ArticlePDF Available

An Optimized Hidden Node Detection Paradigm for Improving the Coverage and Network Efficiency in Wireless Multimedia Sensor Networks

Authors:

Abstract and Figures

Successful transmission of online multimedia streams in wireless multimedia sensor networks (WMSNs) is a big challenge due to their limited bandwidth and power resources. The existing WSN protocols are not completely appropriate for multimedia communication. The effectiveness of WMSNs varies, and it depends on the correct location of its sensor nodes in the field. Thus, maximizing the multimedia coverage is the most important issue in the delivery of multimedia contents. The nodes in WMSNs are either static or mobile. Thus, the node connections change continuously due to the mobility in wireless multimedia communication that causes an additional energy consumption, and synchronization loss between neighboring nodes. In this paper, we introduce an Optimized Hidden Node Detection (OHND) paradigm. The OHND consists of three phases: hidden node detection, message exchange, and location detection. These three phases aim to maximize the multimedia node coverage, and improve energy efficiency, hidden node detection capacity, and packet delivery ratio. OHND helps multimedia sensor nodes to compute the directional coverage. Furthermore, an OHND is used to maintain a continuous node– continuous neighbor discovery process in order to handle the mobility of the nodes. We implement our proposed algorithms by using a network simulator (NS2). The simulation results demonstrate that nodes are capable of maintaining direct coverage and detecting hidden nodes in order to maximize coverage and multimedia node mobility. To evaluate the performance of our proposed algorithms, we compared our results with other known approaches.
Content may be subject to copyright.
sensors
Article
An Optimized Hidden Node Detection Paradigm for
Improving the Coverage and Network Efficiency in
Wireless Multimedia Sensor Networks
Adwan Alanazi * and Khaled Elleithy
Computer Science and Engineering Department, University of Bridgeport, 126 Park Ave, Bridgeport,
CT 06604, USA; elleithy@bridgeport.edu
*Correspondence: aalanazi@my.bridgeport.edu; Tel.: +1-816-703-9142; Fax: +1-203-576-4766
Academic Editor: Jaime Lloret Mauri
Received: 10 June 2016; Accepted: 31 August 2016; Published: 7 September 2016
Abstract:
Successful transmission of online multimedia streams in wireless multimedia sensor
networks (WMSNs) is a big challenge due to their limited bandwidth and power resources.
The existing WSN protocols are not completely appropriate for multimedia communication.
The effectiveness of WMSNs varies, and it depends on the correct location of its sensor nodes
in the field. Thus, maximizing the multimedia coverage is the most important issue in the delivery of
multimedia contents. The nodes in WMSNs are either static or mobile. Thus, the node connections
change continuously due to the mobility in wireless multimedia communication that causes an
additional energy consumption, and synchronization loss between neighboring nodes. In this
paper, we introduce an Optimized Hidden Node Detection (OHND) paradigm. The OHND consists
of three phases: hidden node detection, message exchange, and location detection. These three
phases aim to maximize the multimedia node coverage, and improve energy efficiency, hidden node
detection capacity, and packet delivery ratio. OHND helps multimedia sensor nodes to compute the
directional coverage. Furthermore, an OHND is used to maintain a continuous node– continuous
neighbor discovery process in order to handle the mobility of the nodes. We implement our proposed
algorithms by using a network simulator (NS2). The simulation results demonstrate that nodes are
capable of maintaining direct coverage and detecting hidden nodes in order to maximize coverage and
multimedia node mobility. To evaluate the performance of our proposed algorithms, we compared
our results with other known approaches.
Keywords:
wireless multimedia sensor networks (WMSNs); optimized occlusion-free viewpoint and
multimedia coverage; mobility; coverage; energy efficient hidden node detection
1. Introduction
WMSNs are capable of capturing audio-video information by using low-cost cameras embedded
with sensor nodes. The multimedia sensors provide substantial information related to a particular
area of interest [
1
6
]. However, multimedia applications experience problems due to online media
transmission challenges [
7
]. Several sources of energy waste include idle listening, overhearing, packet
loss due to collisions, and packet overhead in the multimedia sense. One of the major sources of energy
waste are the packet collisions that happen when two nodes try to transmit the packets simultaneously.
As a result, this causes a partial or complete packet loss at the recipient node. The lost packets need
to be discarded or retransmitted, which could be the source of the excess energy consumption waste
and Quality of Service (QoS) degradation. To enable the on-demand multimedia services, we need
to focus on multimedia-supported algorithms in WMSNs to determine the hidden node problems
and compute the directional coverage. The existing IEEE 802.15.4 standard is based on the blind
Sensors 2016,16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Sensors 2016,16, 1438 2 of 19
back off carrier sense multiple approach with collision avoidance (CSMA/CA), where the nodes
check the channel before sending the data frame. If, the channel is free, then the node initiates the
transmission process; otherwise, it reattempts after a certain time. However, this approach is only
suitable when the nodes hear each other, and that could be a rare case in WMSNs [
6
]. In most cases,
the network coverage is much larger than the single node’s coverage area. In general, a well-defined
coverage area does not support WMSNs because the propagation characteristics are uncertain and
dynamic. In this situation, the node is unable to determine the receiver side. The results could be the
probability of hidden node collisions. The restrictions of the multimedia sensing proficiencies relate to
the location coverage and hidden node problem. Once a better location coverage and hidden node
solution for multimedia sensors are discovered, the results will help to improve the capabilities of
WMSNs applications. Additionally, WMSNs restrictions are caused by tall buildings, mountains, and
trees. Hence, directional coverage of multimedia sensors could be completed once they are deployed in
an area of interest. However, proper directional location for multimedia sensors requires correct field
information prior to deployment of sensors. It is also likely that multimedia sensors might change their
location due to mobility over time. This problem can be resolved by dynamic updates of the locations
through location information exchanges [8]. However, multimedia applications have limitations that
will affect the successful media transmission in the sensor networks. Node connectivity is subject to
change because of wireless commotions [
9
,
10
]. When a sensor is responsive to its immediate neighbors,
it must uninterruptedly upload information about its surroundings.
The connectivity is a severe problem subject to the mobility change when the network has been
set up. The sensors nodes try to identify new neighbors to address mobility problems, but a hidden
node problem is a hurdle. Initial neighbor node discovery is typically performed when the sensor node
has no proof about the configuration of its immediate neighbors. In this situation, the sensor node is
unable to communicate with either the base station or the sink station. Thus, immediate neighboring
nodes should be detected as soon as possible to set a path to the base station and contribute to the
operation of the network [11,12]. Hence, in this state, more wide-ranging energy use is justified.
In order to handle the hidden node and coverage problem, our approach contributes the
Obstacle-driven Negative Effect Strategy (ONES) method that handles the negative effect of the
obstacles. The proposed method is designed for those scenarios where the number of the relays are less
than those of relays required for building steady links. In addition, it is particularly suitable for those
multimedia sensor networks that suffer due to several disconnected subdivisions of the network that
are experiencing the issue of obstacles among the subdivisions. The method is validated by applying
several assumptions and definitions. This helps reduce energy consumption and maintaining the QoS.
Furthermore, our paradigm contributes the Optimized Hidden Node Paradigm that involves the
hidden node detection, message exchange phase, and location detection. Our paradigm is different
from existing hidden node approaches, as we focus on the multiple discoveries of nodes rather than
a single discovery of node. The approach is particularly designed for a distributed network as most
of the existing approaches follow the central-based network in the node discovery process, which is
also expensive for location-updating. We focus on improving the QoS and extending the network life.
Thus, a network is divided into different subdivisions and is controlled by a coordinator node, as the
subdivision process helps multimedia sensor nodes cover the entire area efficiently. The network life
extension is justified with the proper selection of a controlling node using metrics such as residual
energy, data forwarding capacity of the node, distance of the node from the base station, and memory
allocation. These metrics are assigned the specific weightage that provides enough chance for each
node to be a coordinator and balance the network. Handling the problem of overlapped subdivisions
in the network particularly when a new joining node attempts to be a part of either subdivision is a
critical issue that has also been handled by a priority-based synchronization.
The random wake-up procedure is applied to reduce the option of repeating collision amongst
the nodes in the same subdivision. The beauty of our random wake-up process that it provides an
opportunity for each node to coordinate with its neighborhood nodes to avoid the collision and initiate
Sensors 2016,16, 1438 3 of 19
a faster discovery process for the new joining hidden node. In addition, each node applies an active
discovery process to detect its coordinator node, whereas in the existing approaches, the coordinator
(head node) is responsible for detecting its nodes in its subdivision, which puts an extra burden on
the coordinator node. However, our approach handles this issue by assigning responsibility to each
node to detect its coordinator node in each subdivision. Unlike existing approaches, our paper also
contributes the novel location detection procedure that helps detect the maximum view, node boundary
and viewpoint of multimedia sensor node; existing approaches either apply a node boundary or node
view to detect the location. The remainder of the paper is organized as follows: Section 2presents the
salient features of the most related work. Section 3presents an obstacle-driven negative effect strategy
method. Section 4presents optimized occlusion-free viewpoint and an energy efficient hidden node
detection algorithms. Section 5discusses the simulation setup and experimental results. Finally, the
conclusions of the entire approach are given in Section 6.
2. Related Work
In this section, related WMSN approaches are discussed. Previous works discuss maximizing
the coverage area and detecting the hidden nodes in the fields of wireless sensor networking, ad-hoc
networks, and robotics. However, little research has been done in wireless multimedia sensor
networking. Some addresses an omnidirectional coverage problem in wireless sensor networks [
13
,
14
],
but it is not suitable for a bidirectional and an occlusion-free viewpoint. Numerous applications require
bi-directional coverage, but existing coverage models are only suitable for traditional wireless sensor
networks (WSNs), and do not support WMSNs. An initial study regarding the coverage of multimedia
sensors is described in [
15
]. In this work, the authors proposed a routing protocol with the field of view
camera placed on the floor. The video sensors are used by oceanographers to monitor the shallows.
Furthermore, triangular view segments are used for calculating the coverage of wireless multimedia
sensor networks in [16].
The neighbor discovery node process is proposed in [
17
] to regulate the new nodes from the base
station. This approach only focuses on finding hidden nodes and not on energy consumption. The base
station starts the node discovery process by broadcasting a HELLO message, and the node initiates
the registration process after receiving the HELLO message. The node can switch channels to find the
best HELLO message, which helps to locate the hidden nodes. In order to reduce the neighbor node
discovery time, [
18
] introduced the HELLO message-based approach to identify the hidden nodes,
but energy efficiency was not considered. In [
19
], an energy efficient node discovery algorithm was
introduced based on temporal patterns of coincidences in order to reach other nodes. However, all
these approaches address the wireless sensor network issues rather than multimedia wireless sensor
network issues. In [
11
], the authors proposed the use of Voronoi diagrams and Delaunay triangulation
to detect the best and worst coverage area in WMSNs. Another approach based on deploying an
additional sensor node is introduced in [
20
] to maximize the coverage area. In this proposed approach,
a two stage process was used for detection of phase coverage boundaries and obstacles by applying
the formula 2×R(where R: sensing radius of sensors).
In [
8
], virtual centripetal force-based coverage-enhancing algorithm was proposed for WMSNs.
In this work, the grid theory, centripetal force model with essential mass and overlapping idea of
the sensors are discussed. This algorithm shuts off any idle multimedia sensory to maximize the
network coverage. Furthermore, the network is extended by redistributing sensors and by applying
centripetal force based on the circular motion. The authors of [
21
] introduced a secure neighbor
discovery process and attempted to protect the wireless sensor networks from different types of
threats. The approach comprises a scalable key-distribution protocol that protects the neighbor nodes
in the presence of malicious nodes. This aims to improve the secure neighbor discovery to guard
the attacks of hidden nodes. The static network is deployed for securing the one-hop neighbor
discovery process. However, the work does not address energy efficiency. The Line of Sight Method
(LOSM) [
22
] is introduced for the wireless personal area network based on visible light communication
Sensors 2016,16, 1438 4 of 19
technology. It handles the issue of the hidden nodes in IEEE 802.15.7 and in particular focuses on QoS
parameters such as end-to-end delay, message loss, and goodput. The idle-pattern-based approach
is introduced and in which the idle patterns are sent by the network coordinator to perpetuate the
communication with other network sensor nodes. However, the work did not focus on energy efficiency
and multimedia contents. The Efficient Beam Scanning Algorithm for Hidden Node (EBSAHN) [
23
]
is proposed for the wireless sensor networks. This approach is based on an efficient Intra cluster
grouping scheme (IC-GS) that helps add a new node into the wireless sensor network. Furthermore, it
avoids the hidden node collision avoidance. The Hidden Node Problem (HNP) [
24
] is introduced in
the wireless sensor network. This approach aims to generate the hidden node relationship for all nodes
and allot the hidden nodes into different clusters. In this approach, time for super-frame is divided
into sub-period, and size of the sub-period depends on the number of the hidden nodes into the cluster.
The approach is primary based on improving the QoS. The Clustering-Based Mechanism for Detecting
the Hidden Nodes (CMDHN) [
25
] is proposed for resolving the hidden node problem in the wireless
sensor networks in order to improve the network performance. Furthermore, delay, throughput, and
energy consumption are major parameters focuses. All existing approaches attempted to determine the
hidden nodes and coverage problems but did not properly focus on energy efficiency, scalability, QoS,
multimedia-content support delivery, and accuracy in multimedia sensor nodes. The characteristics
and limitations of existing approaches are highlighted in Table 1.
3. Obstacle-Driven Negative Effect Strategy Method
Here, we consider the wireless sensor network that is divided into several subdivisions and
suffered because of disconnected subdivisions; it also experiences the issue of the obstacles among the
subdivisions. In the network, each subdivision is controlled by the coordinator node. The coordinator
node has a communication range
Rc
that is the maximum Euclidean distance reached by the node’s
radio. Our network is based on the following assumptions.
Sensors 2016, 16, 1438 4 of 19
technology. It handles the issue of the hidden nodes in IEEE 802.15.7 and in particular focuses on QoS
parameters such as end-to-end delay, message loss, and goodput. The idle-pattern-based approach
is introduced and in which the idle patterns are sent by the network coordinator to perpetuate the
communication with other network sensor nodes. However, the work did not focus on energy
efficiency and multimedia contents. The Efficient Beam Scanning Algorithm for Hidden Node
(EBSAHN) [23] is proposed for the wireless sensor networks. This approach is based on an efficient
Intra cluster grouping scheme (IC-GS) that helps add a new node into the wireless sensor network.
Furthermore, it avoids the hidden node collision avoidance. The Hidden Node Problem (HNP) [24]
is introduced in the wireless sensor network. This approach aims to generate the hidden node
relationship for all nodes and allot the hidden nodes into different clusters. In this approach, time for
super-frame is divided into sub-period, and size of the sub-period depends on the number of the
hidden nodes into the cluster. The approach is primary based on improving the QoS. The Clustering-
Based Mechanism for Detecting the Hidden Nodes (CMDHN) [25] is proposed for resolving the
hidden node problem in the wireless sensor networks in order to improve the network performance.
Furthermore, delay, throughput, and energy consumption are major parameters focuses. All existing
approaches attempted to determine the hidden nodes and coverage problems but did not properly
focus on energy efficiency, scalability, QoS, multimedia-content support delivery, and accuracy in
multimedia sensor nodes. The characteristics and limitations of existing approaches are highlighted
in Table 1.
3. Obstacle-Driven Negative Effect Strategy Method
Here, we consider the wireless sensor network that is divided into several subdivisions and
suffered because of disconnected subdivisions; it also experiences the issue of the obstacles among
the subdivisions. In the network, each subdivision is controlled by the coordinator node. The
coordinator node has a communication range ′ that is the maximum Euclidean distance reached
by the node’s radio. Our network is based on the following assumptions.
Figure 1. Showing the On-Demand Delivery process and obstacles.
Figure 1. Showing the On-Demand Delivery process and obstacles.
Sensors 2016,16, 1438 5 of 19
Table 1. The characteristics and limitations of the existing approaches.
Existing Protocols Bio-Directional
Coverage
Omni-Directional
Coverage
Single-Directional
Coverage Energy-Efficient Hidden Node
Detection Scalability Multimedia Content-
Support Delivery QoS
Energy-efficient Probabilistic Area Coverage
(EPAC) [12]X X
Energy-efficient Node Scheduling Protocol for
Target Coverage (ENSPTC) [13]X X
Coverage Problem in Video-Based (CPV) [14] X X
Optimal Worst-Case Coverage of Directional
Field-of-View (OWCDF) [15]X X
Worst and Best-Case Coverage (WBC) [16] X X X
Delaunay Triangulation-Based Method
(DTM) [17]X X
Virtual Centripetal Force-based
Coverage-Enhancing Algorithm (VCFEA) [18]X
Energy-Efficient Link Assessment (ELA) [19] X X
Secure Neighbor Discovery (SND) [20] X
Line of Sight Method (LOSM) [22] X X
Efficient Beam Scanning Algorithm for Hidden
Node (EBSAHN) [23]X X X
Hidden Node Problem (HNP) [24] X X X
Clustering-Based Mechanism for Detecting the
Hidden Nodes (CMDHN) [25]X X X
Sensors 2016,16, 1438 6 of 19
Assumption 1:
All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted in
Figure 1.
Assumption 2:
There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3:
Let
σ
be the count of subdivision
Sdi
’. The number of available mobile relay nodes responsible
for dealing with on-demand multimedia service ‘Mrn should satisfy the following condition:
Mrn >σ&& Mrn <Trn
where Trn: Total relay nodes.
Definition 1:
Given an undirected graph
GU
with subdivisions (segments) ‘SD’ and edges
E
can be written
as GU=(SD,E). Thus, the set of subdivisions are SD ={Sd1,Sd2,Sd3, ..., Sdn}and edges.
E={(Sdi,Sdj)|Sdi,SdjSD
. Let obstacle-driven negative effect strategy
Ψ
be
Ψ
= (SD,
EΨ
),
where
EΨ
denotes the set of edges of (ONES). Let
Ed=(EEΨ)
,
Sdi,SdjEΨ
and
Sensors 2016, 16, 1438 6 of 19
Sensors 2016, 16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Assumption 1: All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted
in Figure 1.
Assumption 2: There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3: Let ′′ be the count of subdivision ′. The number of available mobile relay nodes
responsible for dealing with on-demand multimedia service ′ should satisfy the following condition:
 >&&
 <

where : Total relay nodes.
Definition 1: Given an undirected graph ′ with subdivisions (segments)′′ and edges ′′ can be
written as =(,). Thus, the set of subdivisions are  = {,,,...,} and edges.
=(
,⎮,∈}. Let obstacle-driven negative effect strategy Ψ’ be Ψ=(SD,), where
′ denotes the set of edges of (ONES). Let =(−),∀,∈ and ℾƸ(,)
˅(,)<(
,), where (,): Euclidean distance of subdivision , :
deleted edges.
Definition 2: The number of relay sensor nodes between subdivisions  can be denoted by
Ƹ(,)∈
,(,). Thus, (,) can be obtained by:
(,)=
(,)
−1 (1)
Definition 3: Let  represents the total number of the relay needed to generate the steady network
topology. Thus,  can be obtained by:
 =
(,)
(,)∈
(2)
Hence, obstacle-driven negative effect strategy can be simplified as  = {,,,...,} that is
the set of subdivisions and set of obstacles are ={
,,,...,}. where cannot be overlapped with
subdivision . Thus, ′ denotes the total relay nodes to create the steady network topology. Let  =
{,,,...,} be the available relays for on-demand multimedia contents, where  satisfies
following condition:
 >
⋀
 <

4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless
multimedia sensor networks because the sensor nodes are deployed in a disseminated manner within
a realistic environment. On the other hand, centralized location deployment is not appropriate for
WMSNs because these networks encompass a large number of multimedia nodes. Furthermore,
updating the location is expensive within a centralized approach when compared to a distributed
approach. Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we
Sensors 2016, 16, 1438 6 of 19
Sensors 2016, 16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Assumption 1: All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted
in Figure 1.
Assumption 2: There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3: Let ′′ be the count of subdivision ′. The number of available mobile relay nodes
responsible for dealing with on-demand multimedia service ′ should satisfy the following condition:
 >&&
 <

where : Total relay nodes.
Definition 1: Given an undirected graph ′ with subdivisions (segments)′′ and edges ′′ can be
written as =(,). Thus, the set of subdivisions are  = {,,,...,} and edges.
=(
,⎮,∈}. Let obstacle-driven negative effect strategy Ψ’ be Ψ=(SD,), where
′ denotes the set of edges of (ONES). Let =(−),∀,∈ and ℾƸ(,)
˅(,)<(
,), where (,): Euclidean distance of subdivision , :
deleted edges.
Definition 2: The number of relay sensor nodes between subdivisions  can be denoted by
Ƹ(,)∈
,(,). Thus, (,) can be obtained by:
(,)=
(,)
−1 (1)
Definition 3: Let  represents the total number of the relay needed to generate the steady network
topology. Thus,  can be obtained by:
 =
(,)
(,)∈
(2)
Hence, obstacle-driven negative effect strategy can be simplified as  = {,,,...,} that is
the set of subdivisions and set of obstacles are ={
,,,...,}. where cannot be overlapped with
subdivision . Thus, ′ denotes the total relay nodes to create the steady network topology. Let  =
{,,,...,} be the available relays for on-demand multimedia contents, where  satisfies
following condition:
 >
⋀
 <

4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless
multimedia sensor networks because the sensor nodes are deployed in a disseminated manner within
a realistic environment. On the other hand, centralized location deployment is not appropriate for
WMSNs because these networks encompass a large number of multimedia nodes. Furthermore,
updating the location is expensive within a centralized approach when compared to a distributed
approach. Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we
(Sdp
,
Sdq)
Ed
Sensors 2016, 16, 1438 6 of 19
Sensors 2016, 16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Assumption 1: All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted
in Figure 1.
Assumption 2: There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3: Let ′′ be the count of subdivision ′. The number of available mobile relay nodes
responsible for dealing with on-demand multimedia service ′ should satisfy the following condition:
 >&&
 <

where : Total relay nodes.
Definition 1: Given an undirected graph ′ with subdivisions (segments)′′ and edges ′′ can be
written as =(,). Thus, the set of subdivisions are  = {,,,...,} and edges.
=(
,⎮,∈}. Let obstacle-driven negative effect strategy Ψ’ be Ψ=(SD,), where
′ denotes the set of edges of (ONES). Let =(−),∀,∈ and ℾƸ(,)
˅(,)<(
,), where (,): Euclidean distance of subdivision , :
deleted edges.
Definition 2: The number of relay sensor nodes between subdivisions  can be denoted by
ℾƸ(,)∈
,(,). Thus, (,) can be obtained by:
(,)=
(,)
−1 (1)
Definition 3: Let  represents the total number of the relay needed to generate the steady network
topology. Thus,  can be obtained by:
 =
(,)
(,)∈
(2)
Hence, obstacle-driven negative effect strategy can be simplified as  = {,,,...,} that is
the set of subdivisions and set of obstacles are ={
,,,...,}. where cannot be overlapped with
subdivision . Thus, ′ denotes the total relay nodes to create the steady network topology. Let  =
{,,,...,} be the available relays for on-demand multimedia contents, where  satisfies
following condition:
 >
⋀
 <

4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless
multimedia sensor networks because the sensor nodes are deployed in a disseminated manner within
a realistic environment. On the other hand, centralized location deployment is not appropriate for
WMSNs because these networks encompass a large number of multimedia nodes. Furthermore,
updating the location is expensive within a centralized approach when compared to a distributed
approach. Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we
r(Sdi
,
Sdj)<r(Sdp
,
Sdq)
, where
r(Sdp
,
Sdq)
: Euclidean distance of subdivision
Sdiand Sdj
,
Ed
:
deleted edges.
Definition 2:
The number of relay sensor nodes between subdivisions
Sdiand Sdj
can be denoted by
Sensors 2016, 16, 1438 6 of 19
Sensors 2016, 16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Assumption 1: All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted
in Figure 1.
Assumption 2: There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3: Let ′′ be the count of subdivision ′. The number of available mobile relay nodes
responsible for dealing with on-demand multimedia service ′ should satisfy the following condition:
 >&&
 <

where : Total relay nodes.
Definition 1: Given an undirected graph ′ with subdivisions (segments)′′ and edges ′′ can be
written as =(,). Thus, the set of subdivisions are  = {,,,...,} and edges.
=(
,⎮,∈}. Let obstacle-driven negative effect strategy Ψ’ be Ψ=(SD,), where
′ denotes the set of edges of (ONES). Let =(−),∀,∈ and ℾƸ(,)
˅(,)<(
,), where (,): Euclidean distance of subdivision , :
deleted edges.
Definition 2: The number of relay sensor nodes between subdivisions  can be denoted by
Ƹ(,)∈
,(,). Thus, (,) can be obtained by:
(,)=
(,)
−1 (1)
Definition 3: Let  represents the total number of the relay needed to generate the steady network
topology. Thus,  can be obtained by:
 =
(,)
(,)∈
(2)
Hence, obstacle-driven negative effect strategy can be simplified as  = {,,,...,} that is
the set of subdivisions and set of obstacles are ={
,,,...,}. where cannot be overlapped with
subdivision . Thus, ′ denotes the total relay nodes to create the steady network topology. Let  =
{,,,...,} be the available relays for on-demand multimedia contents, where  satisfies
following condition:
 >
⋀
 <

4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless
multimedia sensor networks because the sensor nodes are deployed in a disseminated manner within
a realistic environment. On the other hand, centralized location deployment is not appropriate for
WMSNs because these networks encompass a large number of multimedia nodes. Furthermore,
updating the location is expensive within a centralized approach when compared to a distributed
approach. Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we
Sensors 2016, 16, 1438 6 of 19
Sensors 2016, 16, 1438; doi:10.3390/s16091438 www.mdpi.com/journal/sensors
Assumption 1: All the relay nodes are mobile sensor and responsible for on-demand delivery as depicted
in Figure 1.
Assumption 2: There is at least one edge that interconnects the obstacles. Each obstacle is not overlapped
with subdivisions.
Assumption 3: Let ′′ be the count of subdivision ′. The number of available mobile relay nodes
responsible for dealing with on-demand multimedia service ′ should satisfy the following condition:
 >&&
 <

where : Total relay nodes.
Definition 1: Given an undirected graph ′ with subdivisions (segments)′′ and edges ′′ can be
written as =(,). Thus, the set of subdivisions are  = {,,,...,} and edges.
=(
,⎮,∈}. Let obstacle-driven negative effect strategy Ψ’ be Ψ=(SD,), where
′ denotes the set of edges of (ONES). Let =(−),∀,∈ and ℾƸ(,)
˅(,)<(
,), where (,): Euclidean distance of subdivision , :
deleted edges.
Definition 2: The number of relay sensor nodes between subdivisions  can be denoted by
Ƹ(,)∈
,(,). Thus, (,) can be obtained by:
(,)=
(,)
−1 (1)
Definition 3: Let  represents the total number of the relay needed to generate the steady network
topology. Thus,  can be obtained by:
 =
(,)
(,)∈
(2)
Hence, obstacle-driven negative effect strategy can be simplified as  = {,,,...,} that is
the set of subdivisions and set of obstacles are ={
,,,...,}. where cannot be overlapped with
subdivision . Thus, ′ denotes the total relay nodes to create the steady network topology. Let  =
{,,,...,} be the available relays for on-demand multimedia contents, where  satisfies
following condition:
 >
⋀
 <

4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless
multimedia sensor networks because the sensor nodes are deployed in a disseminated manner within
a realistic environment. On the other hand, centralized location deployment is not appropriate for
WMSNs because these networks encompass a large number of multimedia nodes. Furthermore,
updating the location is expensive within a centralized approach when compared to a distributed
approach. Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we
(Sdp,Sdq)EΨ,Mrn(Sdi,Sdj).Thus, Mrn (Sdi,Sdj)can be obtained by:
Mrn (Sdi,Sdj) = r(Sdp,Sdq)
Rc1 (1)
Definition 3:
Let
Trn
represents the total number of the relay needed to generate the steady network topology.
Thus, Trn can be obtained by:
Trn =
N
(Sdi,Sdj)EΨ
Mrn (Sdi,Sdj)(2)
Hence, obstacle-driven negative effect strategy can be simplified as
SD ={Sd1,Sd2,Sd3, ..., Sdn}
that is
the set of subdivisions and set of obstacles are
O={O1,O2,O3, ..., On}
. where
Oj
cannot be overlapped
with subdivision
Sdi
. Thus,
Trn
denotes the total relay nodes to create the steady network topology.
Let
Mrn ={Ra1,Ra2,Ra3, ..., Ran}
be the available relays for on-demand multimedia contents, where
Mrn
satisfies following condition:
Mrn >EdMrn <Trn
4. Optimized Hidden Node Detection Paradigm
An optimized hidden node detection paradigm is introduced for distributed wireless multimedia
sensor networks because the sensor nodes are deployed in a disseminated manner within a realistic
environment. On the other hand, centralized location deployment is not appropriate for WMSNs
because these networks encompass a large number of multimedia nodes. Furthermore, updating
the location is expensive within a centralized approach when compared to a distributed approach.
Our approach consists of three phases:
Hidden Node Detection
Message Exchange Phase
Location Detection
4.1. Hidden Node Detection
Detecting a hidden node in WMSNs is the critical problem that affects the network performance.
Hence, the efficient neighbor discovery process helps in hidden node detection. In this phase, we focus
Sensors 2016,16, 1438 7 of 19
on a continuous neighbor discovery process to determine the hidden nodes. Each sensor node uses
a coordination-driven approach, and we chose the 1-hop multiple neighborhood discovery process
rather than the particular node-discovery in the network that helps detect all hidden nodes at the1-hop
neighborhood. As a result, the network consumes a minimum amount of energy and has a collision-free
process. In this approach, the nodes share the schedule as discussed in [
26
]. The network is divided
into different subdivisions, and each subdivision is controlled by a coordination node. However, our
approach selects the coordination node based on the residual energy, data forwarding capacity of the
node, distance of the node from the base station, and memory allocation. Each node continues to play
a role as the coordinator until it possesses the higher weightage as compared with other nodes of the
subdivision. The higher weightage is calculated by assigning different values as residual energy is
assigned 33% weightage, the nearest distance of the node from the base station gets 25%, the data
forwarding capability gets 15%, and the memory allocation resource gets 27% weightage. We tried to
use different combinations of the weightage for each metrics, but we obtained the optimal results with
our chosen weightage numbers. Thus, each coordination node is responsible for detecting the new
hidden node when joining the network. Each new joining node is required to send a synchronization
message to the coordination node, each subdivision has only one coordination node, which means there
is less possibility of collision to handle synchronization message. In case a new joining node sends the
coordination requests to two different subdivisions, if a node is located close to the subdivisions that
are overlapped, the coordinator node that receives the first synchronization message that entertains
the node. On other hand, the coordination node that receives a later synchronization message that
responds to the new joining node, but that node has already become part of the other subdivision.
As a result, the possibility exists for energy waste of the coordination node. However, the wasted
energy of the coordination node is negligible. The coordination node replies to the new node and
sends a message to all the nodes in the same subdivision to store records about the new joining node.
After getting a message from the coordination node, all the nodes send a message to the new node.
When it receives these messages, the new node sends a message to the coordination node about the
nodes who replied to the previous message. The goal is to confirm the new node request and inform
all the nodes in the same subdivision about the new member and the new node has the information
about the other neighboring nodes. The synchronization message is dispersed over the entire links of
the network to link with the coordination node. This is the way that the coordination node determines
a new joining node is detected. When the coordination node has information about a new joining
node that broadcasts within its subdivision nodes, the coordinating node will send out a message to
all the nodes in the same subdivision. The synchronization process of a new joining node with the
coordination node process is depicted in Figure 2.
The hidden node detection process applies a random wake-up procedure to reduce the option of
repeating collisions amongst the nodes in the same subdivision. In this phase, each node coordinates
with its neighborhood nodes during the wake-up period to avoid collisions and make a faster discovery
process of any new joining hidden node. The wake up time period is very small, and the time of
forwarding the HELLO message is even smaller. In this case, there is a possibility that two nodes can be
active at same time and initiate the neighbor node discovery process. Therefore, we use a scheduling
method to control the wake up process of two nodes at the same time. During the scheduling, the
nodes are required to be synchronized with each other and report to the coordination node. During the
scheduling, each receiving node chooses time slots and obtains the data during those time slots.
The time slot process is performed without contradicting the schedule of the other node. This is the
reason that the neighbor nodes are subdivided into different subdivisions, where each node chooses its
slot assigned to that subdivision. Each sensor node decides randomly when to initiate the transmission
of a HELLO message. If its message does not strike with another HELLO message, the node is referred
to as a discovered node. We can also determine residual energy and the load of each node after the
node discovery process occurs.
Sensors 2016,16, 1438 8 of 19
Sensors 2016, 16, 1438 8 of 19
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N × N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ’C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (: coordinator node-1; : coordinator node-2; : newly joined-node; :
beacon message for joining the subdivision; : node synchronization process; :
subdivision of the network; : listening)
2. Input: ()
3. Output: ()
4. Set attempts for  // New node intends to join the subdivision of the network
5. // New node sends beacon message for joining the network
6.  &&  // Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. ℓ∈
then // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
8. 
∈ // coordinator node-1 allows ‘ɀ′ the new node to be part of subdivision
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N
×
N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ‘C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1
: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (Cn1: coordinator node-1; Cn2: coordinator node-2; Nj: newly joined-node; Bm:
beacon message for joining the subdivision; Sn: node synchronization process; Nsdi:
subdivision of the network; `: listening)
2. Input: (Bm)
3. Output: (Sn)
4. Set Njattempts for Nsd // New node intends to join the subdivision of the network
5. NjBm// New node sends beacon message for joining the network
6. Bm`Cn1&& Cn2// Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. If Cn1`BmNjthen // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
Sensors 2016,16, 1438 9 of 19
Algorithm 1:Cont.
8. Cn1
Sensors 2016, 16, 1438 8 of 19
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N × N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ’C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (: coordinator node-1; : coordinator node-2; : newly joined-node; :
beacon message for joining the subdivision; : node synchronization process; :
subdivision of the network; : listening)
2. Input: ()
3. Output: ()
4. Set attempts for  // New node intends to join the subdivision of the network
5. // New node sends beacon message for joining the network
6.  &&  // Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. ℓ∈
then // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
8. 
ɀ
∈ // coordinator node-1 allows ‘ɀ′ the new node to be part of subdivision
NjNsd // coordinator node-1 allows
Sensors 2016, 16, 1438 8 of 19
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N × N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ’C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (: coordinator node-1; : coordinator node-2; : newly joined-node; :
beacon message for joining the subdivision; : node synchronization process; :
subdivision of the network; : listening)
2. Input: ()
3. Output: ()
4. Set attempts for  // New node intends to join the subdivision of the network
5. // New node sends beacon message for joining the network
6.  &&  // Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. ℓ∈
then // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
8. 
ɀ
∈ // coordinator node-1 allows ‘ɀ′ the new node to be part of subdivision
the new node to be part of subdivision
9. Cn2
Sensors 2016, 16, 1438 9 of 19
9. Ґ // coordinator node-2 discards ’Ґ’ the request initiated by new node
10. Else if 
ɀ
∈  // If coordinator node-2 gets beacon message from new node for
joining the subdivision
11. ɀ∈ // coordinator node-2 allows ‘ɀ′ the new node to be part of subdivision
12. Ґ // coordinator node-2 discards ’Ґ’ the request initiated by new node
13. End if
14. End else
Here, we determine the energy consumed for four types of jobs. Thus, the residual energy of ‘C’
and ‘Cn can be calculated as follows:
 =󰇟{(×())+(
×())}×󰇠 (3)
Equation (3) shows the consumed energy by coordinator node for synchronization:
=
+
(4)
Equation (4) shows the consumed energy for broadcasting the message (disclosing the
information of newly joined nodes to the subdivision nodes):
=
(+
)×
 (5)
Equation (5) shows the consumed energy by coordinator node for scheduling with subdivision
nodes:
 =󰇟{(×())+(
×())}×󰇠 (6)
Equation (6) shows the consumed energy by coordinator node for collecting and forwarding the
data.
We can determine the residual energy of the coordinator node and non-coordinator nodes based
on the energy consumption for four tasks given by Equations (7) and (8), respectively:
 =
 ( +
+
+
) (7)
 =
 ( +
+
) (8)
Determining the node’s load is significant for hidden node discovery. The load factor ′N
requires the buffer capacity that is calculated d by using the Equation (9).
=
 (9)
Table 2 shows the details of the notations used and their respective explanations.
Table 2. Notations used and their description.
Notation Description
 Consumed energy of coordinator node for synchronization
Number of synchronized messages by each newly joining node
Energy consumed by radio of multimedia sensor
Energy consumed for amplifying
Number of synchronizing nodes
d Distance between coordinator and newly joined node
Energy consumed for broadcasting
Energy consumed by coordinator node for scheduling
C Coordinator node
Cn Non-coordinator nodes
 Coordinator’s initial energy
Nj// coordinator node-2 discards
Sensors 2016, 16, 1438 9 of 19
9. Ґ // coordinator node-2 discards ’Ґ’ the request initiated by new node
10. Else if 
ɀ
∈  // If coordinator node-2 gets beacon message from new node for
joining the subdivision
11. ɀ∈ // coordinator node-2 allows ‘ɀ′ the new node to be part of subdivision
12. Ґ // coordinator node-2 discards ’Ґ’ the request initiated by new node
13. End if
14. End else
Here, we determine the energy consumed for four types of jobs. Thus, the residual energy of ‘C’
and ‘Cn can be calculated as follows:
 =󰇟{(×())+(
×())}×󰇠 (3)
Equation (3) shows the consumed energy by coordinator node for synchronization:
=
+
(4)
Equation (4) shows the consumed energy for broadcasting the message (disclosing the
information of newly joined nodes to the subdivision nodes):
=
(+
)×
 (5)
Equation (5) shows the consumed energy by coordinator node for scheduling with subdivision
nodes:
 =󰇟{(×())+(
×())}×󰇠 (6)
Equation (6) shows the consumed energy by coordinator node for collecting and forwarding the
data.
We can determine the residual energy of the coordinator node and non-coordinator nodes based
on the energy consumption for four tasks given by Equations (7) and (8), respectively:
 =
 ( +
+
+
) (7)
 =
 ( +
+
) (8)
Determining the node’s load is significant for hidden node discovery. The load factor ′N
requires the buffer capacity that is calculated d by using the Equation (9).
=
 (9)
Table 2 shows the details of the notations used and their respective explanations.
Table 2. Notations used and their description.
Notation Description
 Consumed energy of coordinator node for synchronization
Number of synchronized messages by each newly joining node
Energy consumed by radio of multimedia sensor
Energy consumed for amplifying
Number of synchronizing nodes
d Distance between coordinator and newly joined node
Energy consumed for broadcasting
Energy consumed by coordinator node for scheduling
C Coordinator node
Cn Non-coordinator nodes
 Coordinator’s initial energy
the request initiated by new node
10. Else if Cn2
Sensors 2016, 16, 1438 8 of 19
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N × N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ’C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (: coordinator node-1; : coordinator node-2; : newly joined-node; :
beacon message for joining the subdivision; : node synchronization process; :
subdivision of the network; : listening)
2. Input: ()
3. Output: ()
4. Set attempts for  // New node intends to join the subdivision of the network
5. // New node sends beacon message for joining the network
6.  &&  // Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. ℓ∈
then // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
8. 
ɀ
∈ // coordinator node-1 allows ‘ɀ′ the new node to be part of subdivision
NjNsd // If coordinator node-2 gets beacon message from new node for
joining the subdivision
11. Cn2
Sensors 2016, 16, 1438 8 of 19
Figure 2. Coordination node synchronization process.
Let us assume that each sensor node communicates at the distance of the single-hop node to
detect the hidden nodes. Each sensor node sends the HELLO message ‘Hm’ at the distance ‘d’ within
the subdivision ‘Sd’ and is located at the N × N area of WMSN. The residual energy of the two types
of multimedia sensor nodes, the coordinator node ’C’ and the non-coordinator node ‘Cn’, can be
determined as follows.
The coordinator node performs four types of jobs that include the synchronization with newly
joined nodes, broadcasting the information of newly joined nodes inside the subdivision, scheduling
the transmission of subdivision, and data collection from non-coordinator nodes of the subdivision.
The synchronization process between the newly joined node and the coordinator is explained in
Algorithm 1.
Algorithm 1: Priority-based synchronization process between coordinator and newly joined nodes
1. Initialization: (: coordinator node-1; : coordinator node-2; : newly joined-node; :
beacon message for joining the subdivision; : node synchronization process; :
subdivision of the network; : listening)
2. Input: ()
3. Output: ()
4. Set attempts for  // New node intends to join the subdivision of the network
5. // New node sends beacon message for joining the network
6.  &&  // Beacon message sent by new node, but it is heard/listened by two
coordinator nodes when two subdivisions are overlapped.
7. ℓ∈
then // If coordinator node-1 gets beacon message from new node for
joining the subdivision.
8. 
ɀ
∈