Professional Documents
Culture Documents
HST - 0621-657
HST - 0621-657
HST - 0621-657
Secure Data Collection in Clustered Wireless Sensor Networks using Fuzzy based
scheme todetect Malicious Data Collector
Abstract—Energy efficiency in resource constrained network among the group become the Cluster Head (CH). The CH
like Wireless Sensor Network is one of the very important goal is responsible for collecting the data from other nodes in the
to be met for the successful adoptability of any new scheme group and forwarding the same towards base station. Many
proposed. Secure Data collection is one of the important task
in wireless sensor network. This paper address a novel energy literatures [3], [4], [5], [6], [7], [8] have proved that clustering-
efficient scheme for secure data collection in clustered wireless based approach is efficient compared to distributed approach
sensor network. The proposed scheme ensures the confidentiality with respect to network lifetime. To further improve network
of the data from the point it is sensed until it reaches the central lifetime aggregation /in network processing [9], [10] is also
point that is Base station. Also the scheme offers lightweight used. Another way to improve network lifetime is using mobile
authentication method to detect malicious data collector. The
proposed authentication scheme first classifies the node using node called Mobile Data Collector (MDC) for data collection.
Fuzzy logic if classification is not conclusive then it invokes Mobile nodes move in a predefined path all over a network to
second level authentication using ECC public key cryptosystem. collect data from sensor node or cluster heads [11], [12], [13],
The experimental results show that around 50% of the times [14], [15], [16], [17]. When mobile nodes are used for data
during multiple MDC visits it is possible to complete the collection other nodes in the network will save their energy as
authentication without using second level authentication which
involve compute intense cryptographic algorithms. Also the time they are not participating in data forwarding. In this type of
complexity experiments reveal that ECC based authentication is networks where MDC are used for data collection, it is also
efficient in terms of involved computations. important to address the path that has to be taken by MDC
Index Terms—Authentication, Mobile data Collector, Fuzzy in order to cover the network during data collection. Mobil-
rule, Malicious Node, Clustered WSN. ity patterns for mobile data collector are discussed in [18].
Most of the WSN applications demand secure data collection
I. INTRODUCTION thereby restricting attacker/malicious users from accessing
Wireless Sensor Networks (WSN) composed of sensor data in an unautho- rized way. Many schemes [19], [20],
nodes which can sense, process and transmit data to the [21]are proposed for se- cure data collection using mobile node
central node called Base Station (BS). Sensor nodes are and identify malicious mobile data collectors. In secure data
resource constrained in nature with limited computation, stor- collection using MDC, we have two important points which
age and communication capabilities and are battery driven. require attention when it comes to achieving security. First,
As the participating devices in WSN are sensor nodes, achieving confidentiality of the sensed data until it reaches
it is very important to conserve the energy. This energy the BS. Second important thing is detection of Malicious
conservation leads to longevity of the participating nodes MDC. Always security comes with a cost. In a resource
thereby increasing the overall network lifetime. Therefore, constrained network like WSN, in order to optimize the energy
the schemes/protocols/algorithms proposed for various WSN efficiency we always look towards light weight schemes. Here
operations for any category of applications will always look lightweight refers to the schemes which incur less load on
forward for energy efficiency. WSN has wide range of applica- nodes with respect to computation and communication. Fuzzy
tions [1] such as area monitoring, military surveillance, health logic based methods have been proved to be simple and
care and industrial applications. The range of application of also light weight in terms of computation and communication
WSNs now extends to home automation, and also to other compared to complex cryptographic algo- rithms [22], [23],
areas of human activity [2]. These networks can be distributed [24]. Efficient data gathering approach using fuzzy logic is
or clustered, in distributed WSN each node participates in proposed in [22]. Here fuzzy logic is used for selecting optimal
forwarding the data to BS. Where as in clustered WSN nodes route to be followed while gathering data in a WSN so that
organised into small groups called clusters and one node it optimizes the overall energy. In [23] a fuzzy based trust
Vol 50 1 Issue 6
Journal of Huazhong University of Science and Technology ISSN-1671-4512
model is proposed and evaluated to identify malicious nodes in four different propagation schemes for information delivery.
clustering. The scheme proposed in [24] identifies Also the results show that the schemes are able to achieve
misbehaving nodes before they actually participate in data security of the transmitted information. The compressive sens-
transmission. In this paper we propose a scheme for secure ing (CS) based data collection schemes can effectively reduce
data collection using Mobile Data Collector in clustered the transmission cost of wireless sensor networks (WSNs)
WSN. The scheme addresses two issues. First one is by exploring the sparsity of compressible signals. This work
achieving confidentiality of the collected data and the second is elaborated in [31]. In [19], three different protocols for
is to identify malicious MDC. Both the issues are addressed in secure data collection based on different assumptions and
an energy efficient way. To achieve confidentiality of the constraints are presented. Here the network model considered
data, suitable symmet- ric block cipher algorithm as in [25], is a clustered WSN with a mobile data collector traversing
[26] and public key cryptosystems [27] are used. To identify within the network to collect the data from cluster head. The
malicious MDC we use two level scheme. First level we try to proposed protocols are able to identify malicious MDC and
classify MDC as Malicious(M) or Non-Malicious (NM) based also can maintain confidentiality of the collected data.
on the proposed Fuzzy mechanism. If fuzzy mechanism fails
to identify MDC as malicious or non-malicious then we III. PROPOSED SCHEME
invoke the second level authentication where we use light In this section we present the details of the fuzzy based
weight public key cryptosystem, that is ECC algorithm [27] in scheme proposed for secure data collection. Firstly, we explain
order to verify the authenticity of MDC. The motivation for the network structure considered in the scheme. Next part
this two level scheme is, in WSN we deal with resource elaborates about achieving confidentiality of the collected data.
constrained devices where use of cryptographic algorithms How data is secured from eavesdropping attacks from the
every time for authentication become a costly affair in terms time it is collected from the sensor nodes till it reaches Base
of computation. This proposed two level scheme reduces the Station. Third part of the section summarizes the two level
burden of heavy computation as it uses cryptographic authentication scheme for identifying malicious MDC. Also
algorithm for identifying malicious node only when fuzzy data transfer from MDC to BS is presented. Finally we are
based mechanism fails to classify the MDC correctly. going to discuss about updating the fuzzy variables for the
Experimental results of the proposed scheme shows that next round data collection by MDC.
around 50% of the times during multiple MDC visits only
fuzzy method is sufficient to complete the authentication. A. Network Structure
Other 50% of the times as fuzzy method fails to identify a node The network consists of n number of sensor nodes denoted
correctly as a Non-malicious node, we will be invoking second as Si where i ranges from i = 1, ...n which are used for
level ECC based public key cryptosystem to complete the sensing a particular type of data. These sensor nodes are
MDC authentication. We also conducted experiments to show organised into groups called clusters, in a deployed network let
that indeed the ECC based second level authentication scheme us assume that we have m such clusters. One node with in the
is light weight in terms of computation overhead incurred cluster is selected as Cluster Head CHj where j = 1, ...m.
on Cluster head and MDC during this authentication process. Role of cluster head is uniformly rotated among the nodes
Therefore by restricting compute intense cryptographic algo- in cluster. Cluster head selection and rotation is done as
rithms we can improve overall network lifetime significantly. explained in [32]. After deployment cluster formation and use
of secret keys for secure communication within the cluster
II. LITERATURE SURVEY is as explained in [19]. Every sensor node Si forwards the
In the literature, we find different proposals addressing the sensed data to cluster head. The data transfer from CH to
use of Mobile data collectors to collect the sensed data in BS is done using Mobile Data Collector (MDC). The MDC
wireless senor networks. In [28] a survey of such schemes is considered here are special nodes with higher memory and
presented. In this survey the authors have covered different processing capability which can move in the monitoring area.
categories of data collection and have also compared each MDC are deployed by BS to collect the data at regular interval.
category in detail with respect to different parameters. In the These MDC traverse in the monitoring area to collect the data
work presented in [29] a detailed survey on use of mobile data and the collected data is carried to BS for further processing.
collector for data collection in WSN is discussed. The authors Every time when a MDC visits CHi , the cluster head initiates
have classified the survey into two major groups: data collec- authentication process to check whether the MDC is Malicious
tion using mobile sinks and data collection using mobile relay or not. Once the authenticity of MDC is verified, CHi transfers
nodes. In each group few important protocols are discussed the data to MDC which is carried to BS.
and also compared with each other. In the paper they have also
discussed open challenges with respect to data collection using B. Achieving confidentiality of the collected Data
mobile data collectors. Another set of mechanisms/schemes For any secure communication we need shared secret keys.
for efficient data collection in WSN are presented in [17]. In this scheme, we need appropriate secret keys for achieving
Authors have discussed all the schemes in brief. To achieve confidentiality and also for two level authentication. Let us
security, authors in [30] have proposed schemes to propagate discuss in detail various keys used in the proposed scheme
the information by dividing it into shares. They have proposed
Vol 50 2 Issue 6
Journal of Huazhong University of Science and Technology ISSN-1671-4512
Vol 50 3 Issue 6
Journal of Huazhong University of Science and Technology ISSN-1671-4512
Vol 50 4 Issue 6
Journal of Huazhong University of Science and Technology ISSN-1671-4512
TABLE II
At Cluster Head
SAMPLE RECORDED VALUES AFTER APPLYING MAX-MIN
RSA ECC
1.2
TV VI PCH Class
0.7 0.9 0.1 Non-Malicious 1
Time in seconds
0.5 0.2 0.4 Malicious 0.8
0.6 0.2 0.3 Malicious 0.6
0.9 0.7 0.1 Non- Malicious
0.4
0.3 0.2 0.9 Unable to Decide
0.2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
MDC classification using Fuzzy Method Iterations
60
50
Fig. 3. Time Complexity recorded at Cluster Head considering both RSA
40 and ECC cryptosystems.
30
20
3. Similarly time complexity at MDC is also recorded for the
10
second level authentication both by using RSA and ECC. The
0 graph in 4 shows the execution time required by RSA and
NM M UD
ECC algorithms for different iterations. Overall observation
on time complexity illustrates that the ECC based algorithm is
Fig. 2. Graph Representing the classification of MDC as NM, M or UD for computationally better compared to RSA algorithm. Therefore
the experiments conducted by considering different factor vector r. ECC based algorithm for second level authentication is a
suitable choice for resource constrained networks like WSN.
The performance analysis of RSA versus ECC is also studied
which the node belongs to are recorded. The results are in the literature [33]. Based this literature we experimented
tabulated. Few such recorded results are shown in Table with both RSA and ECC and concluded that ECC is suitable
2. The experiments are repeated and the graph is plotted for WSN applications.
for the outcome ie., the percentage of times the node is
classified as NM without initiating second level authentication. V. CONCLUSION
The graph is shown in 2. The recorded results show that In this work we have addressed Secure Data collection
around 50% of the time it is possible to classify a MDC using Mobile Data Collector. The proposed scheme ensures
as Malicious by using just Fuzzy method without using the confidentiality of the data from the time it is sensed by
second level authentication. This is a significant benefit one sensor node till it reaches the BS. For identifying malicious
can expect when we are dealing with resource constrained MDC we use lightweight authentication scheme comprising
environment like WSN where saving energy consumption of two levels. First level uses Fuzzy logic to classify a node.
is very crucial. The next set of experiments are conducted The fuzzy method classifies a node into one of the three
to study the time complexity of the proposed scheme. For classes Non-malicious, Malicious or Unable to Decide.If fuzzy
studying time complexity of the scheme we have considered method classifies a MDC as either Malicious or Unable to
the complete authentication part staring from invoking the Decide, second level authentication using ECC public key
fuzzy method until the decision by CH to transfer the data cryptosystem is used to confirm indeed the node is Malicious
or not to transfer. Here if fuzzy method fails to classify the or not. If this second level identifies the MDC as malicious,
node then public key based authentication scheme is invoked
as second level. For experimentation we have considered both
RSA and ECC based public key cryptosystem. First set of At MDC
experiments are conducted wherein we use RSA algorithm ECC RSA
0.6
factor vector r and every iteration running time is recorded
0.4
and tabulated. Next part of the experiments are conducted by
0.2
using ECC based public key cryptosystem for second level
0
authentication. For the same set of randomly selected factor 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
vector using ECC algorithm experiments are repeated and Iteration
running time is recorded. A comparative graph illustrating the
time complexity of the scheme recorded at cluster head CH Fig. 4. Time complexity recorded at MDC considering both RSA and ECC
by using RSA and ECC public key cryptosystems is shown in cryptosystems.
Vol 50 5 Issue 6
Journal of Huazhong University of Science and Technology ISSN-1671-4512
Vol 50 7 Issue 6