Performance Evaluation of Congestion-Aware

Routing Protocols for Underwater Sensor Networks

with Multimedia Data

Shalini Jain1, Emmanuel S Pilli, M. C Govil D Vijay Rao

Department of Computer Engineering Institute for System Studies and Analyses
Malviya National Institute of Technology (MNIT) Metcalf House, Delhi, India
Jaipur, India doctor.rao.cs@gmail.com.
1
Shallu.jain@gmail.com.

Abstract Performance advancements in acoustic Underwater sensor nodes are designed to withstand the harsh
communication technology have fostered to provide technical isolated environment of the deep seas to perform reliably over
platform for numerous interdisciplinary applications ranging long time periods without repairing, servicing, or recharging.
from bathymetry, hydrographic surveys, disaster prevention, to As these nodes are incapable of independent or deliberate
tactical surveillance over underwater sensor networks (UWSN). lateral movement, they must be deployed in a spatial
Supporting real-time data transmission over error-prone UWSN arrangement that serves their purpose optimally, while being
is increasingly important as these networks become more widely stable against disruption by drifting due to undercurrents [2].
deployed. Existing UWSN routing protocols caters the
requirements of non-real time applications where as delay UWSN node deployments are sparse and hence prone to
sensitive applications requires solutions that can improve network congestion with a large volume and burst data input
efficiency and reliability, often dynamically throughout the event load such as multimedia applications. Such systems have
detection and data transmission session. Congestion control is many applications where real-time monitoring is essential as
vital to achieve a high throughput and a long network lifetime.
in military, surveillance and threat detection [1]. Network
In particular, it is important for a routing protocol to provide
congestion control by incorporating metrics like throughput, congestion occurs when load on the node or link exceeds its
delay, packet loss ratio, etc. This paper presents a comparative carrying capacity, which can lead to packet drops, packet
analysis of UWSN routing protocols over real time multimedia interferences, queuing delay or blocking of new connections
data, using H.265/HEVC encoded video sequences. The aim of and thus decay the quality of service (QoS). Such
this study is to extend the in-built support in UWSN routing repercussions are responsible for reduction in the network
protocols for transmission of real time data traffic. The throughput. As a consequence of reduced throughput may
applicability of congestion control protocols in UW framework result into excessive retransmissions to compensate for packet
has been explored and the performance characteristics of the loss leading the system into the state of congestion, even after
protocols have been studied under different multimedia load
the initial load has been reduced to a level which would not
conditions with varying mobility. This methodology has been
illustrated using the case studies in the military and ocean normally have induced network congestion. In order to
monitoring domains. accomplish the networks mission, maintaining the reliability
of the sensed data and improving the QoS, congestion control
Keywords UWSN; Congestion protocols; Multimedia becomes an important aspect.
traffic; Congestion Aware protocols; Simulation; Aqua-Sim In this paper, we propose a methodology to model a given set
of sensor nodes as point sources with specified transmission,
I. INTRODUCTION detection characteristics and deployment as a graph. The
Underwater Sensor Networks (UWSN) is multipurpose grids of proposed methodology has been simulated on Aqua-Sim
autonomous acoustic transceiver-receiver nodes with numerous software [3] that models the characteristics of the sensors, UW
interdisciplinary applications ranging from bathymetry, communication medium and node and link characteristics. We
hydrographic surveys, disaster prevention, to tactical explore the applicability of congestion control protocols in the
surveillance [1]. Acoustic transmission is the only viable option UW framework and study the performance characteristics of
for underwater communication as of now, as radio waves suffer the DBR [4] and VBF [5] protocols using H.265/HEVC
from severe attenuation and optical transmission from severe encoded real time video traces [6]. We illustrate this
scattering. As acoustic waves are essentially pressure waves
methodology using two case studies in the military and ocean
that traverse transversally across the medium, they mainly
monitoring domains. The underwater protocol performance is
undergo losses due to spherical spreading and attenuation due
to medium and signal characteristics [1]. The restrictive nature evaluated in various scenarios with varying network load and
of seawater imposes harsh constraints on the effective ranges, mobility.
energy efficiency, and reliability of the network elements.
The remainder of the paper is organized as follows. In Section source reduces its rate if congestion occurs and ACKs are lost.
II, we discuss the related work. In Section III, we discuss about Notably, this ACK based congestion detection is not reliable in
congestion in UWSN. In section IV, we identify the challenges UWSN because of the high and variable delays, which
imposed by underwater sensor networks while realizing increases the system response time considerably.
multimedia data over UWSNs. In Section V, simulation setup
and results are discussed. Finally, concluding remarks with Deshpande et al. [9], assumes that the initial reporting rate of
directions for future work are given in Section VI. the nodes near the sink is to be kept minimum and the rate
increases as it go away from the sink node. It contains three
II. RELATED WORK modules; Congestion Detection Algorithm, Differed Reporting
Rate Algorithm and Adaptive Flow Rate Control Algorithm
In this section we review the only available technique for (AFRC). The algorithm increases efficiency using differed rate
congestion control in UWSN and several existing congestion and is independent of topology but in underwater environment,
control techniques for terrestrial sensor networks. there is varying propagation delay that makes estimation
difficult.
A. Congestion control techniques for UWSNs:
The congestion problem is more thoroughly investigated in
The first paper that deals with the congestion in underwater [10], a strategy is proposed that uses relative success ratio
wireless sensor networks is presented in [7]. The proposed (RSR) value- the ratio of the number of packets transmitted
algorithm draws analogy from the cycle that phytoplankton and from MAC layer to the number of packets forwarded by
zooplankton undergo. The author elaborates a protocol that network layer to determine the congestion level at other nodes.
takes the system back to an equilibrium state, in case there This value is broadcast periodically by a node informing others
exists congestion, which is inspired by the way equilibrium is regarding its congestion state. The nodes take decision of
established among the Phytoplanktons and Zooplanktons. The selecting a route depending upon distance and RSR value. The
protocol tries to do this by minimizing the energy consumption use of a different node also helps to prevent continuous energy
and packet loss. It also detects shadow zones, where dissipation of same nodes. RSR value also helps to regulate
accounting waves cannot travel. data generation rate thus mitigating congestion and improving
However, the assumptions made restrict its functioning for a efficiency. Also critical data can be prioritized and can be
static tree like topology as the nodes are tethered at the seabed. saved from dropping. However, RSR value does not take into
This type of architecture supports connectivity between various account the lossy environment which can also affect the
nodes but if a dynamic architecture is considered, as free float throughput. Window size needed to accurately estimate RSR is
monitoring is required in many applications, such connectivity a critical parameter. Too small a window size, RSR estimate
is difficult to ensure, which is a critical issue for the success of may be inaccurate and too large shall increase processing time.
this algorithm. Also, bad channel and shadow zone mitigation Also, there is no provision for contention prevention.
requires topology reorganization which is difficult in tethered In addition, the congestion control protocols discussed in
architecture. [8][11][12] do not consider congestion due to shadow zones,
which is a severe drawback, since there are frequent packet
B. Congestion control techniques for terrestrial sensor losses due to high bit error rate of the underwater acoustic
networks: channel.

Many congestion controls, techniques [1] have been proposed In this paper, source node is simulated using H.265/HEVC
in literature for terrestrial sensor networks (land based). encoded real time video traces [6]. Our aim is to extend the in-
However, they are not directly applicable due to harsh built support and challenges present in UWSN routing
underwater environment and radio does not work well in water. protocols for transmission of real time Ocean monitoring
multimedia data. Event detecting nodes sends a file in the form
The classical TCP's congestion control protocol not suited for of a video trace consisting of two parameters:
UWSN as the long round trip time (RTT) in underwater
environment affect the throughput. It performs poorly as it Frame size
assumes that congestion is the only cause of packet loss and Inter-packet time
triggers rate reduction when packet loss is detected. Whereas,
in UWSNs packets can loss due to shadow zone as well [1][2] The performance is evaluated in various scenarios with varying
which results into reduced throughput. Therefore, we need network load and mobility in military and ocean monitoring
efficient mechanisms to find the cause of packet loss. domain. Furthermore, results are presented to show how
effectively the network resources are utilized. We use one
Furthermore, in [8], the performance analysis of CODA is geographic-based routing protocol i.e., VBF [5] and another
discussed. It is a well-designed protocol for congestion non-geographic-based routing protocols, DBR [4] for
detection and handling. When congestion is detected at each performance evaluation.
intermediate, node rests on the queue length and channel
sampling, backpressure messages are sent to the neighbors to III. CONGESTION IN UWSNS
slow down their sending rates and further propagating An underwater sensor network consists of a number of
backpressure towards the source. If source reports data at a nodes relaying data towards one or more sinks. As data
higher rate than a preconfigured threshold, it must receive a collection in sensor networks is primarily event based [13],
steady feedback of ACKs to maintain that rate. However, the
there is a sudden increase in the load that the nodes as well as B. Acoustic Physical Layer constraints
the link can handle. Therefore, a convergence zone is formed
with the sink at its apex, which increases the chances of In UWSNs, RF waves does not work well, instead acoustic
congestion at sink and the nodes around it. This can lead to signals are used which are highly unreliable in harsh
packet drops and packet interferences causing congestion and underwater environment. Due to limited bandwidth constraints
thus decreasing the QoS. of acoustic links, large delay spread and high time-variability
due to slow sound propagation in underwater environment,
In order to accomplish the networks mission, maintaining acoustic channels suffer from severe rate performance
the reliability of the sensed data and improving the QoS, limitations. It is also prone to large-latency and high bit error
congestion control becomes an important aspect. The generic rates caused by very high attenuation and Doppler distortion.
reasons for congestion are - All the above mentioned factors may lead to unpredictable
Buffer overflow channel capacity and increases the probability of frequent route
failure. These effects can be reduced to some limit, but it may
Lack of synchronization further drop the bandwidth.
Many to one nature of the network (sources to a
C. Mobility induced constraints
common sink)
Encountering these problems becomes more difficult when Due to under water current, sea animals and other activities
underwater constraints are considered. This area of research in large numbers of underwater sensor nodes are mobile from
case of UWSNs has not been much explored. Although a low to medium to speed. Nodes can move in a 3D volume
number of congestion control and avoidance schemes are without following any mobility pattern except the nodes
available for terrestrial sensor networks but they are not anchored at the sea bed or surface. Underwater nodes may
directly applicable underwater due to propagation delays, move at the speed of 2-3 knots/hour in a usual underwater
dynamic environment, node drift etc. condition. Mobility may leads to frequent routes change or
even break, which may put extra burden of control packets.
In order to make a network robust, a proactive strategy
that avoids congestion must be available and also a provision Delay sensitive applications might not be able bear the
for congestion mitigation is required to recover the network if rerouting time, since delay is a major problem when meeting
it is rendered impoverished by congestion. timeliness requirement of a real time multimedia data
transmission for applications like submarine detection.
IV. CHALLENGES FOR MULTIMEDIA DATA OVER UWSN
Due to years of research, the knowledge about terrestrial D. Multi-hop induced constraints
sensor/ land based network has been evolved and to a certain Underwater sensor nodes use acoustic signals which
extent, the reliability issues associated with it are resolved. On inherently have low range as compared to RF signals. Also,
the other hand, in UWSNs reliability is a major concern due to multiple hops are required to route the packets to destination.
inhospitable underwater conditions and noisy communication Along with this, limited battery of a node necessitates it to have
media. Also, discovering and maintaining an optimal path is a assistance of other nodes for data delivery. However, multi-hop
challenging task underwater due to mobility of nodes with delivery poses several obstacles like high delay, link failure,
water current causing frequent route failure and topology and increased interference in neighboring links. For instance,
changes. Furthermore, their characteristics and properties i.e., the metric end-to-end delay which increases linearly with the
high propagation delays, multipath, very high attenuation and number of hops may result into decreased network packet
Doppler distortion causes high bit error rates. These problems delivery ratio, resulting into reduced throughput. It is very
cause more trouble in case of multimedia data because these challenging in UWSNs network, to determine a path that
constraints limit the number of legitimate samples required to satisfies congestion control metrics. Later in this paper, we will
extract information from the data. Given below are the show that routing protocols alone cannot achieve robust video
challenges identified while using multimedia traffic over transmission in scenarios consisting of large number of
UWSNs. multimedia data sources and also in case of highly mobile
nodes moving in random directions.
A. Congestion constraints
Network congestion is a problem occurring during the E. Battery constraints
transmission of multimedia traffic and large files over the In underwater network nodes, battery power is limited and the
network. This happens because the sender tries to utilize all possibilities of recharging it with solar energy or Ocean
available bandwidth every time they have data to send. Due to Thermal Energy have not been much explored. For long term
this competition for bandwidth among nodes, the network is monitoring applications, node energy imposes a great
rendered impoverished by congestion. In UWSNs, this challenge as due to battery discharge the nodes will die out
problem is elevated further due to bad channel conditions. and will not be available for further monitoring thus inflating
Congestion makes it difficult to satisfy the draconian QoS the cost of deployment because of continuous node
requirements in high multimedia traffic applications. replacement.
Also, underwater sensors are prone to failures because of B. Performance Metrics
fouling and corrosion. One needs to design energy efficient The data delivery ratio and end-to-end delay were the
congestion control protocol for UWSN that incorporate fault metrics considered for the evaluation of the performance of our
tolerant approaches. routing protocol. The ratio of total data packets received at
destination i.e., sink to total data packets sent by source node
V. PERFORMANCE STUDY
i.e., underwater nodes is known as the Packet Delivery Ratio
In this section, we present the performance evaluation of (PDR). The average delay for all of data packets received
DBR and VBF multi-hop routing protocols for multimedia successfully at the sinks on the surface of the water is known as
traffic over underwater sensor networks. We show that routing the End-to-End delay (EED).
protocols alone shall not be able to achieve robust video
transmission in scenarios consisting of large number of video C. Results
sources and where nodes are moving with water current in This section presents the inferences from the series of
random directions. simulation results based on packet delivery ratio and average
end-to-end delay.
A. Simulation Configuration 1) Varying Network Scalability
We have run simulations using Aqua-Sim [3], underwater Figure 1-2 show packet delivery ratio and average end-to-end
sensor network simulation software, to evaluate the delay with increase in network size in terms of area and
performance of DBR, and VBF routing protocols during the number of nodes. Each source node is simulated with video
transmission of multimedia traffic. The chosen video stream is trace file consisting of frames of variable sizes.
the well-known CIF Sony Demo, whose size may be suitable
for current wireless mobile device displays [6]. The ISO/IEC Table 1: Simulation Parameters
and ITU Joint Video Team (JVT) working group distributes
reference software for SVC called JSVM (Joint Scalable Video Simulation Parameters Values
Model) that is used for the implementation of the H.265/HEVC
coding technique [6]. The video trace file used to simulate the Simulator Aqua-Sim V1.28
video sources during simulation is generated using JSVM the
frame rate is set to 30 frames per second (FPS). The MAC Simulation Time 800sec
protocol for channel access is Underwater-MAC. The
transmitter power for each node is set to 30 db and the Area 800 X 800 X 8000 meter
bandwidth is set to 20kbps. We have placed 50 to 200 hundred
nodes in an area of 800mX800mX800m 3-D area. We have Number of Nodes 50 to 200
used multi-sink architecture, since DBR with multiple sinks
has a better performance than single sink architecture. We have
Routing Protocols DBR, VBF
deployed three sinks randomly at the water surface. Sensor
node follows the random-walk mobility pattern, whereas the
sink is assumed to be stationery. We considered ideal Mobility model Random
environment (i.e., no obstacle) and also considered energy
expenditure only during transmission. Data Rate 20kbps
Transmitter power of nodes has been chosen carefully
because a very small transmitter power may lead to high packet No. of Simulation Runs 08
losses due to large number of hops from source to destination.
On the other hand, large transmission power will cause Speed of Sound 1500m/s
interference with neighboring transmissions. To evaluate the
performance, the video trace files obtained from [6] are Packet size 3500 bytes
converted into a format that is suitable for Aqua-Sim simulator.
At network layer a single frame is transmitted using a single T 15' C
packet. In this way, we are able to expend the network with
real-life video traffic instead of relying on VBR or CBR flows. PH 08
In non-geographic-based routing protocols, we use DBR for
performance evaluation and VBF as geographic-based
S 35ppt
protocols. The network parameters used during simulations are
presented in Table I.
Type of environment Shallow water
Our evaluation is done over 800 sec because this amount of
time is required to transmit the complete video trace file from a
source node. All sources are selected randomly and all
simulation results presented in this paper are average results of Background traffic was created in each scenario using
simulation processes. couple of CBR sessions. As shown in Figure 1, when the
network size is small, both the routing protocols demonstrate
lower EED values. When the network size increases, EED in process. On the other hand, the EDD in DBR is low as
VBF increases rapidly due to the 3 way handshake mechanism compared to VBF in large sized networks.
of VBF. As the number of nodes increases, the control
overheads increase. 2) Varying Node Mobility
Figure 3 and Figure 4 depict EED and PDR with increase in
node mobility. VBF EED increases with the mobility speed of
nodes. This is because due to high mobility the network
topology changes frequently triggering frequent virtual route
pipe reformation. We further note from Figure 3 and Figure 4
that the delay is not affected much initially but with increase in
mobility however, it increases again due to the large proportion
of lost traffic. PDR and EED do not change much with node
speed in case of DBR. The reason is that all routing decisions
are made locally based on nodes depth information. No
topology or route information needs to be exchanged among
neighboring nodes as compared to VBF.
Figure 1: Effect of Network Scalability on End to end delay

Figure 5: Effect of Network Load on End to end delay

Figure 2: Effect of Network Scalability on Packet Delivery Ratio.

Figure 6: Effect of Network Load on Packet Delivery Ratio

Figure 3: Effect of Network Mobility on End to end delay
As the size of network increases, the number of intermediate
nodes on a route also increases. Due to this, the probability of
route failure increases. We can observe from Figure 2 that only
DBR shows acceptable results for moderate sized networks.
This is because DBR has low control overheads and better
dynamic adaptability towards link failure due to its flooding
nature as compared to VBF. As shown in Figure 2, the PDR of
both routing protocols increases as the size of network
increases.
3) Varying Network Load
Finally, Figure 5 and 6 show the EED and PDR with increase
in network traffic. We increase number of video sources in
Figure 4: Effect of Network Mobility on Packet Delivery Ratio network by two in each round of simulation. Figure 5 depicts
This also increases the size of the routing table at each node that the increase in EED of DBR is much smaller as compared
resulting in increase of time required for route discovery to other protocols. Good amount of bandwidth is wasted due to
interference caused by one or two hop neighbors who lie in a [10] Bhuiyan, I. Gondal, J. Kamruzzaman, CAM: Congestion Avoidance and
node's transmission or carrier-sensing range. Mitigation in wireless sensor network in Vehicular Technology
Conference (VTC 2010 Spring), IEEE, pp. 15-21.
Furthermore, Figure 6 shows that the PDR of all routing [11] O.B. Akan, I.F. Akyildiz, Event-to-sink reliable transport in wireless
protocols decreases with increase in number of video sources. sensor networks, IEEE/ACM Transactions on Networking ,2005, pp.
The decrease in PDR is more rapid when the number of 1003-1016.
sources increases beyond 8. As the collisions due to [12] J. Kang, Y. Zhang, B. Nath, TARA: topology-aware resource
adaptation to alleviate congestion in sensor networks, IEEE Transactions
interference and contention on channel increases it reduces the on Parallel and Distributed Systems , 2007, pp. 919-931
PDR in network. [13] X. Yin, R. Huang, Y. Fang, S. Li, A fairness-aware congestion control
scheme in wireless sensor networks, in IEEE Trans. Veh Technology.,
VI. CONCLUSION pp 5225-5234.
In this work, we presented a comparative analysis of
underwater sensor networks routing protocols over real time
multimedia data. The key objective was to provide high
transmission quality for the data traffic under conditions of
congestion. Our analysis extends the in-built support for real
time multimedia data transmission in wireless routing
protocols. A series of simulations were conducted to evaluate
the strength of the existing UWSN routing protocols for to be
able to achieve the desired performance for multimedia data
over wireless networks in constrained congestion situation. The
performance of UWSN routing protocol has been analyzed and
compared in terms of, packet delivery ratio and average end-to-
end delay. We observe through simulation in Aqua-Sim that,
DBR outperforms VBF routing protocol in congestion and
constrained situation. In summary, the performance evaluation
results show that transmitting multimedia data using traditional
UWSN routing protocols alone is not sufficient to provide
acceptable QoS to end users. we also strongly present that
routing protocols alone cannot achieve robust video
transmission in scenarios consisting of large number of
multimedia data sources and also in case of highly mobile
nodes moving in random directions due to water current

References
[1] I.F. Akyildiz, D. Pompili, T. Melodia, Underwater acoustic sensor
networks: Research Challenges, in Ad Hoc Networks, pp.257-281.
[2] J.H Cui, J.Kong, M.Gerla, S.Zhou, The challenges of building mobile
underwater wireless networks for aquatic applications, IEEE Network,
2006, pp. 12-18.
[3] 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 MTS, pp. 17-21.
[4] H. Yan, Z. J. Shi, J.H. Cui, DBR: Depth-based routing for underwater
sensor networks, in Proceedings of the 7th International IFIP-TC6
Networking Conference on AdHoc and Sensor Networks, Springer, pp.
72-86.
[5] Xie, P., Cui, J.H, Lao, L, VBF: Vector-based forwarding protocol for
underwater sensor networks. In: Proceedings of IFIP Networking,
Springer 2006, pp.1216-1221.
[6] Patrick Seeling and Martin Reisslein. Video Transport Evaluation With
H.264 Video Traces. IEEE Communications Surveys and Tutorials
2012, pp. 1142-1165.
[7] M. C. Domingo, Marine communities based congestion control in
underwater wireless sensor networks, in Information Sciences, pp. 203-
221.
[8] C.Wan, S. Eisenman, A. T. Campbell, CODA: Congestion detection and
avoidance in sensor networks, in ACM SenSys, pp.266 279.
[9] V. Deshpande, P. Chavan, V. Wadhai, J. Helonde, Congestion control in
wireless sensor networks by using differed reporting rate, in World
Congress on Information and Communication Technologies (WICT), pp.
209-213.