2016 IEEE TrustCom-BigDataSE-ISPA
TrustCom/BigDataSE/ISPA
Data Sharing in Secure Multimedia Wireless Sensor
Networks
Muhammad Usman, Mian Ahmad Jan, Xiangjian He*, Priyadarsi Nanda
School of Computing and Communications
Faculty of Engineering and Information Technology
University of Technology Sydney, Australia
*Corresponding author: Xiangjian.He@uts.edu.au
AbstractThe use of Multimedia Wireless Sensor Networks
(MWSNs) is becoming common nowadays with a rapid growth
in communication facilities. Similar to any other WSNs, these
networks face various challenges while providing security, trust
and privacy for user data. Provisioning of the aforementioned
services become an uphill task especially while dealing with real-
time streaming data. These networks operates with resource-
constrained sensor nodes for days, months and even years
depending on the nature of an application. The resource-
constrained nature of these networks makes it difcult for the
nodes to tackle real-time data in mission-critical applications such
as military surveillance, forest re monitoring, health-care and
industrial automation. For a secured MWSN, the transmission
and processing of streaming data needs to be explored deeply.
The conventional data authentication schemes are not suitable
for MWSNs due to the limitations imposed on sensor nodes
in terms of battery power, computation, available bandwidth
and storage. In this paper, we propose a novel quality-driven
clustering-based technique for authenticating streaming data in
MWSNs. Nodes with maximum energy are selected as Cluster
Heads (CHs). The CHs collect data from member nodes and
forward it to the Base Station (BS), thus preventing member
nodes with low energy from dying soon and increasing life span
of the underlying network. The proposed approach not only
authenticates the streaming data but also maintains the quality
of transmitted data. The proposed data authentication scheme
coupled with an Error Concealment technique provides an
energy-efcient and distortion-free real-time data streaming. The
proposed scheme is compared with an unsupervised resources
scenario. The simulation results demonstrate better network
lifetime along with 21.34 dB gain in Peak Signal-to-Noise Ratio
(PSNR) of received video data streams.
Index TermsMultimedia WSNs, Surveillance, Fire Monitor-
ing, Health-Care, Automation, Streaming, Authentication, Clus-
tering, Error Concealment, PSNR.
I. I NTRODUCTION
Adaptive and secured video data streaming is grabbing re-
searchers attention in the eld of Multimedia Wireless Sensor
Networks (MWSNs). Such type of networks are deployed in
many applications such as multimedia surveillance, storage of
potentially relevant activities, trafc avoidance, enforcement
and control systems, advanced health care, environmental
monitoring, industrial process control and many more [1].
These networks, similar to any other WSNs, have limited
resources such as energy, computational power, transmission
range, available bandwidth, data rate and storage [2]. These
limitations render the support for resource-intensive opera-
tions which ultimately affect the Quality of Service (QoS)
2324-9013/16 $31.00 2016 IEEE
DOI 10.1109/TrustCom.2016.113
10.1109/TrustCom/BigDataSE/ISPA.2016.113
in these networks [3]. Such challenges in MWSNs demand
proper network security, better scheduling mechanisms for the
transmission of video data streaming, and energy optimization.
An ideal solution to fulll these demands is to design a reliable
and highly efcient Medium Access Control (MAC) protocol.
In MWSNs, MAC layer can guarantee efcient resource
utilization with reliable data transmission, thus maximizing
the network throughput and minimizing energy consumption.
MWSNs are deployed for mission-critical tasks for an
unspecied duration of time. Therefore, security considera-
tions need to be in place at the time of network design.
The resource-constrained nature of these networks coupled
with their unique characteristics, such as dynamic topology,
in-network processing, error-prone communication links and
scalability, makes the security provisioning challenging and
complicated. In addition, these networks are left unattended
without human intervention and base station supervision.
Instead, sensor-collected data is harvested intermittently by
a base station. Since data packets are retained on individual
sensors, securing these packets is both important and challeng-
ing. Sensor nodes operating in unattended environments face
a higher risk of security breaches. If any one of these nodes is
compromised, its sensitive data and security parameters will be
retrieved by an adversary to participate in malicious activities.
These networks face a diverse range of security challenges
at various layers, and they are vulnerable to a wide range
of attacks such as Sybil, homing, wormhole, sinkhole, se-
lective forwarding, resource-exhaustion and Denial-of-Service
(DoS) [4]. As a result, security and protection of sensors
against malicious behaviours is extremely important. However,
the resource-starving nodes cannot withstand the resource-
demanding nature of real-time multimedia trafc towards pro-
visioning of highly-robust, stronger and sophisticated security
services in MWSNs. Therefore, designing an effective, energy-
efcient and secured solution for any MWSN is a challenging
issue.
Traditional error resilience schemes provide better data pro-
tection against data alteration during transmission. However,
these techniques are not suitable for sensor nodes because they
require high computational power from the underlying pro-
cessing devices which ultimately lead to energy consumption
[5], [6]. Schemes like watermarking are robust against net-
work impairments. However, they incur excessive processing
overhead, so they are not suitable for network devices having
590

Fig. 1. Clustering-based multimedia wireless sensor network
limited battery power [7]. The cryptography-based techniques
are used to protect sensitive data by embedding them in
other data. They are light-weighted, but they cannot resist
against network impairments [8][11]. The stenography-based
techniques are also used to embed the secret data into other
data at bit level. However, these techniques are also unable to
resist against network impairments and are ineffective while
dealing with larger quantity of data [12].
A wide range of techniques are available to authenticate
the video data streams in error-prone wireless networks. Un-
like watermarking, cryptography and stenography-based tech-
niques, which are considered as pre-processing schemes, the
data authentication schemes are post-processing techniques.
These techniques make use of simple but specic signatures
or hash functions to classify an incoming data packet as
authorized or unauthorized [13][18]. The video data packets
can be given higher priority in MWSNs, as video processing
requires high computational power and computing resources
and is less tolerant to bit errors [19][23].
In this paper, we propose a centralized cluster-based hi-
erarchical data authentication approach for real-time video
streaming over MWSNs. Multimedia sensor nodes are grouped
into geographical clusters, and designated nodes, also named
cluster heads (CHs) [24], are assigned the task to collect data
from their neighbouringhood as shown in Figure 1.
Cluster-based hierarchical protocols are iterative in nature
and operate in rounds [25]. In each round, a new set of cluster
heads are elected to perform resource-intensive operations
such as data aggregation, data fusion and data transmission
to a centralized base station. Our proposed approach secures
MWSNs and maintains the received video quality by combin-
ing a lightweight data authentication with an Error Conceal-
ment (EC) technique, respectively. The major contributions of
our proposed scheme are as follow.
1) Our proposed scheme operates in two phases. First phase
deals with cluster formation and data authentication
using hashing. With the help of a hierarchical cluster-
based architecure and a lightweight data authentication
scheme, smooth video streaming can be achieved with-
591
out excessive energy consumption.
2) In MWSNs, packets drop frequently at various intervals.
For this purpose, a simple EC method is presented in
the second phase as a backup technique to maintain
the quality of received videos. Our approach provides a
proper analysis of proposed authentication scheme along
with the packet drops and perceived video quality.
The rest of the paper is organized as follows. The related
work is provided in Section II. The proposed scheme is briey
described in Section III. In Section IV, the experimental set-up,
comparison and evaluation are presented. Finally, the paper is
concluded in Section V.
II. R ELATED W ORK
In this section, rst we discuss the related work on security
issues faced by MWSNs, followed by data authentication and
error concealment in these networks.
In [26], the authors presented a survey which summarized
the challenges incurred by any MWSN while securing it. These
challenges and security issues need to be taken into account
while designing efcient and robust authentication protocols
for these networks. Three types of attacks, i.e., snoozing, sleep
deprivation and network substituition in WSNs, were reported
in [27]. The unattended nature of deployed region, privacy
assurance of the data and multimedia in-network processing
are some of the main challenges faced by these networks at
the design stage of secured authentication protocols. Fake data
injection is a commonly found type of attacks in MWSNs,
which was reported in [28]. Detecting this type of attacks
helps minimize the malicious intrusions. In [29], the authors
conducted a research on secure data collection in an unse-
cured WMSN. Secure data collection for sensitive networks is
important and poses crucial challenges.
Resources allocation was combined with stream authen-
tication to improve the video quality along with integrity
protection [30]. Although this scheme works well, it involves
crypto-hash functions, which are not strong enough to tackle
network impairments. To maintain end-to-end video quality
in wireless channels, a joint channel authentication and rate
allocation scheme was proposed in [31]. This scheme only
considers the authentication and rate allocation issues using
an unequal error protection. However, the proposed scheme
demands nodes with constant energy source. In [32], wa-
termarking was used to authenticate real-time digital video
streaming. This technique is suitable for both H.264/AVC and
H.265/HEVC video standards, however, it produces quality
degradation in received videos. Joint ngerprinting, decryption
and authentication were used together to protect the video
streams transmitted in a multi-casting wireless environment
[33]. A robust watermarking framework was proposed to
authenticate the HEVC encoded video streams in unsecured
wireless networks [34]. Data encryption was applied on H.264
encoded videos using codeword substitution to protect and
authenticate the sensitive data from network intruders [35].
The techniques presented in [33][35] work well without
disturbing the coding structure of encoded video stream.
However, they demand heavy computations which make them
infeasible for WSNs.
 Likewise security in MWSNs, the EC is also a new research
eld in MWSNs. So far researchers have focused on proper
scheduling for video transmission in MWSNs. However, such
schedulings cannot eliminate burst losses in unreliable wireless
channels [36]. The error concealment techniques are very
popular in multimedia streaming to estimate the lost and/or
missing information. However, they do not provide a robust
and defensive solution for network impairments [37]. Lost
frames of multiview videos, transmitted over MWSNs were
concealed using motion vector extrapolation [38]. Although
it produces a better objective quality, it is suitable only for
multiview videos. The impact of network impairment on Qual-
ity of Experience in video streaming over wireless network
was analyzed in [39]. This study reports a deep subjective
evaluation and uses frame copy algorithm to perform EC. An
efcient and fast EC technique was proposed in [40] to support
real-time video streaming over heterogeneous transmission
networks. The technique produces better objective quality with
less computational time.
III. Q UALITY-D RIVEN DATA AUTHENTICATION AND
C OMMUNICATION
In an encoded video bitstream, there are always N en-
coded frames. Therefore, it becomes impractical to design and
evaluate a data authentication scheme for a complete video
bitstream in an energy-constrained MWSN. Furthermore, the
designing of an authentication scheme heavily depends on
the encoding mode of a video codec. The authentication
provisioning becomes more difcult and resource consuming
if the videos are encoded in intra mode. Current video stan-
dard, i.e. H.265/HEVC, is used to generate encoded video
bitstreams [41]. Therefore, videos are encoded in inter and
unsliced modes using H.265 in our experiments. As in inter
encoding mode, the successor video frames are dependent on
former video frames, therefore, it becomes easy to perform
authentication between video frames using a crypto-hashing
function. The crypto-hashing is suitable for inter encoding, as
the crypto-hash tag of current video packet is linked with the
most recent previously processed video packet, which does not
increase the processing load over the resource-starving sensor
nodes [42]. The packet drop effects in received videos are
covered by default error concealment technique of H.265.
In our proposed scheme, rst we perform authentication
of multimedia wireless sensor nodes followed by a cluster-
based data transmission to a centralized base station via cluster
heads. The authentication ensures that only legitimate nodes
can participate in data transmission to the base station. After
authenticating the sensor nodes, the next step is to secure
the transmitted video data over wireless channels through
data authentication. The proposed scheme is summarized in
Algorithm 1.
A. Authentication
During authentication, the identities of multimedia sensor
nodes are validated and a secured methodology is adopted
for an optimal selection of cluster heads. After an optimal
selection, the multimedia sensor nodes form clusters with
592
their nearby CHs. The CHs then collect the data from the
multimedia sensor nodes within their clusters and forward it
to the BS.
1) Node Authentication: Initially, each multimedia wireless
sensor node i broadcasts a control packet to the Base Station
(BS). This packet contains nodes identity (IDi ), location (Li )
and residual energy (Ei ). A threshold n is set for the maximum
number of multimedia sensor nodes at any location, where
n N . The purpose of setting the threshold is to reduce
network congestion, caused by video data trafc, coming from
different multimedia nodes. After receiving control packets
from n multimedia wireless sensor nodes, coming from a
specic location, the BS performs a comparison among Ei
of all the nodes. The nodes with highest Ei are elected as
Cluster Heads (CHs). Soon after performing the election, the
BS replies them back with update packets. Those update
packets contain an acknowledgement for their authentication,
a notication for CH selection along with locations and
identities of their neighbouring multimedia sensor nodes. The
BS elects 5% of multimedia wireless sensor nodes as CHs in
an entire sensor eld in each round. Along with the update
packets, the BS also sends broadcast packets. The broadcast
packets contain the identities of elected CHs in that particular
round. After receiving the update and broadcast packets, each
authenticated and elected multimedia wireless sensor node
replies back with an acknowledgement packet to the BS.
2) Cluster Formation: Each CH advertises itself to its
nearest neighbouring sensor nodes which evaluate the re-
ceived signal strength from multiple CHs. A neighbouring
node associates itself with a CH having the strongest signal
strength to form a cluster as shown in Figure 1. The selection
of CH and formation of cluster is known as set-up phase.
The completion of set-up phase is followed by initiation
of steady-state phase during which each CH allocates Time
Division Multiple Access (TDMA) slots within its cluster
for sharing the transmission medium. This concept of slot
allocation enables sensing nodes to remain inactive for most
of their lifetime and at the same time avoids contention for
transmission over a wireless link. Each CH collects data within
its cluster in a particular round and transmits it directly to BS.
3) Cluster Head Communication: After successful authen-
tication, each CH aggregates data from its member nodes
within its cluster and transmits to the BS. During this step,
each CH performs data fusion to eliminate redundant packets
of overlapping regions of nearest neighbours. As a result,
highly accurate and precise data is delivered to the BS. More-
over, elimination of redundant data packets reduces the energy
consumption of deployed multimedia sensor nodes which
in turn reduces network congestion, in-network processing,
latency and packets drop. Although, the nodes capture data
based on threshold values, however, it is highly probable that
more than one nodes in the same geographical region may
capture the same video. As a result, the captured data has a
high level of redundancy and it is the job of CH to eliminate
such redundancy. After all, the BS in any sensor network is
interested only in brief description of events happening rather
than a huge volume of redundant data.
B. Video Communication Algorithm 1: Data Authentication and Communication
1) Video Encoding: The rst step in video streaming is Algorithm
video encoding, formally known as video compression. In our Input: CV - Captured Video
proposed scheme, we use HEVC standard to achieve that task. Output: OV - Output Video
In general, there are two encoding modes, i.e. intra and inter. /*Send coordinates and orientation to Base Station (BS)*/
In intra mode, each video frame is treated independently and for i = 1 to N do
the focus is on quality rather than compression. In contrary, control(Ei ,Li ,IDi );
the inter mode involves video frames dependency and the end
focus is on compression rather than quality. Each mode can /*Wait for authentication and CH selection*/
further be extended to sliced and unsliced versions. In sliced for i = 1 to n do
version, each video frame is further subdivided into smaller compare(Ei ,IDi );
data chunks and as a result, a sequence of data packets update(IDi );
traveling on transmission network, will contain only one video broadcast(IDi );
frame. On the other side, in unsliced mode, one data packet ack(IDi );
contains a whole video frame. end
In our proposed scheme, the multimedia wireless sensor /*Cluster formation*/
nodes, e.g. cameras, will be monitoring the objects. If an event for i = 1 to m do
is located, the cameras will start recording, i.e. Captured Video advertise(Ei ,IDi ,Li );
(CV). The CVs are encoded in inter but unsliced mode from a for j = 1 to n do
[IDi ,Li ] = receive();
pool of encoding options. We choose these parameters based
steady state(IDi ,Li );
on the following reasons.
ack(IDi ,Li );
The inter mode provides us an opportunity to utilize the end
dependency between frames to maintain an authentica- end
tion strategy. In this mode, any unrelated/modied video /*Encoding and transmission*/
frame can easily get diagnosed with the help of crypto- for i n do
hash function. In inter mode, each current video frame is wait for event();
dependent on its ancestor video frame for motion estima- if event == CV then
tion and compression purposes. Similarly, the crypto-hash bitstream = encode(CV,inter,unsliced);
tag of current video frame is dependent on its ancestor DPs = NL(bitstream);
video frame. Therefore, if one dependency fails, the other transmit(DPs ) to CH;
type works as a backup for authentication of video data end
packet. end
If any video frame gets lost/corrupted during transmis- /*Data authentication, concealing and transmission*/
sion, it can easily be noticed and estimated back with the for DPs CH do
help of a suitable error concealment technique. if crypto-hashing is True then
transmit(DPs ) to BS;
After encoding, a bitstream will be generated. The next
else
step is to convert the bitstream into Data Packets (DPs ) for discard(DPs );
transmission purpose. The DPs are generated by Network conceal(DPs );
Layer (NL) of HEVC test model, which are then transmitted transmit(DPs ) to BS;
to the CH. end
2) Data Authentication: It is impractical to build an authen- end
tication graph for all the video packets in video bitstream, as a /*Transmission of DPs to server*/
video may contain thousands of frames and it demands heavy for DPs BS do
processing which is not permissible with resource-constrained transmit(DPs ) to server;
MWSNs. In our proposed approach, the data authentication end
process does not put heavy computational load on sensor nodes /*Decoding and storing*/
to verify the incoming DP . for DPs server do
In order to maximize the probability of authentication, the bitstream = transform(DPs );
crypto-hash tag of each DP is connected with its former DP . end
In that way, a chain will be created where each former DP OV = decode(bitstream);
will be veried by its ancestor DP . Therefore, each current Store OV;
DP is formally veried by a series of DPs . Each DP contains
two crypto-hash tags except the rst and last DP in a series
of video frames. In our proposed approach, two crypto-hash
links are selected for an entire Group Of Pictures (GOP)
in an encoded video bitstream, where rst link denotes the
crypto-hash tag of even numbered video packets in a GOP

593
while the second link denotes the crypto-hash tag of odd
numbered video packets in a GOP. This distribution reduces
the computational load for authenticating a particular DP . In
general, if the transmission Channels Error Rate is denoted by
(CER), payload of video DP by PL(n1) , payload of crypto
signature by PC and payload of hash function by PH , then
the probability to verify a DP can be calculated as:
P = (1 CER)(PL(n1) +PC +PH ) (1 CER)(PL(n) +PH )
(1)
where n represents the video packet or frame number.
In a video stream, the frames always occur sequentially.
If frame K 1 is already veried, then we may assume
that the verication for frame K, which is linked with frame
K 2 can be skipped in our proposed approach. Based on
that assumption, the proposed authentication scheme becomes
light-weighted, less expensive in term of computation and
can easily be applied on any energy-constrained multimedia
wireless sensor node. However, in some applications, such
as military surveillance and health-care monitoring, every
incoming video packet needs to be veried.
3) Error Concealment Technique: The Error Concealment
(EC) techniques play an important role to maintain an accept-
able video quality as receivers side, especially if the videos
are streaming through lossy networks. The issue of packet
drops is always there in MWSNs, which makes the presence
of an EC technique as a feasible option. The choice of an
EC technique is very important in MWSNs. Usually, the main
part in any EC technique of inter domain is the estimation of
Motion Vectors (MVs) of missing video frames. The MVs are
estimated by a Motion Estimation (ME) procedure, which is
by far, the most complex and resource-consuming process in
any video standard/codec. Since, the sensor nodes have limited
energy, therefore, we cannot apply such EC techniques directly
to MWSNs.
The default EC method in any video codec is Frame Copy
(FC), which we use in our proposed scheme to ght against
packet drops issue. If frame K gets lost during transmission,
this method simply copies frame K 1 to ll up the gap
of missing frame K. This EC technique does not require any
extra processing, which means energy is not consumed on part
of each node. As a result, it is suitable for our proposed scheme
and can be applied to any real-time video transmission.
4) Video Decoding: After performing the data authentica-
tion and/or EC at CH, the next step is to broadcast the DPs
to the BS, which forwards them to the server. As soon as
the DPs arrive at the server, it performs three operations, i.e.
transform DPs to bitstream, decoding of bitsream to generate
Output Video (OV), and store the OV in internal storage.
C. Energy Evaluation
After getting a synchronization between video stream en-
coding, authentication and decoding, the video packet authen-
tication is achieved through crypto-hash links. In our approach,
the size of the GOP can be set manually, but it can be set
even by default. Furthermore, in order to make the evaluation
procedure simple, the starting video packets in either even or
594
odd numbered link must be given a higher priority, as the
forthcoming DPs authentication is linked with them. Let i
denote the crypto-hash tag of an even numbered video packet
in a GOP, i.e. {0, 2, 4, ...}, while j denote the crypto-hash tag
of an odd numbered video packet in a GOP, i.e. {1, 3, 5, ...}.
The crypto-hash tags are connected in reverse order, e.g. both
in even and odd frames sets, the crypto-hash tag of frame K
will be connected with frame K1 . The crypto-hash assignment
puts extra processing load in terms of processing extra data,
i.e. crypto-hash tags, on the processing node. Except the rst
and last video packets in a GOP, remaining video packets will
be dealing with dual tags, i.e. one tag from previous frame and
the second one from the upcoming frame. If the total energy,
consumed by an entire GOP is represented by ET , then it can
be computed by following equation:
ET = E0 + E1
k2

+ Ei + Ei2
i=2
k3
(2)

+ Ej + Ej2
j=3
+EK2 + EK1
IV. E XPERIMENTAL S ETUP
In this section, we summarize the experimental setup and
simulation results and performance comparisons. We use a
High Denition (HD) test video le, named Rush hour,
in YUV format, in our experiments. The YUV is an un-
compressed format and is recommended by video codecs.
The video is encoded through HM 16.9 [43] test model of
H.265/HEVC standard. We encode the test video in inter and
unsliced mode. For data authentication, we use Secure Hash
Algorithm-2 (SHA-2). This algorithm computes and compare
the hash value of current data packet with an expected hash
value, i.e. of previous data packet (in our case) to ensure data
integrity and validity. Our network consists of N multimedia
wireless sensor nodes in a sample area of 100 100 square
meter. To set up the parameters for network communication,
we follow the the specication sheet of LOTUS mote [44]. The
Cluster-based Hierarchical Routing (CHR) protocol is used for
dual purpose, i.e. for transmission and cluster head selection.
This protocol determines which node should be selected, not
only as a cluster head but to process and authenticate the
video packet. The last phase is the reconstruction of test
video le. During that phase, rstly, the video data packets
are transformed into a video bitstream. Secondly, the video
bitstream is decoded by same HM 16.9 test model to produce
video le in YUV format. Thirdly, the reconstructed video le
is compared with the original source le to compute the PSNR
to evaluate the received video quality.
In this section, we analyze our proposed scheme for differ-
ent network performance metrics.
A. Resilience to Attacks
Our proposed authentication scheme can efciently de-
tect various attacks. The security properties of our proposed
scheme in comparison to related schemes are shown in Table authentication scheme, the packet loss rate reaches up to 30%
I. in some rounds. Our proposed scheme only protects MWSN
TABLE I. Comparison of Security Properties
from external and internal attacks. The drop in packet is due
to the operation of nodes in a hostile environment using error-
Attacks LEACH LEACH-C Tsais scheme Our Scheme prone communication links. The drop in packets is contributed
Replay Insecure Insecure Insecure Secured much toward these error-prone links. In Fig. 3, the nodes drop
Sybil Insecure Insecure Insecure Secured
DoS Attack No No Yes Yes
Replication No No No Yes

As shown in the table, our proposed scheme is highly

efcient to detect various attacks such as replay, Sybil, Denial-
of-Service (DoS) and node replication. LEACH and LEACH-C
only provides cluster-based hierarchy for data transmission to
a centralized base station. These schemes do not provide any
protection against attacks and as such are vulnerable to all the
aforementioned attacks. Tsais scheme, on the other hand, is
capable to detect only DoS attack, however, it is not capable
to detect replay, DoS and node replication attacks.

B. Energy Consumption
The amount of energy consumed by MWSNs in presence
and absence of authentication is shown in Fig. 2. As the gure
shows, our proposed scheme has little impact on the energy
consumption of multimedia sensor nodes. These nodes con-
sume lesser amount of energy in absence of authentication, i.e.,
our proposed scheme. However, in absence of authentication,
the nodes are vulnerable to a wide range of attacks such as
replay, DoS, eavesdropping, Sybil and replication.
Fig. 3. Packet Loss Rate
packets up to certain rounds, i.e., 1.2 104 rounds. After that,
the packet loss rate is very marginal.
D. Network Throughput
The network throughput is dened as the number of packets
successfully delivered to a base station. Each cluster head
aggregates, fusion and transmits the data of member nodes to
a centralized base station. The network throughput is shown in
Fig. 4. As shown in the gure, the network throughput is much

Fig. 2. Energy Consumption in Authentication

C. Packet Loss Rate

Similar to any sensor network, MWSNs are frequently
deployed in remote and hostile environments. These environ-
ments coupled with little or no human intervention exposes
the nodes to various threats. As a result, the nodes drop a
signicant amount of packets. The error-prone communication
links further aid to packet loss.
In Fig. 3, the percentage of packets lost in presence of
our proposed scheme is shown. Despite the presence of an
Fig. 4. Network Throughput
higher up to 1.45 104 rounds. After these number of rounds,
the nodes have either little or no energy. As a result, the
network throughput also decreases. The quality of the received
videos after decoding can be estimated through some objective
metric, e.g. Peak Signal-to-Noise Ratio (PSNR). As shown in
Fig. 3, the packets loss ratio is very high. In such situations,

595
the Error Concealment (EC) techniques play an important role.
As we have adopted frame copy as an EC technique in our
proposed approach, we cannot expect a higher average PSNR.
In our experiments, 21.34 dB PSNR is recorded on average,
though it can be improved by applying more sophisticated EC
techniques.
V. C ONCLUSION
Multimedia Wireless Sensor Networks (MWSNs) has
gained popularity among the research community in recent
years. These networks have the potential to provide various
types of multimedia data to an end-user and/or a base station.
These networks have found their applications in various do-
mains such as military surveillance, healthcare monitoring of
industrial products, intrusion detection, bridge suspension and
agricultural monitoring. Despite their potential for monitoring
a wide range of applications, these networks are resource-
constrained due to the presence of resource-starving sensor
nodes at the core of their architecture. Similar to any sensor
nodes, a multimedia sensor node also operates in remote and
hostile environments and as such, it is vulnerable to a wide
range of malicious attacks. The presence of resource-starving
nodes requires lightweight but energy-efcient and secured
authentication schemes. In this paper, we aim to provide
a lightweight but highly secured authentication scheme for
resource-constrained MWSNs. Our proposed scheme elects an
optimal percentage of cluster heads by using the resources
of a centralized base station. Inherently, the base station is
a resource-rich entity and has the ability to elect an optimal
percentage of CHs. The interaction between the base station
and each elected cluster head is authenticated prior to cluster
formation. The clusters are formed through setup and steady-
state processes between cluster heads and their neighbouring
multimedia sensor nodes. Our proposed scheme not only
ensures that the data is coming from legitimate nodes but also
transmits highly-critical and delay-sensitive authenticated data
to the base station.
R EFERENCES
[1] I. F. Akyildiz, T. Melodia, and K. R. Chowdhury, A survey on wireless
multimedia sensor networks, Computer networks, vol. 51, no. 4, pp.
921960, 2007.
[2] P. Rawat, K. D. Singh, H. Chaouchi, and J. M. Bonnin, Wireless sensor
networks: a survey on recent developments and potential synergies, The
Journal of supercomputing, vol. 68, no. 1, pp. 148, 2014.
[3] M. A. Jan, P. Nanda, X. He, and R. P. Liu, Pasccc: Priority-based
application-specic congestion control clustering protocol, Computer
Networks, vol. 74, pp. 92102, 2014.
[4] M. A. Jan, Energy-efcient routing and secure communication in
wireless sensor networks, Ph.D. dissertation, 2016.
[5] S. Das, D. M. Bull, and P. N. Whatmough, Error-resilient design tech-
niques for reliable and dependable computing, Device and Materials
Reliability, IEEE Transactions on, vol. 15, no. 1, pp. 2434, 2015.
[6] E. P. Kim, J. Choi, N. R. Shanbhag, and R. A. Rutenbar, Error resilient
and energy efcient mrf message-passing-based stereo matching.
[7] D. E. Boubiche, S. Boubiche, and A. Bilami, A cross-layer
watermarking-based mechanism for data aggregation integrity in het-
erogeneous wsns, Communications Letters, IEEE, vol. 19, no. 5, pp.
823826, 2015.
[8] K.-A. Shim, A survey of public-key cryptographic primitives issues in
wireless sensor networks.
596
[9] K.-A. Shim and C.-M. Park, A secure data aggregation scheme based
on appropriate cryptographic primitives in heterogeneous wireless sensor
networks, Parallel and Distributed Systems, IEEE Transactions on,
vol. 26, no. 8, pp. 21282139, 2015.
[10] J. Choi, J. Bang, L. Kim, M. Ahn, and T. Kwon, Location-based key
management strong against insider threats in wireless sensor networks.
[11] L. Harn and C.-F. Hsu, Predistribution scheme for establishing group
keys in wireless sensor networks, Sensors Journal, IEEE, vol. 15, no. 9,
pp. 51035108, 2015.
[12] A. Merlo, M. Migliardi, and L. Caviglione, A survey on energy-aware
security mechanisms, Pervasive and Mobile Computing, vol. 24, pp.
7790, 2015.
[13] C. Zhu, H. Nicanfar, V. Leung, and L. T. Yang, An authenticated
trust and reputation calculation and management system for cloud and
sensor networks integration, Information Forensics and Security, IEEE
Transactions on, vol. 10, no. 1, pp. 118131, 2015.
[14] S.-I. Chu, Y.-J. Huang, and W.-C. Lin, Authentication protocol design
and low-cost key encryption function implementation for wireless sensor
networks.
[15] K. Cho, B.-G. Lee, and D. H. Lee, Low-priced and energy-efcient
detection of replicas for wireless sensor networks, Dependable and
Secure Computing, IEEE Transactions on, vol. 11, no. 5, pp. 454466,
2014.
[16] C. Benzaid, K. Lounis, A. Al-Nemrat, N. Badache, and M. Alazab,
Fast authentication in wireless sensor networks, Future Generation
Computer Systems, vol. 55, pp. 362375, 2016.
[17] K. Grover and A. Lim, A survey of broadcast authentication schemes
for wireless networks, Ad Hoc Networks, vol. 24, pp. 288316, 2015.
[18] A. Al-Riyami, N. Zhang, and J. Keane, Impact of hash value truncation
on id anonymity in wireless sensor networks, Ad Hoc Networks, 2016.
[19] N.-E. Rikli and M. Alabdulkarim, Cross-layer-based adaptive video
transport over low bit-rate multihop wsns, Electrical and Computer
Engineering, Canadian Journal of, vol. 37, no. 4, pp. 182191, 2014.
[20] L. Cheng, J. Niu, M. Di Francesco, S. K. Das, C. Luo, and Y. Gu,
Seamless streaming data delivery in cluster-based wireless sensor
networks with mobile elements.
[21] S.-H. Ou, C.-H. Lee, V. S. Somayazulu, Y.-K. Chen, and S.-Y. Chien,
On-line multi-view video summarization for wireless video sensor
network, Selected Topics in Signal Processing, IEEE Journal of, vol. 9,
no. 1, pp. 165179, 2015.
[22] J. H. Ko, B. A. Mudassar, and S. Mukhopadhyay, An energy-efcient
wireless video sensor node for moving object surveillance, Multi-Scale
Computing Systems, IEEE Transactions on, vol. 1, no. 1, pp. 718, 2015.
[23] A. Kudeshia and A. K. Jagannatham, Optimal viterbi based total
variation sequence detection (tvsd) for robust image/video decoding in
wireless sensor networks, Signal Processing Letters, IEEE, vol. 21,
no. 6, pp. 722726, 2014.
[24] M. A. Jan, P. Nanda, and X. He, Energy evaluation model for
an improved centralized clustering hierarchical algorithm in wsn. in
WWIC. Springer, 2013, pp. 154167.
[25] M. A. Jan, P. Nanda, X. He, and R. P. Liu, Enhancing lifetime and
quality of data in cluster-based hierarchical routing protocol for wireless
sensor network, in High Performance Computing and Communications
& 2013 IEEE International Conference on Embedded and Ubiquitous
Computing (HPCC EUC), 2013 IEEE 10th International Conference
on. IEEE, 2013, pp. 14001407.
[26] M. Guerrero-Zapata, R. Zilan, J. M. Barcelo-Ordinas, K. Bicakci,
and B. Tavli, The future of security in wireless multimedia sensor
networks, Telecommunication Systems, vol. 45, no. 1, pp. 7791, 2010.
[27] A. Gabrielli, L. V. Mancini, S. Setia, and S. Jajodia, Securing topology
maintenance protocols for sensor networks, Dependable and Secure
Computing, IEEE Transactions on, vol. 8, no. 3, pp. 450465, 2011.
[28] F. Bao, I.-R. Chen, M. Chang, and J.-H. Cho, Hierarchical trust man-
agement for wireless sensor networks and its applications to trust-based
routing and intrusion detection, Network and Service Management,
IEEE Transactions on, vol. 9, no. 2, pp. 169183, 2012.
[29] K. Lin, X. Ge, X. Wang, C. Zhu, and H.-G. Ryu, Research on secure
data collection in wireless multimedia sensor networks, Computer
Communications, vol. 35, no. 15, pp. 19021909, 2012.
[30] W. Wang, C. Wang, and M. Zhao, Resource optimized ttsh-ura for
multimedia stream authentication in swallowable-capsule-based wireless
body sensor networks, Biomedical and Health Informatics, IEEE Jour-
nal of, vol. 18, no. 2, pp. 404410, 2014.
[31] X. Zhu and C. W. Chen, A joint source-channel adaptive scheme for
wireless h. 264/avc video authentication, Information Forensics and
Security, IEEE Transactions on, vol. 11, no. 1, pp. 141153, 2016.
[32] M. Fallahpour, S. Shirmohammadi, M. Semsarzadeh, and J. Zhao,
Tampering detection in compressed digital video using watermarking,
Instrumentation and Measurement, IEEE Transactions on, vol. 63, no. 5,
pp. 10571072, 2014.
[33] C.-Y. Lin, K. Muchtar, C.-H. Yeh, and C.-S. Lu, Secure multicasting
of images via joint privacy-preserving ngerprinting, decryption, and
authentication, Journal of Visual Communication and Image Represen-
tation, 2016.
[34] T. Dutta and H. P. Gupta, A robust watermarking framework for high
efciency video coding (hevc)encoded video with blind extraction
process, Journal of Visual Communication and Image Representation,
vol. 38, pp. 2944, 2016.
[35] D. Xu, R. Wang, and Y. Q. Shi, An improved scheme for data hiding
in encrypted h. 264/avc videos, Journal of Visual Communication and
Image Representation, vol. 36, pp. 229242, 2016.
[36] J. Wu, Y. Shang, B. Cheng, B. Wu, and J. Chen, Loss tolerant band-
width aggregation for multihomed video streaming over heterogeneous
wireless networks, Wireless personal communications, vol. 75, no. 2,
pp. 12651282, 2014.
[37] M. Usman, X. He, M. Xu, and K. M. Lam, Survey of error concealment
techniques: Research directions and open issues, in Picture Coding
Symposium (PCS), 2015. IEEE, 2015, pp. 233238.
[38] Y. Zhou, W. Xiang, and G. Wang, Frame loss concealment for mul-
tiview video transmission over wireless multimedia sensor networks,
Sensors Journal, IEEE, vol. 15, no. 3, pp. 18921901, 2015.
[39] J. Nightingale, Q. Wang, C. Grecos, and S. Goma, The impact of
network impairment on quality of experience (qoe) in h. 265/hevc video
streaming, Consumer Electronics, IEEE Transactions on, vol. 60, no. 2,
pp. 242250, 2014.
[40] M. Usman, X. He, K. Lam, M. Xu, S. Bokhari, and J. Chen, Frame
interpolation for cloud-based mobile video streaming, Multimedia,
IEEE Transactions on, vol. 18, no. 5, pp. 831839, 2016.
[41] G. J. Sullivan, J.-R. Ohm, W.-J. Han, and T. Wiegand, Overview of
the high efciency video coding (hevc) standard, Circuits and Systems
for Video Technology, IEEE Transactions on, vol. 22, no. 12, pp. 1649
1668, 2012.
[42] D. He, S.-C. Chan, and S. Tang, A novel and lightweight system
to secure wireless medical sensor networks, Biomedical and Health
Informatics, IEEE Journal of, vol. 18, no. 1, pp. 316326, 2014.
[43] Hm 16.6, a test model of h.265/hevc standard, https://hevc.hhi.
fraunhofer.de/svn/svn HEVCSoftware/trunk/, accessed: 2016-04-13.
[44] Lotus mote, multimedia wireless sensor node, http://www.memsic.
com/wireless-sensor-networks/, accessed: 2016-04-13.

597