Erca energy efficient routing and reclustering

Advanced Computational Intelligence: An International Journal (ACII), Vol.3, No.1. January 2016
DOI: 10.5121/acii.2016.3102 11
ERCA: ENERGY-EFFICIENT ROUTING AND RE-
CLUSTERING ALGORITHM FOR CCEF TO EXTEND
NETWORK LIFETIME IN WSNS
Muhammad K. Shahzad1
, Jae Kwan Lee1
, and Tae Ho Cho1,
1 College of Information and Communication Engineering, Sungkyunkwan University,
Suwon 440-746, Republic of Korea
ABSTRACT
The pervasive application of wireless sensor networks (WNSs) is challenged by the scarce energy constraints
of sensor nodes. En-route filtering schemes, especially commutative cipher based en-route filtering (CCEF)
can saves energy with better filtering capacity. However, this approach suffer from fixed paths and inefficient
underlying routing designed for ad-hoc networks. Moreover, with decrease in remaining sensor nodes, the
probability of network partition increases. In this paper, we propose energy-efficient routing and re-clustering
algorithm (ERCA) to address these limitations. In proposed scheme with reduction in the number of sensor
nodes to certain thresh-hold the cluster size and transmission range dynamically maintain cluster node-density.
Performance results show that our approach demonstrate filtering-power, better energy-efficiency, and an
average gain over 285% in network lifetime.
KEYWORDS
Wireless sensor networks, energy-efficiency, network lifetime, re-clustering, filtering-power.
1. INTRODUCTION
Several en-route filtering schemes have been presented [3-6] which save energy by improved filtering
of false report attacks in wireless sensor networks (WSNs). However, routing protocols used are not
considered to save more energy. Commutative cipher based en-route filtering (CCEF) [1] can save
up to 32% energy in case of large number of injected fabricated reports. However, CCEF suffers from
the following problems; network lifetime is inefficient, fixed paths selection, and distance based
routing originally designed for ad-hoc networks. Against different of attacks or fabricated traffic ratio
(FTR) the security response is constant in CCEF. In this paper, we define FTR as number of attacks
divided by total number of events and security response is number of verification nodes assigned in
a path according to current FTR. The underlying routing in CCEF is greedy perimeter stateless
routing (GPSR) [2] which is not suited for energy constraint WSNs. In past [7] it has been observed
that unbalanced communication loads results in network partition or energy-hole problem which have
drastic effects on network lifetime.
In pre-deterministic key distribution based CCEF (PKCCEF) [8] with energy-efficient routing, saves
efficiency (16.05%) and prolongs network lifetime (81.01%). In this paper, energy-efficient routing

Corresponding author

12
and re-clustering algorithm (ERCA), significantly extends network lifetime (avg. 285%) while
maintains filtering-power. The reason for improvement in energy-efficiency and network lifetime is
due to dynamic re-clustering and sensor range as number of sensor nodes are reduced over time.
Furthermore, attacks information is also obtained without causing energy consumption at sensor
nodes. This FTR method will be explained in section 4.2 in detail. In this paper, to evaluate energy
consumption, we will use first order radio model [17, 18]. The main contributions of this paper are:
 Improved energy-efficiency
 Significant network lifetime extension
 Improved security
2. COMMUTATIVE CIPHER BASED EN-ROUTE FILTERING (CCEF)
In this section, query-response model and verification example of CCEF is explained as shown in
Figure 1.
In order to establish a session, the 𝐵𝑆 transmits a query message [𝑄𝑖𝑑, 𝐶𝐻𝑖𝑑, {𝑘 𝑠} 𝐾𝑛 𝐶𝐻
] to the 𝐶𝐻
that contains the Query ID (𝑄𝑖𝑑 ), 𝐶𝐻 ID (𝐶𝐻𝑖𝑑 ), and 𝑘 𝑠 key encrypted with the CH’s 𝑘n, i.e.,
{𝑘 𝑠} 𝐾𝑛 𝐶𝐻
as shown in Figure 1(a). In order to establish a session, a plain text 𝑘 𝑤 key is dropped on
each en-route node. Verification noes are randomly selected based on probabilistic method. In a
sensor field events are randomly generated which the 𝐵𝑆 is interested in know certain event in an
area in the field. In response the nodes receiving the event’s information, select a 𝐶𝐻, generate
response.
An event report is endorsed by the event sensing nodes and forward it to the 𝐶𝐻. Upon receiving the
query, the 𝐶𝐻 uses 𝑘 𝑛 key to decrypt the 𝑘 𝑠 key to check validity of the query. The 𝐶𝐻 compress
the 𝑀𝐴𝐶𝑠 generated by above event sensing nodes by using a simple 𝑋𝑂𝑅 operation. The response
message is sent along with the 𝑀𝐴𝐶𝑠 and the 𝐼𝐷𝑠 of these nodes, as illustrated in Figure 1(b). In
response, the 𝐶𝐻 sends a reply message [𝑄𝑖𝑑,𝑅, {𝐸𝑖𝑑, 𝐹𝑖𝑑, 𝐺𝑖𝑑}, 𝑀𝐴𝐶𝑠, 𝑀𝐴𝐶 𝑛], to establish a session.
For the predetermined session duration, reports are generated by the event-receiving nodes in
collaboration with the 𝐶𝐻 and are forwarded to the 𝐵𝑆.
Figure 1. Query-reponse and route setup process in CCEF

13
For event reports verification, the sensor nodes with 𝑘 𝑤 keys first extract 𝑄𝑖𝑑 to determined that the
query is valid, and if so, use 𝑘 𝑤 to verify the 𝑀𝐴𝐶𝑠 without knowing 𝑘 𝑠 by using the property of
commutative cipher. When the report in the response message reaches the 𝐵𝑆, it generates the 𝑀𝐴𝐶 𝑛
and verifies it along with 𝑀𝐴𝐶𝑠. This will verify the 𝐶𝐻 and all of the report-endorsing neighbours
if both tests are validated. If not, either the 𝐶𝐻 or one or more endorsing nodes is compromised.
3. PROPOSED SCHEME
In this section, system models and overview is presented.
3.1. System models
3.1.1. Network model
The 𝑁 number of sensor nodes are randomly deployed within field 𝐹 of area 𝐴 = 𝐹ℎ × 𝐹𝑤 with
radius of 𝑅 as illustrated in the Figure 2. This sensor field comprise of 𝑁 number of sensor nodes
represented by: {𝑆1, 𝑆2, 𝑆3, … , 𝑆 𝑛} respectively. The location of the 𝐵𝑆 is right mid edge of the
sensor field. The clusters are represented by: {𝐶1, 𝐶2, 𝐶3, … , 𝐶 𝑛}. At startup phase all the cluster are
of equal size with 𝐴 = 𝐶ℎ × 𝐶 𝑤. In each cluster equal number of nodes are randomly distributed,
referred as node density or ñ. As the number of sensor nodes decreases with time due to energy
depletion (i.e., remaining energy become zero and no more communication is possible), re-clustering
and transmission range are adapted to maintain ñ. It will be further elaborated in section 3.2.3.
Following are the assumptions related to this model:
 Network is composed of static homogenous nodes
 Communication links are symmetric
 Nodes can adjust transmission power as per distance
3.2.2. Energy consumption model
In this paper, we have used the first-order radio model [17] with a free space (𝑑2
power loss) channel
model. A sensor node is composed of a radio communication components, data processing unit, micro
sensor unit, antenna to transmit or receive data, amplifier, and power supply. In this paper for
simplicity, we only consider the energy dissipation that is associated with the radio component and
Figure 2. Sensor field

14
electronics. A simple and commonly used first-order radio model block diagram is shown in Figure
3. The packet of 𝑚 bits propagation at a distance 𝑑 between the transmitter and receiver antennas,
require the transmission energy of 𝐸 𝑇𝑥
(𝑘, 𝑑) as shown by Equation (1).
𝐸 𝑇𝑥
(𝑚, 𝑑) = 𝐸𝑒𝑙𝑒𝑐 × 𝑚 + 𝐸 𝑎𝑚𝑝 × 𝑚 × 𝑑λ
(1)
𝐸𝑒𝑙𝑒𝑐 is the energy used by the electronics of the circuit and 𝐸𝑒𝑙𝑒𝑐 × 𝑚 is the energy used by the
transmitter to sent 𝑚 bits. Furthermore, 𝐸 𝑎𝑚𝑝 is the energy required by the amplifier, and 𝜆 is the
path loss constant. Similarly, 𝐸 𝑅 𝑥
(𝑚) is the energy required to receive k-bits as shown in Equation (2).
𝐸 𝑅 𝑥
(𝑘𝑚) = 𝐸𝑒𝑙𝑒𝑐 × 𝑚 (2)
The energy used by transmitter amplifier for transmission is 𝐸 𝑎𝑚𝑝 = 100𝑝𝐽/𝑏𝑖𝑡/𝑚2
. In addition the
energy used by circuitry of transmitter and receiver is 50 𝑛𝐽/𝑏𝑖𝑡.
3.2.3. Clustering and re-clustering method
Over the time sensor nodes are depleted as the energy level reaches to zero. The probability of
depletion of the sensor nodes is higher near the 𝐵𝑆 and on the paths. This reduces the number of
event sensing nodes or report endorsing nodes which make it easier for the adversary to forge a report.
This not only make easier to compromise security but also shortens the network lifetime. The
depletion of nodes around sink and one established paths cause energy-holes which have adverse
effect on network lifetime. This indicates that maintaining number of sensor nodes in a cluster can
help security and extend network lifetime.
However, physical installation of sensor nodes may be both costly and hazardous task. A simple
solution can be adjusting cluster size and transmission range to maintain cluster node density.
Suppose ñ is the cluster node density at the network deployment or our desired number of nodes to
be maintained in a cluster. In this paper as 10% of total number of nodes are depleted we apply cluster
node density adaptive method to re-cluster the sensor network. The re-clustering formula used in this
paper is illustrated with Equation (3).
𝐶𝑠𝑖𝑧𝑒 𝑘
=
ñ𝐹ℎ
2
𝑁 𝑛𝑘
(3)
As the total number of sensor nodes (𝑁𝑛𝑘) decreases, the cluster size at the 𝑘 𝑡ℎ
step is increased given
ñ and the field higher and width. We assume same height and width of a cluster.
The number of 𝐶𝐻𝑠 in a row (𝑁𝐶𝐻𝑠 𝑟
) or column (𝑁𝐶𝐻𝑠 𝑐
) of the sensor field is depicted by Equation
(4).
𝑁𝐶𝐻𝑠 𝑟
= 𝑁𝐶𝐻𝑠 𝑐
= √𝑁𝑛𝑘 ñ⁄ (4)
The cluster, 𝑘 height (𝐶 𝑘ℎ) or width (𝐶 𝑘𝑤)in field 𝐹 with height 𝐹ℎ and width 𝐹𝑤is defined by
Equations (5) and (6).
𝐶 𝑘ℎ = 𝐹ℎ 𝑁𝐶𝐻𝑠 𝑟
⁄ (5)
𝐶 𝑘𝑤 = 𝐹𝑤 𝑁𝐶𝐻𝑠 𝑐
⁄ (6)
Figure 3. First order radio model for energy consumption

15
Similarly, new range, 𝑅 𝑘 at the 𝑘 𝑡ℎ
step is given by Equation (7)
𝑅 𝑘 =
𝐶 𝑘ℎ
𝜕
(7)
Where 𝜕 = 𝐶𝑖ℎ/𝑅𝑖 represent the system or the design parameter.
Initially all nodes are with same initial energy and there are no depleted nodes. Re-clustering
algorithm keep track of the number of nodes depleted due to energy depletion. Our proposed method
take care of the network topology conditions (i.e., range, cluster size etc.) and can adjust these
parameters to maintain ñ. This does not only help extending network lifetime but also help
maintaining security by taking care of minimum nodes in a given cluster.
3.3. System Overview
This section illustrates; setup initialization, key distribution, and forwarding node selection for the
proposed method.
3.3.1. Network construction phase
A sensor field of area (500 × 500) m2
with 1000 densely deployed sensor nodes is considered. In
this phase sensor nodes are assumed secure during the network construction at boot-up process. It is
also assumed that the 𝐵𝑆 cannot be compromised during this phase. The sensor nodes have a fixed
initial energy of 1 Joules. At this phase, the randomly deployed nodes are granted unique 𝐼𝐷𝑠 and 𝑘 𝑛
keys. Furthermore, each node knows it’s location through some location mechanism to help it
calculate its distance from the 𝐵𝑆.
3.3.2. Key distribution phase
The 𝑘 𝑤 keys are distributed pre-deterministically for each session to a randomly selected percentage
of sensor nodes on a path. The 𝑘 𝑤 is pre-deterministically disseminated before a session in the
network, while 𝑘 𝑠 is transmitted securely to the 𝐶𝐻 in a query message. In contrast in CCEF, keys
are distributed to all nodes on the path in the query message, whereas, the filtering nodes are assigned
using probabilistic method 𝑝 =
1
𝛼ℎ
; where 𝛼 is design parameter and ℎ is distance or number of
hopes. In ERCA, keys are only possessed by a predetermined percentage of nodes in the path and the
same nodes are used as filtering nodes in response message. The percentage of nodes to be granted
𝑘 𝑤 keys are selected based on current FTR or attacks. Due to this reason, based on current FTR, we
can dynamically determine a corresponding secure path. This method can dynamically select a
suitable path depending upon the attack information. A session is changed after 𝑡 time or after a node
is depleted.
3.3.3. Route set-up phase
In the proposed method the route setup process is similar to CCEF with improved key distribution
method and a forwarding node selection method.
The routing in proposed method is dynamic instead fixed path routing considering different
parameters in addition of distance only as in CCEF. In order to make scheme to respond to different
FTR, we consider presence of keys on nodes. If there are more attacks or higher FTR a path with
more nodes having keys is preferred. This means with higher FTR ratio a path with more verification
nodes (i.e., nodes with 𝑘 𝑤 keys) is selected and vice versa. The 𝐵𝑆 sends a query message to

16
establish a path through the 𝐶𝐻 in an area of interest. As events can take place randomly in different
clusters, multiple sessions can be established at a given time between the 𝐵𝑆 and 𝐶𝐻𝑠.
3.3.4. Forwarding node selection method
Proposed method for selecting the next forwarding node among number of given candidate nodes to
create a path is shown in Equation (8).
𝐹𝑛 = arg 𝑚𝑎𝑥 {𝑘 × (1 −
β
2
) + (𝑑 + 𝑒) ×
𝛼
2
} (8)
Where 𝛼 and are is the system design parameter, 𝑑 is the shortest distance of the closest neighbor
to the sink, 𝑒 is the residual energy, and 𝑘 is the presence of 𝑘 𝑤 key on that node. All are variable
and normalized by one and the node with the highest evaluation among candidate nodes is selected
as forwarding node. By repeating this process a path is created between the set of the 𝐵𝑆 and the
source 𝐶𝐻𝑠.
4. PERFORMANCE EVALUATION
4.1. Experimental Environment
A 1000-node sensor node are randomly distributed in field of area 500 × 500 𝑚2
consisting of 𝑘 =
100 clusters. In each cluster, a fixed number of nodes ñ = 10 are positioned at random locations.
Each of the sensor nodes has an initial range of 50 𝑚 which is used to determine the neighbours,
candidate, and forwarding nodes. The 𝐵𝑆 is located at (500, 250) m and aware of the node 𝐼𝐷𝑠,
locations, and node keys (𝑘 𝑛) of all of the sensor nodes. The communication links are assumed to be
bidirectional. Each node also assumes a unique 𝐼𝐷 and knows its 𝑘 𝑛 key. Table 1 illustrates the
experimental parameters for the performance analysis. The values of 𝐸𝑒𝑙𝑒𝑐 and 𝐸 𝑎𝑚𝑝 are chosen to
achieve an acceptable ratio of 𝐸 𝑏/𝑁0 [18]. The data packet or message size of 200 bits (referred as
one time step) and a round is defined as four time steps of data received at the 𝐵𝑆.
Table 1. Experiment parameters
Parameters CCEF ERCA
Sensors nodes 1000 -
Sensor field size (500 × 500) m2 -
BS location (250, 0) m -
𝑅𝑖 50 m Variable
Cluster h/w 50 m Variable
𝐸𝑒𝑙𝑒𝑐 for Tx and Rx 50 nJ/bit -
𝐸 𝑎𝑚𝑝 100 pJ/bit/m2 -
Node energy 1 Joules -
MAC verification 20 mJ -
Data packet 32 bytes -
Round 128 bytes -
FTR 30% -
Path loss constant (λ) 2 -

17
4.2. Attack information
In CCEF and ERCA the communication is query driven. In this model a query message is initiated
by the 𝐵𝑆 to inquire about occurrence of an event in an area of interest. In response, a number of
message reports of the event are transmitted during the session. Now, we explain how we can
determine the number of attacks or FTR information without causing extra energy or message at the
sensor nodes.
The number of event reports through designated 𝐶𝐻 are known to the 𝐵𝑆. A legitimate report
received at the 𝐵𝑆 will increment its counter by one to determine total number such reports. This
need no extra message at the sensor nodes. For fabricated report there can be two cases or either
report will be dropped en-route due verification is failed or at the 𝐵𝑆. In first case, the 𝐵𝑆 after
waiting for designated time window 𝑡 will consider it dropped and increment the fabricated report
counter by one. Similarly, verification failure at the 𝐵𝑆 will cause the fabricated report counter by
one since it is also known.
Since, we can determine total number of legitimate and fabricated reports, by using equation (9), the
current FTR ratio can be calculated for 𝑛 events
𝐹𝑇𝑅 = ∑
𝐹 𝑅
𝐹 𝑅+ 𝐿 𝑅
𝑛
𝑒=1 (9)
4.3. Experimental results
The metrics used to evaluate network lifetime and energy-efficiency are; first node depleted (FND),
half nodes depleted (HND), and total node depleted (TND). The depleted node is one which have
used its total energy and cannot participate in communication. As mentioned earlier one round is four
time steps 800 bits of data received at the 𝐵𝑆. One time step is equal to the one packet size of 200
bits. A measure of better network lifetime performance is more number of rounds before node(s)
depletion or maximum number of nodes depleted at the end of the simulation experiment. In case of
energy-efficiency which scheme consume less average energy per round is more energy-efficient as
per FND and HND metrics. The performance is evaluated at different network sizes (nodes) so some
fluctuations are present in graphs.
4.3.1. Network Lifetime
In Figure 4 the network lifetime performance comparison of CCEF and EECA is shown using FND
performance metric. The x-coordinate depicts network size in terms of number of sensor nodes and
y-coordinate represents number of rounds. EECA shows a significant performance gain of an average
3.309 times or 330.90% over CCEF. For all network sizes (number of nodes) proposed scheme
performance better than CCEF. The performance improvement of EECA over CCEF is 2.402 folds
as shown in the Figure 5 for performance metric HND. The network lifetime performance of EECA
is better than CCEF in most cases except at network size of 300 and 200 nodes. As shown in Table 2
the average performance gain is 2.856 times in two cases.
The Figure 6 illustrates performance comparison of CCEF and ERCA for %age of nodes depleted at
the end of the simulation experiment for different network sizes. Except for the first case the ERCA
performance is significantly better as compare the CCEF. The performance improvement is achieved
due to more balance network energy consumption strategies which balance communication loads

18
over large group of sensor nodes. Results have indicated that energy-efficient routing, dynamic path
selection based on network conditions, and re-clustering help in solving energy-hole problem.
Table 2. Network lifetime gain of EECA over CCEF
metric FND HND avg.
lifetime 3.309 2.402 2.856
Figure 4. Network lifetime first node depleted (FND)
Figure 5. Network lifetime half nodes depleted (HND)
Figure 6. Network lifetime %nodes depleted (TND)

19
4.3.2. Energy efficiency
In this section energy-efficiency performance of EECA and CCEF is discussed. Figure 7 shows that
EECA have advantage over CCEF in energy-efficiency using FND metric. In all cases for different
network sizes EECA performs better than CCEF in average energy-efficiency per round. The
proposed scheme improved on average energy saving of 3.50% for different network sizes.
The case of performance improvement using HND metric is shown in the Fig. 8. In this case the
average energy saving is 3.46%. The energy saving performance is summarizes in Table 3.
Table 3. Energy-efficiency gain of EECA over CCEF
metric FND HND avg.
energy % 3.50 3.46 3.48
Figure 7. Energy-efficiency first node depleted (FND)
Figure 8. Energy-efficiency half nodes depleted (HND)

20
4.3.3. Security
In this section filtering-power of two compared schemes in highlighted. With pre-deterministic key
re-distribution help achieving better performance as compare to the original scheme. The Figure 9
shows that EECA filtering-power in comparison with CCEF. Average filtering-power of for CCEF
and ERCA are 71.31% and 75.93% respectively.
5. RELATED WORK
In this section, research work related to our proposed scheme is explained. En-route filtering scheme
such as CCEF [1] can save energy by early dropping of false report attacks. CCEF forms a secret
association among the nodes and the 𝐵𝑆 for entire session where each node possesses its own
witness. The en-route sensor nodes do not require to share a symmetric key, thereby offering stronger
security protection than the existing symmetric key sharing schemes. Re-clustering is required,
because; 1) when number of sensor nodes are less then 𝑡 nodes the security is compromised and 2)
network lifetime suffer from adverse effect. Security is compromised with less number of nodes in a
cluster due to the need of less number compromised keys are needed to forge a report. It has adverse
effects on network lifetime as with fewer nodes in a cluster limits the reachability.
The underlying routing protocol in CCEF is GPSR [3], which forwards packets based on a greedy
approach in terms of the next-hop closer to the destination. GPSR does not consider residual energy
level of the sensor nodes to determine the next hope. However, for energy constraint WSNs, it suffer
from number of constraint; 1) consider distance only 2) fix path routing and 3) inefficient network
lifetime.
Recently, several security schemes has been proposed in order to achieve energy-efficiency. One
approach is to filter false reports en-route as early as possible. To cater this challenge several en-route
filtering has been proposed. Statistical en-route filtering (SEF) [3] has first addressed the false report
detection problems in WSNs by finding the number of compromised sensor nodes. It presents the
general en-route filtering framework to serves as the source of subsequent en-route filtering security
protocols. Dynamic en-route filtering (DEF) [4] uses the novel hill climbing method for key
dissemination to filter false reports earlier, requires a key chain for authentication by each node in
the chain.
Figure 9. Filtering power of CCEF and EECA

21
The interleaved hop-by-hop authentication scheme (IHA) [5] can determine false data reports in case
of no more than 𝑡 nodes are compromised. It presents an upper bound to the number of hops a false
report can travel before getting dropped in the presence of 𝑡 colluding nodes. IHA is also based on
GPSR routing as in CCEF and suffer from similar limitations. In a probabilistic voting-based filtering
scheme (PVFS) [6], the number of message authentication controls (MACs) or votes is used to detect
both fabricated reports with false votes and false votes on valid report attacks. The study [7] explores
the uneven consumption of the energy in sink based networks. This problem results in energy holes
having drastic effects on network lifetime.
In order to address, related limitations, several variants of above en-route filtering schemes has been
propose to increase energy-efficiency and/or extend network lifetime. PKCCEF [8] improves CCEF
which by using energy aware routing, improves network lifetime and saves energy. In research [9], a
key index-based routing for filtering false event reports in the WSN is proposed. In this approach
each node selects a path from the source to the destination based on the key index information of its
neighbour nodes.
In work [10], a fuzzy-based path selection method (FPSM) is proposed which improves the detection
of false reports in the WSN. In this approach each cluster selects paths by considering the detection
capacity and the energy-efficiency. These schemes do not utilized re-clustering to further improve
network lifetime. The paper [11] addresses the limitations of IHA, which is based on a single fixed
path between the source and the sink. In order to address this problem, a multipath interleaved hop-
by-hop authentication (MIHA) scheme is proposed. It creates multiple paths and select another
available path if current path has re are 𝑡 compromised nodes. Results demonstrate that it is more
energy-efficient and can filter more attacks.
In work [12], it is observed that uneven distribution of the communication loads often results in
energy hole problem. In order to counter this problem optimal and adjustable transmission ranges are
dynamically assigned to improve the network lifetime. Performance evaluation shows the near
optimal solution to prolong network lifetime both in uniform and non-uniform sensor nodes
deployment. The effect of cluster size on energy consumption has been study by authors in an energy-
efficient multi-hop hierarchical routing protocol (MHRP) for WSNs [13].
The authors in [14] studies the cluster size issue in a practitioner reference of the communication
required for data collection. In [15], the authors present the optimal cluster size considering the
network lifetime and energy-efficiency. These approaches demonstrates that optimal clustering sizes
or re-clustering can extend the network lifetime.
In the adaptive decentralized re-clustering protocol (ADRP) [16], the 𝐶𝐻𝑠 and the next 𝐶𝐻𝑠 are
determined by considering the residual energy level of each node and the average energy of each
cluster. Results shows that improved network lifetime as compared to the other clustering protocols.
However, these schemes does not support mobile sink. The paper [17] presents several radio
transmission model. In order to calculate the energy dissipation in this paper, we first order radio
model [18].

22
6. CONCLUSIONS AND FUTURE WORK
By distributing and balancing the communication loads over a larger group nodes EECA has been
able cater with energy hole or network partition problem. Keys are pre-deterministically re-
distributed on different paths to respond to different FTR ratios or attack frequency. This enable
attack based dynamic path selection based routing. This helps in load balance over multiple paths
alternatively which extend network lifetime.
We have saved energy by better detection of fabricated reports which limits number of hopes. In case
of higher FTR, more verification nodes are assigned results in higher filtering of fabricated report.
Whereas, when FTR is low, less number of verification nodes are selected resulting is less number
of verifications for legitimate reports. Another main reason for significant network lifetime extension
is re-clustering ability to reach nodes when node density in a cluster is reduced with depletion of
sensor nodes. To maintain the node density proposed scheme adjust cluster size and transmission
range. In future work we aim to achieve more energy-efficiency and improved filtering-power by
selecting filtering nodes using fuzzy logic instead of pre-deterministic or probabilistic methods. The
filtering capacity can further improved by using Generic Algorithm (GA) to optimized fuzzy
membership functions.
ACKNOWLEDGEMENTS
This research was supported by the Basic Science Research Program through the National Research
Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (No.
2013R1A2A2A01013971).
REFERENCES
[1]. Hao Yang and Songwu Lu. Commutative cipher based en-route filtering in wireless sensor
networks. 60th Vehicular Technology Conference, 2004, vol. 2, pp. 1223-1227.
[2]. B. Karp and H. T. Kung. GPSR: Greedy perimeter stateless routing for wireless networks. ACM
MobiCom, 2000, pp. 243-254.
[3]. F. Ye, H. Luo, S. Lu, and L. Zhang. Statistical en-route filtering of injected false data in sensor
networks. In IEEE Proceedings of INFOCOM, 2004, pp. 839-850.
[4]. Zhen Yu, and Yong Guan. A dynamic en-route filtering scheme for data reporting in wireless
sensor networks. IEEE/ACM Transactions on Networking, 2010, vol. 18(1), pp.150-163.
[5]. S. Zhu, S. Setia, S. Jajodia, and P. Ning. An interleaved hop-by-hop authentication scheme for
filtering of injected false data in sensor networks. Proceedings of EEE Symposium on Security
and Privacy, 2004, pp. 259-271.
[6]. Feng Li and Jie Wu. A probabilistic voting-based filtering scheme in wireless sensor networks.
Vancour, Canada, ACM IWCMC, 2006, pp. 27-32.
[7]. Liu, Tao. Avoiding energy holes to maximize network lifetime in gradient sinking sensor
networks. Wireless Personal Communication. Springer Science + Business Media, LLC, 2012,
pp. 581-600.

23
[8]. Muhammad K Shahzad and Tae Ho Cho, Extending the Network Lifetime by Pre-deterministic
Key Distribution in CCEF. Wireless Sensor Networks. Wireless Networks, 2015, DOI
10.1007/s11276-015-0941-0.
[9]. S. Y. Moon and T. H. Cho. Key index-based routing for filtering false event reports in wireless
sensor networks. IEEE Transaction on Communication. Tokyo. Japan, 2012, vol. E95-B (9), pp.
2807-2814.
[10].Hae Young LEE and Tae Ho CHO. Fuzzy-based path selection method for improving the
detection of false reports in sensor networks. IEICE Transaction on Information and System,
2009, pp. 1574-1576.
[11].P.T. Nghiem and T.H. Cho. A multi-path interleaved hop by hop en-route filtering scheme in
wireless sensor networks. Computer Communications, 2010, vol. 33(10), pp. 1202-1209.
[12].Chao Songa, Ming Liu, Jiannong Cao, Yuan Zheng, Haigang Gong, and Guihai Chen.
Maximizing network lifetime based on transmission range adjustment in wireless sensor
networks. Computer Communications, 2009, pp. 1316–1325.
[13].Jin Wang, Xiaoqin Yang, Yuhui Zheng, Jianwei Zhang and Jeong-Uk Kim. An energy-efficient
multi-hop hierarchical routing protocol for wireless sensor networks. International Journal of
Future Generation Communication and Networking, 2012, vol. 5(4), pp. 89-98.
[14].Anna Forster, Alexander Forster and Amy L. Murphy. Optimal cluster sizes for wireless sensor
networks: An experimental analysis. Lecture Notes of the Institute for Computer Sciences.
Social Informatics and Telecommunications Engineering, 2010, vol. 28, pp 49-63.
[15].Amini N, Vahdatpour A, Xu W, Gerla M, and Sarrafzadeh M. Cluster size optimization in sensor
networks with decentralized cluster-based protocols. Computer Communications, 2012, vol. 35,
pp.207-220.
[16].Fuad Bajaber, Irfan Awan. Adaptive decentralized re-clustering protocol for wireless sensor
networks. Journal of Computer and System Sciences, 2011, vol. 77, pp. 282–292.
[17].Swarup Kumar Mitra, Mrinal Kanti Naskar. Comparative study of radio models for data
gathering in wireless sensor network. International Journal of Computer Applications, 2011, vol.
27(4), pp. 49-57.
[18].Wendi Rabiner Heinzelman, Anantha Chandrakasan, and Hari Balakrishnan, “Energy-Efficient
Communication Protocol for Wireless Sensor Networks,” Proceedings of the Hawaii
International Conference on System Sciences, 2000, pp. 1-10, January 4-7
Authors
Muhammad Khuram Shahzad received a B.E.I.T degree from the University of Lahore and an M.S. degree in
Information Technology from the National University of Science and Technology, Isalamabad, Pakistan in 2004 and
2007, respectively. He is now a Ph.D. scholar in the College of Information and Communication Engineering at
Sungkyunkwan University, South Korea. His research interests include wireless sensor networks and graph gheory.
Jae Kwan Lee received his B.S. degrees in computer information from BaekSeok Unive rsity, Korea, in February 2013.
He completed his M.S. program from Sungkynkuwan University, Korea, in 2015. He is currently a PhD student in the
College of Information and Communication Engineering at Sungkyunkwan University. His research interests include
wireless sensor network security, intelligent system and modelling & simulation.

24
Tae Ho Cho (Corresponding author) received a Ph.D. degree in Electrical and Computer Engineering from the
University of Arizona, USA, in 1993, and B.S. and M.S. degrees in Electrical Engineering from Sungkyunkwan
University, Korea and the University of Alabama, USA, respectively. He is currently a Professor in the College of
Information and Communication Engineering at Sungkyunkwan University, Korea. His research interests include
wireless sensor networks, intelligent systems, modeling and simulation, and enterprise resource planning.

Erca energy efficient routing and reclustering

More Related Content

What's hot (15)

Viewers also liked (17)

Similar to Erca energy efficient routing and reclustering (20)

Recently uploaded (20)

Erca energy efficient routing and reclustering