitude of data transmission and reception by for-
warder nodes. This phenomenon over penalizes the forwarder
nodes, resulting in increase of void nodes due to failure of most
of neighbors after gradual passing time steps. To deal with over
penalizing of forwarder nodes, EAGDGOR-IA uses Enhanced
Adaptive NADV (EA-NADV) that considers residual energy
fraction of node while calculating node priority as forwarder
node. EA-NADV also adaptively decreases priority with the
decrease in energy level of node. Next hop forwarder node
is elected on base of highest EA-NADV, calculated as per
mathematical expression below.
Furthermore to control void nodes scenario due to dead
nodes, EAGDGOR-IA uses Threshold Energy (TE) to elect
node as potential forwarder node. Any node having energy
less than TE cannot be elected as potential forwarder. TE can
be determined by below mathematical equation.
Where Eois initial energy of node.
Void hole is a scenario in which a node failed to found
any neighbor in range close to surface in comparison to
its own depth to forward data to surfaced sonobuoys. Void
hole occur due to many reasons including movement through
water currents, complete energy dissipation of close surface
neighbor node and high bit error rate of propagation channel.
Several different methods are proposed for void hole recovery
including vertical depth adjustment, replacing physical power
source of sensor and forward data by long route. However
EAGDGOR-SM incorporate the concept of mobile sinks in
EAGDGOR-IA and evaluated the performance.
Mobile sinks Mn=[M1,M
3, ........Mn]are randomly
deployed inside deployment area as shown in ﬁgure 2. Mobile
sink does not have any energy constraint as it can recharge
its battery through solar power by moving to the surface after
consumption of 80% battery. During data forwarding operation
if any node ﬁnds itself in void hole through received beacon
messages. It broadcast the void hole declaration message with
its location coordinates to further transfer the message to
mobile sink. All the neighbor nodes receiving the void hole
declaration message further broadcast until it is received by
any mobile sink. On receiving the message by mobile sink,
it moves to the void hole node location and collects data to
further transfer it to surface sonobuoy.
In EAGRMC-SM we replaced all surfaced sinks with mo-
bile sinks Mn=[M1,M
3, ........Mn]to retrieve infor-
mation directly from nodes. Distribution of all mobile sinks
are uniform with in the deployment area. All nodes broadcast
beacon message to inform neighbor nodes regarding mobile
sinks location in range. Nodes transfer data to neighbor nodes
in neighbor DC to further handover it to nearest mobile sink in
range. During the void hole scenario with any node, respective
node broadcast void hole declaration message with location
coordinates which is further broadcast by neighbor nodes. On
receiving void hole declaration message by mobile sink, it
change its location to collect data from respective void hole.
In EAGRMC-SM mobile sinks are having additional memory
to store received data from nodes. The energy consumption for
movement of mobile sink does not have any impact because
after few rounds mobile sinks move to surface to transfer data
to data collection point over radio signal and get recharged.
In EAGRMC-SM all nodes send data directly to mobile
sinks in range or neighboring nodes in case of mobile sink
absence. In both cases source node calculates EA −NADV
for mobile sink in case of more than one sink in range as per
equation below to tranfer data.
VII. SIMULATIONS AND RESULTS
In this section, we discuss performance of our proposed
algorithms EAGDGOR-IA, EAGDGOR-SM, EAGRMC-SM
against recent and popular previously proposed routing
schemes GEDAR, GDGOR-IA, GDGOR-SM and GRMC-
SM. All evaluated protocols are compared based on extensive
simulations in Aqua-Sim . Performance of all competing
protocols are evaluated based on fraction of void nodes, packet
delivery ratio and energy consumption of each protocol as
all nodes are having limited energy. Deep onward analysis
of proposed protocols are carried out by changing trafﬁc
stress as well. Our objective is to analyze and evaluate how
our enhanced adaptive NADV criteria and modiﬁed neighbor
selection works with varying network density from low to high
densities. More details are discussed as follows.
A. Initial Topology Construction
To perform simulations, testbed topology of 150 to 450
nodes are deployed randomly inside the cubic region of 1500m
x 1500m x 1500m along with 45 sonobuoys over water surface.
Sonobuoys act as a sink to received data from underwater
sensor nodes trough acoustic link and send data to nearby
data collection point over radio link. Data generation follows
the poisson distribution with λ=[0.1,0.05] packets/min
for extremely low trafﬁc load scenarios, λ=[0.01,0.15]
packets/min for minimum trafﬁc loads and λ=[0.2,0.25]
packets/min for trafﬁc of medium load. Extended three di-
mensional version is adopted for meandering current mobility
/cite34 GEDAR to consider the effect of sub-surface currents
in simulations. Jet speed is considered to be 0.3 m/s. Nodes
may move out of deployment region due to horizontal mobility.
In all experiments, transmission range (Rc)of all nodes
are ﬁxed to 250m and data transmission rate is 50kbps. To
avoid collision at MAC layer we use Carrier Sense Multiple
Access (CSMA) protocol. Size of the packets depends on
payload and varies accordingly. We considered payload of 150
bytes including 20 bytes beacon message information. Power
consumption of each node depends on data transmission, data
receiving, idle state and depth adjustment. Energy consump-
tion values are Pt=2W,Pr=0.1W,Pt=0.01Wand
Em= 1500mJ/m respective to transmission, reception, idle
and depth adjustment operation. Average of 100 independent
runs of simulations are performed to get near optimal results
against each plotted value.
B. Performance Metrics
Performance metrics considered are discussed in this section
Fraction of Void Nodes It is the fraction of nodes that are
unable to ﬁnd any sonobuoy in range or potential neighbor to
carry forward data towards surfaced sonobuoys.
Packet Delivery Ratio It is the ratio of packets successfully
received at destination that is surface sonobuoys over total
transmitted packets for sensor nodes during network operation.
Ratio is deﬁned by following mathematical expression:
SUMSs is sum of all packets received at
3...........sn]and SUMNn is sum of all
packets generated by nodes Nn=[n1,n
C. Analysis Fraction of Void Nodes
In ﬁgure 4, comparison of fraction of void nodes in pro-
posed schemes EAGDGOR-IA, EAGDGOR-SM, EAGRMC-
SM against previously proposed routing schemes GEDAR,
GDGOR-IA, GDGOR-SM and GRMC-SM are shown. Results
show that proposed all three schemes have the least number of
void nodes as compare to previous state of art and recent four
schemes due to incorporation of fraction on energy in NADV.
Underwater sensors are equipped with limited energy. Most
of the nodes become dead due to high energy consumption
involved in acoustic underwater communication. Many nodes
deprive from potential neighbors with the increase in quantity
of dead nodes. Adaptive energy fraction in NADV section of
forwarder nodes will decrease the node priority to be selected
as next hop forwarder with the decrease in energy. This results
in signiﬁcant decease in void nodes. The results will keep on
improving with lowering the number of void nodes in all three
schemes along with increasing network densities. Under high
node density, the chances of node isolation is decreased with
the increase of potential neighbors as forwarder nodes.
$ % #!
$ % !"#!
Fig. 4: Fraction of void nodes
EAGDGOR-SM and EAGRMC-SM have further decreased
in void nodes fraction due to having mobile sonobuoys where
as EAGDGOR-IA considers the sonobuoy positioned at the
surface. They collect data themselves directly from void
nodes by diving into the water and moving to the coverage
area of void nodes.
D. Analysis Packet Delivery Ratio
Packet delivery ratio of all the proposed schemes are signif-
icantly better as compare to existing due to enhanced interfer-
ence avoidance mechanism as shown in ﬁgure 5. The packet
delivery ratio also increases with the increase in network
densities due to more number of nodes within communication
range, resulting is ﬁnding a potential forwarder node at very
less distance. In dense network topologies there is frequent
rotation of next hop potential forwarder node. This results
in keeping the maximum nodes alive due to swapping and
increase in packet delivery ratio.
EAGDGOR-IA performs better than GDGOR-IA with a
signiﬁcant improvement. EAGRMC-SM have also shown in-
crease in packet delivery ratio as compare to GRMC-SM.
Proposed schemes have better performance due to inclusion
of advance link quality metric while selection propcess of
potential neighbor to frword packet. EAGDGOR-SM have
comparatively same performance as compare to GDGOR-SM.
' ( ! "#
' ( ! "&$
' ( $%"&$
Fig. 5: Packet delivery ratio
In this paper, proposed routing protocols efﬁciently con-
trols unbalanced energy dissipation of forwarder nodes and
enhances network life span signiﬁcantly. Proposed work also
effectively decreases the fraction of void node occurrence in
network along with increasing packet delivery ratio due to
incorporation of packet delivery probability in EAGDGOR-IA.
Three dimensional division of network into cubes allows opti-
mal and speciﬁc selection of forwarder node depending on ge-
ographic location. Deployment of mobile sink in EAGDGOR-
SM and EAGRMC-SM results in signiﬁcant decrease in void
holes. Proposed work results in overall decrease in energy cost
and void holes in all scenarios along with increase in packet
 Akyildiz, I.F.; Pompili, D.; Melodia, T. Underwater acoustic sensor
networks: Research challenges. Ad Hoc Netw. 2005, 3, 257279.
 Vasilescu, I.; Kotay, K.; Rus, D.; Dunbabin, M.; Corke, P. Data collec-
tion, storage, and retrieval with an underwater sensor network. In Pro-
ceedings of the 3rd International Conference on Embedded Networked
Sensor Systems, San Diego, CA, USA, 24 November 2005; pp. 154165.
 Coutinho, R.W.L.; Boukerche, A.; Vieira, L.F.M.; Loureiro, A.A.F.
GEDAR: Geographic and opportunistic routing protocol with depth
adjustment for mobile underwater sensor networks. In Proceedings of the
2014 IEEE International Conference on Communications (ICC), Sydney,
Australia, 1014 June 2014; pp. 251256.
 Coutinho, R.W.; Boukerche, A.; Vieira, L.F.; Loureiro, A.A. A novel
void node recovery paradigm for long-term underwater sensor networks.
Ad Hoc Netw. 2015, 34, 144156.
 Coutinho, R.W.; Boukerche, A.; Vieira, L.F.; Loureiro, A.A. Geographic
and opportunistic routing for underwater sensor networks. IEEE Trans.
Comput. 2016, 65, 548561.
 Liu, M.; Ji, F.; Guan, Q.; Yu, H.; Chen, F.; Wei, G. On-surface wireless-
assisted opportunistic routing for underwater sensor networks. In Pro-
ceedings of the 11th ACM International Conference on Underwater
Networks and Systems, Shanghai, China, 2426 October 2016; p. 43.
 13. Li, Z.; Yao, N.; Gao, Q. Relative distance based forwarding protocol
for underwater wireless networks. Int. J. Distrib. Sens. Netw. 2014, 10,
 Dhurandher, S.K.; Obaidat, M.S.; Gupta, M. A novel geocast technique
with hole detection in underwater sensor networks. In Proceedings of
the 2010 IEEE/ACS International Conference on Computer Systems and
Applications (AICCSA), Hammamet, Tunisia, 1619 May 2010; pp. 18.
 Guo, Z.; Colombi, G.; Wang, B.; Cui, J.H.; Maggiorini, D.; Rossi,
G.P. Adaptive routing in underwater delay/disruption tolerant sensor
networks. In Proceedings of the Fifth Annual Conference onWireless
on Demand Network Systems and Services (WONS 2008), Garmisch-
Partenkirchen, Germany, 2325 January 2008; pp. 3139.
 Yan, H.; Shi, Z.J.; Cui, J.H. DBR: Depth-based routing for underwater
sensor networks. In Proceedings of the International Conference on
Research in Networking, Singapore, 59 May 2008; pp. 7286.
 Y Noh, U Lee, P Wang, BSC Choi, M Gerla, “VAPR: Void-Aware
Pressure Routing for Underwater Sensor Networks” - IEEE Transactions
on Mobile Computing, 2013
 Noh, Y.; Lee, U.; Lee, S.; Wang, P.; Vieira, L.F.M.; Cui, J.; Gerla, M.;
Kim, K. Hydrocast: Pressure routing for underwater sensor networks.
IEEE Trans. Veh. Technol. 2016, 65, 333347.
 P Nicopolitidis, GI Papadimitriou, AS Pomportsis “Adaptive data broad-
casting in underwater wireless networks”- IEEE Journal of Oceanic,
 H. Yan, Z. J. Shi, and J.-H. Cui, DBR: Depth-based routing for
underwater sensor networks, in Proc. 7th Int. IFIP-TC6 Netw. Conf.
Ad Hoc Sensor Netw., Wireless Netw., Next Generation Internet, 2008,
 Z. Zhou, Z. Peng, J.-H. Cui, Z. Shi, and A. C. Bagtzoglou, Scalable
localization with mobility prediction for underwater sensor networks,
IEEE Trans. Mobile Comput., vol. 10, no. 3, pp. 335348, Mar. 2011.
 L. M. Brekhovskikh and Y. Lysanov, Fundamentals of Ocean Acoustics.
New York, NY, USA: Springer, 2003.
 M. Stojanovic, On the relationship between capacity and distance in
an underwater acoustic communication channel, in Proc. 1st ACM Int.
Workshop Underwater Netw., 2006, pp. 4147.
 L. Freitag, M. Grund, S. Singh, J. Partan, P. Koski, and K. Ball,
The WHOI micro-modem: An acoustic communcations and navigation
system for multiple platforms, in Proc. MTS/IEEE Oceans, 2005, pp.
 H. Yang, B. Liu, F. Ren, H. Wen, and C. Lin, Optimization of energy
efﬁcient transmission in underwater sensor networks, in Proc. IEEE
Global Telecommun. Conf., 2009, pp. 16.
 T. Rappaport, Wireless Communications: Principles and Practice. En-
glewood Cliffs, NJ, USA: Prentice-Hall, 2002.
 Z. Yu, C. Xiao, and G. Zhou, Multi-objectivization-based localization
of underwater sensors using magnetometers, IEEE Sens. J., vol. 14, no.
4, pp. 10991106, Apr. 2014.
 P. Xie, Z. Zhou, Z. Peng, H. Yan, T. Hu, J. H. Cui, Z. Shi, Y. Fei,
S. Zhou, “Aqua-Sim: An NS-2 based simulator for underwater sensor
networks. In Proceedings of the MTS/IEEE Biloxi-Marine Technology
for Our Future: Global and Local Challenges” (OCEANS 2009), Biloxi,
MS, USA, 2629 October 2009; pp. 17.