Recent advances of secure clustering protocols in wireless sensor networks

Đăng ngày 4/2/2019 3:59:05 PM | Thể loại: | Lần tải: 0 | Lần xem: 9 | Page: 14 | FileSize: 0.31 M | File type: PDF
Recent advances of secure clustering protocols in wireless sensor networks. In this article, we review the most recent secure clustering protocols in WSNs. We start with a description of the security requirements for WSNs and discuss the existing security schemes. We analyze to what extend they have been applied to the clustering structure of WSNs. Then, we review secure clustering protocols in emerged recent years. Finally, we present a set of criteria which must be applied to build a secure clustering algorithm.
International Journal of Computer Networks and Communications Security
VOL. 2, NO. 11, NOVEMBER 2014, 400–413
Available online at: www.ijcncs.org
ISSN 2308-9830
Recent Advances of Secure Clustering Protocols in Wireless
Sensor Networks
MOHAMED ELHOSENY1, HAMDY K. EL-MINIR2, A. M. RIAD3 AND XIAOHUI YUAN4
1, 4 Department of Computer Science and Engineering, University of North Texas, U.S.A.
1, 3 Department of Information Systems, Mansoura University, Egypt
2 Department of Electrical Engineering, Kafr El-Sheikh University, Egypt
E-mail: 1mohamed.elhoseny@unt.edu, 2hamdy elminir@eng.kfs.edu.eg, 3amriad2000@yahoo.com,
4xiaohui.yuan@unt.edu
ABSTRACT
Wireless Sensor Networks (WSNs) have been employed in many real-world applications that greatly
improve our life. The ubiquitous WSNs make security a prime issue, and new technologies have been
developed recently. In this article, we review the most recent secure clustering protocols in WSNs. We start
with a description of the security requirements for WSNs and discuss the existing security schemes. We
analyze to what extend they have been applied to the clustering structure of WSNs. Then, we review secure
clustering protocols in emerged recent years. Finally, we present a set of criteria which must be applied to
build a secure clustering algorithm.
Keywords:
Wireless
Sensor
Networks,
Secure
Cluster
Formation,
Secure
Routing,
Secure
data
Aggregation, Security Attacks.
1
INTRODUCTION
There
have
been
reviews
on
the
security
procedures and threats of WSNs [5]–[7], and others
Wireless
Sensor
Network
(WSN)
consists
of
discuss
the
procedures
related
to
the
clustering
sensor
devices
transpired
in
tangible
insecure
model
specifically
[8]–[11].
However,
most
of
environments in order to collect data. There have
these
reviews
evaluated
the
secure
clustering
been many applications in health care monitoring,
algorithms based on two main processes only, i.e.,
environmental monitoring, industrial logging, etc.
CH selection and cluster formation [10], [12], [13].
The data collected could be sensitive and relevant
Other works discussed the existing secure routing
to privacy, which makes security a prime issue [1],
protocols
of
clustering
model
with
the
aim
of
[2]. Unlike conventional networked devices, factors
protecting the data transmission CHs and the base
such
as
open
communication
medium,
limited
station [8], [14]–[16]. But for any secure clustering
computational
capabilities
of
nodes,
and
the
algorithm, a set of criteria must be used to be an
disadvantages of bandwidth constraint make WSN
effective one.
more susceptible to malicious attacks [3].
To
increase
network
life
and
reduce
energy
consumption, cluster model was proposed [4]. In
this model the energy of sensor nodes are reserved
by
involving
them
in
multi-hop
communication
within
a
particular
cluster
and
performing
data
assembling and fusion as shown in Figure 1. Each
cluster
has a head node that is responsible for
gathering data from all nodes within the cluster and
sending the aggregated message to the base station.
Fig. 1. Clustering Model for Wireless Sensor Network
401
M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014
These procedures include secure CH building,
from inside, outside, or both [5]. Table I lists the
secure cluster formation, secure data aggregation
common types of attack in WSNs.
from cluster members, secure data routing from
These attacks aim to affect the transmitted data
CHs to the base station, robustness against different
with one of the following threats [17]:
types of attack, efficiency in terms of WSN
resources limitations, and ability to deal with
dynamic clustering environment. For example, the
survey paper [10] considers that all selected
algorithms that it explains are secure and efficient.
• Interruption: is an attack on the availability of
the network. Its main aim is to make an asset of
the system, i.e., sensor node, unavailable or
unusable. Denial of Service attacks [5] have
However, they do not pay much attention on energy
constraints when different security mechanisms are
used. This is very crucial because technique based
become very
interruption.
well-known
example
of
probabilistic
and
deterministic
strategies
have
• Interception: is an attack on confidentiality.
different impacts on energy consumption which
will affect the network efficiency and performance
directly. In addition, the survey [10] did not address
the performance requirements study (e.g. memory
requirement, computation overhead etc.), which is
The sensor network can be compromised by
the attacker to gain unauthorized access to
sensor node or data store within it. Spoofing
attack is a well-known example.
more important because it is strictly bound to the
consumed energy. Based on that, this work not
limited to a specific point in secure clustering, it is
an attempt to combine all these processes, i.e.,
secure cluster building and secure data
transmission, as evaluation criteria for some secure
• Modification: is an attack on integrity of the
system. It this attack unauthorized party not
only accesses the data but also modifies the
content of a message being transmitted in a
network.
clustering algorithm.
The rest of this paper is organized as follows:
section 2 review the common types of attacks for
WSN and the source of each one. Section 3
• Fabrication: is an attack on authentication in
which the attacker make an insertion of
messages in a network and tries to make it as it
explains the evaluation criteria that is used in this
paper to evaluate the secure clustering algorithms.
Section 4 discusses the most popular security
mechanisms which can be used with the clustering
is sent from authorized node.
• Methods to address WSN security attacks aim
at the following aspects [18]:
model in WSN. Section 5 reviews some of existing
secure clustering algorithms with the strengths and
the limitations of each one. After that, section 6
• Preventing Attacks: It aims
attack before it happens.
to prevent any
Any proposed
analyzes the discussed secure clustering algorithms
technique
will
have
to
defend
against
the
using the proposed evaluation
section 7 summarizes the paper.
criteria.
Finally,
targeted attack.
• Detecting Attacks: If an attacker manages to
2
SECURITY ISSUES IN WSNs
pass
the
measures
taken
by
the
prevention
Security attacks against WSNs can be
categorized into two types: active attacks and
passive attacks. In passive attacks, attackers are
typically hidden and aim to monitor the
mechanism, the security solution would
immediately switch into the detection phase of
the counter attack in progress and specifically
identify the nodes that are being compromised.
communication link to collect data. The common
examples of passive attacks are eavesdropping,
node malfunctioning, node destruction and traffic
analysis types. In active attacks, the attacker affects
• Removing Attacks: It aims to mitigate any
attack after it happens by removing the affected
nodes and securing the network.
the operations of network, i.e., the transmitted data.
For example, the network services may degrade or
terminate as a result of these attacks. The common
3
EVALUATION CRITERIA FOR SECURE
CLUSTERING IN WSNs
examples of active attacks are Denial-of-Service
(DoS), hole attacks, flooding and Sybil types [6].
The source of the attack can come to the network
In this section we discuss the criteria which we
will use to evaluate the existing secure clustering
method.
402
M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014
A.
Completeness
without any alteration, a technique like hash
Secure clustering is a sequential process that
must guarantee the security goals, i.e.
confidentiality, integrity, and availability, in each
phase. This process consists of two stages: cluster
function can be used.
• Confidentiality: Confidentiality prevents
sensitive information from reaching the wrong
party, while making sure that the right party
building and data transmission. The cluster building
stage starts with cluster formation in which the
cluster heads (CHs) are determined and nodes are
assigned to the CHs. The next stage, i.e., data
can in fact get it. So, while communicating the
data in the network, no one can understood
except intended recipient.
transmission, aims to protect the collected data
during its transferring from nodes to the base
station. It has two main steps: data aggregation and
data routing to base- station. Data aggregation is the
process of transmitting data from nodes to the CH
inside the cluster. Then CHs forward the data to the
base station through a specific path known as
• Availability: Availability requires that WSN
assets, i.e., data, are available to authorized
parties, i.e. CH and base station, at appropriate
time and not prevented through this time. It is a
requirement intended to assure that WSN work
promptly and service is not denied to authorize
routing process. Finally, the base station receives
parties
when
they
request
them.
So,
with
the data and extracts the meaning, and then the
process will start again as shown at Figure 2. To
achieve secure clustering, these steps shall be
enforced. In this paper, we evaluate the existing
availability services of a network should be
available always even in presence of an
internal or external attacks.
secure clustering methods and show to what extend
each method is. We use S − CH, S − CF , S − DA,
and S − DR to indicate to the four phases
respectively.
• Freshness: Freshness is a central goal which is
violated by replay attacks in which the attacker
retransmits an old message to occupy system
resources or confusing the receiver, i.e., base
station.
Generally,
it
ensures
that
no
old
messages have been replayed.
In
order
to
evaluate
each
of
the
existing
clustering algorithms from the robustness point of
view, we will use two notations: P −R and A−R to
indicate its work against passive attack and active
attack respectively.
C.
Robustness
A secure clustering algorithm must be as robust
Fig. 2. Secure clustering process consists of four steps:
Secure Cluster Heads Selection, Secure Clusters
Formation, Secure Data Aggregation, and Secure
Routing of Data to the base station. The arrows depict
data flow.
as possible. The degree of robustness is measured
by the count of attacks that the algorithm prevents.
It also depends on the kind of attack, whether it is
active or passive. The previous list of attacks is
used through this paper to evaluate the robustness
B.
Achieving Security Goals
of each of the secure clustering algorithms.
Secure
clustering
algorithm
must
achieve
the
D.
Efficiency
security goals, i.e., integrity, confidentiality,
availability, and freshness to avoid attacks and
threats as much as possible. These goals can be
summarized as the following [19]:
Secure clustering algorithm must take into
consideration the WSN resource limitations, i.e.,
sensor memory size, energy, and computation
powers. That is refers to preventing the complex
• Integrity: Data must not be changed in transit,
and steps must be taken to ensure that data
cannot be altered by unauthorized party. To
insure that data reaches to the intended receiver
security procedures that may decrease the network
lifetime. It must balance between the security issue
and the network performance. This is refers to the
efficiency of the secure clustering algorithm. We
will evaluate efficiency the secure clustering
HƯỚNG DẪN DOWNLOAD TÀI LIỆU

Bước 1:Tại trang tài liệu slideshare.vn bạn muốn tải, click vào nút Download màu xanh lá cây ở phía trên.
Bước 2: Tại liên kết tải về, bạn chọn liên kết để tải File về máy tính. Tại đây sẽ có lựa chọn tải File được lưu trên slideshare.vn
Bước 3: Một thông báo xuất hiện ở phía cuối trình duyệt, hỏi bạn muốn lưu . - Nếu click vào Save, file sẽ được lưu về máy (Quá trình tải file nhanh hay chậm phụ thuộc vào đường truyền internet, dung lượng file bạn muốn tải)
Có nhiều phần mềm hỗ trợ việc download file về máy tính với tốc độ tải file nhanh như: Internet Download Manager (IDM), Free Download Manager, ... Tùy vào sở thích của từng người mà người dùng chọn lựa phần mềm hỗ trợ download cho máy tính của mình  
9 lần xem

Recent advances of secure clustering protocols in wireless sensor networks. In this article, we review the most recent secure clustering protocols in WSNs. We start with a description of the security requirements for WSNs and discuss the existing security schemes. We analyze to what extend they have been applied to the clustering structure of WSNs. Then, we review secure clustering protocols in emerged recent years. Finally, we present a set of criteria which must be applied to build a secure clustering algorithm..

Nội dung

International Journal of Computer Networks and Communications Security VOL. 2, NO. 11, NOVEMBER 2014, 400–413 Available online at: www.ijcncs.org ISSN 2308-9830 Recent Advances of Secure Clustering Protocols in Wireless Sensor Networks MOHAMED ELHOSENY1, HAMDY K. EL-MINIR2, A. M. RIAD3 AND XIAOHUI YUAN4 1, 4 Department of Computer Science and Engineering, University of North Texas, U.S.A. 1, 3 Department of Information Systems, Mansoura University, Egypt 2 Department of Electrical Engineering, Kafr El-Sheikh University, Egypt E-mail: 1mohamed.elhoseny@unt.edu, 2hamdy elminir@eng.kfs.edu.eg, 3amriad2000@yahoo.com, 4xiaohui.yuan@unt.edu ABSTRACT Wireless Sensor Networks (WSNs) have been employed in many real-world applications that greatly improve our life. The ubiquitous WSNs make security a prime issue, and new technologies have been developed recently. In this article, we review the most recent secure clustering protocols in WSNs. We start with a description of the security requirements for WSNs and discuss the existing security schemes. We analyze to what extend they have been applied to the clustering structure of WSNs. Then, we review secure clustering protocols in emerged recent years. Finally, we present a set of criteria which must be applied to build a secure clustering algorithm. Keywords: Wireless Sensor Networks, Secure Cluster Formation, Secure Routing, Secure data Aggregation, Security Attacks. 1 INTRODUCTION Wireless Sensor Network (WSN) consists of sensor devices transpired in tangible insecure environments in order to collect data. There have been many applications in health care monitoring, environmental monitoring, industrial logging, etc. The data collected could be sensitive and relevant to privacy, which makes security a prime issue [1], [2]. Unlike conventional networked devices, factors such as open communication medium, limited computational capabilities of nodes, and the disadvantages of bandwidth constraint make WSN more susceptible to malicious attacks [3]. To increase network life and reduce energy consumption, cluster model was proposed [4]. In this model the energy of sensor nodes are reserved by involving them in multi-hop communication within a particular cluster and performing data assembling and fusion as shown in Figure 1. Each cluster has a head node that is responsible for gathering data from all nodes within the cluster and sending the aggregated message to the base station. There have been reviews on the security procedures and threats of WSNs [5]–[7], and others discuss the procedures related to the clustering model specifically [8]–[11]. However, most of these reviews evaluated the secure clustering algorithms based on two main processes only, i.e., CH selection and cluster formation [10], [12], [13]. Other works discussed the existing secure routing protocols of clustering model with the aim of protecting the data transmission CHs and the base station [8], [14]–[16]. But for any secure clustering algorithm, a set of criteria must be used to be an effective one. Fig. 1. Clustering Model for Wireless Sensor Network 401 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 These procedures include secure CH building, secure cluster formation, secure data aggregation from cluster members, secure data routing from CHs to the base station, robustness against different types of attack, efficiency in terms of WSN resources limitations, and ability to deal with dynamic clustering environment. For example, the survey paper [10] considers that all selected algorithms that it explains are secure and efficient. However, they do not pay much attention on energy constraints when different security mechanisms are used. This is very crucial because technique based probabilistic and deterministic strategies have different impacts on energy consumption which will affect the network efficiency and performance directly. In addition, the survey [10] did not address the performance requirements study (e.g. memory requirement, computation overhead etc.), which is more important because it is strictly bound to the consumed energy. Based on that, this work not limited to a specific point in secure clustering, it is an attempt to combine all these processes, i.e., secure cluster building and secure data transmission, as evaluation criteria for some secure clustering algorithm. The rest of this paper is organized as follows: section 2 review the common types of attacks for WSN and the source of each one. Section 3 explains the evaluation criteria that is used in this paper to evaluate the secure clustering algorithms. Section 4 discusses the most popular security mechanisms which can be used with the clustering model in WSN. Section 5 reviews some of existing secure clustering algorithms with the strengths and the limitations of each one. After that, section 6 analyzes the discussed secure clustering algorithms using the proposed evaluation criteria. Finally, section 7 summarizes the paper. from inside, outside, or both [5]. Table I lists the common types of attack in WSNs. These attacks aim to affect the transmitted data with one of the following threats [17]: • Interruption: is an attack on the availability of the network. Its main aim is to make an asset of the system, i.e., sensor node, unavailable or unusable. Denial of Service attacks [5] have become very well-known example of interruption. • Interception: is an attack on confidentiality. The sensor network can be compromised by the attacker to gain unauthorized access to sensor node or data store within it. Spoofing attack is a well-known example. • Modification: is an attack on integrity of the system. It this attack unauthorized party not only accesses the data but also modifies the content of a message being transmitted in a network. • Fabrication: is an attack on authentication in which the attacker make an insertion of messages in a network and tries to make it as it is sent from authorized node. • Methods to address WSN security attacks aim at the following aspects [18]: • Preventing Attacks: It aims to prevent any attack before it happens. Any proposed technique will have to defend against the targeted attack. • Detecting Attacks: If an attacker manages to 2 SECURITY ISSUES IN WSNs pass the measures taken by the prevention Security attacks against WSNs can be categorized into two types: active attacks and passive attacks. In passive attacks, attackers are typically hidden and aim to monitor the communication link to collect data. The common examples of passive attacks are eavesdropping, node malfunctioning, node destruction and traffic analysis types. In active attacks, the attacker affects the operations of network, i.e., the transmitted data. For example, the network services may degrade or terminate as a result of these attacks. The common examples of active attacks are Denial-of-Service (DoS), hole attacks, flooding and Sybil types [6]. The source of the attack can come to the network mechanism, the security solution would immediately switch into the detection phase of the counter attack in progress and specifically identify the nodes that are being compromised. • Removing Attacks: It aims to mitigate any attack after it happens by removing the affected nodes and securing the network. 3 EVALUATION CRITERIA FOR SECURE CLUSTERING IN WSNs In this section we discuss the criteria which we will use to evaluate the existing secure clustering method. 402 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 A. Completeness without any alteration, a technique like hash Secure clustering is a sequential process that must guarantee the security goals, i.e. confidentiality, integrity, and availability, in each phase. This process consists of two stages: cluster building and data transmission. The cluster building stage starts with cluster formation in which the cluster heads (CHs) are determined and nodes are assigned to the CHs. The next stage, i.e., data transmission, aims to protect the collected data during its transferring from nodes to the base station. It has two main steps: data aggregation and data routing to base- station. Data aggregation is the process of transmitting data from nodes to the CH inside the cluster. Then CHs forward the data to the base station through a specific path known as routing process. Finally, the base station receives the data and extracts the meaning, and then the process will start again as shown at Figure 2. To achieve secure clustering, these steps shall be enforced. In this paper, we evaluate the existing secure clustering methods and show to what extend each method is. We use S − CH, S − CF , S − DA, and S − DR to indicate to the four phases respectively. function can be used. • Confidentiality: Confidentiality prevents sensitive information from reaching the wrong party, while making sure that the right party can in fact get it. So, while communicating the data in the network, no one can understood except intended recipient. • Availability: Availability requires that WSN assets, i.e., data, are available to authorized parties, i.e. CH and base station, at appropriate time and not prevented through this time. It is a requirement intended to assure that WSN work promptly and service is not denied to authorize parties when they request them. So, with availability services of a network should be available always even in presence of an internal or external attacks. • Freshness: Freshness is a central goal which is violated by replay attacks in which the attacker retransmits an old message to occupy system resources or confusing the receiver, i.e., base station. Generally, it ensures that no old messages have been replayed. In order to evaluate each of the existing clustering algorithms from the robustness point of view, we will use two notations: P −R and A−R to indicate its work against passive attack and active attack respectively. C. Robustness Fig. 2. Secure clustering process consists of four steps: Secure Cluster Heads Selection, Secure Clusters Formation, Secure Data Aggregation, and Secure Routing of Data to the base station. The arrows depict data flow. B. Achieving Security Goals Secure clustering algorithm must achieve the security goals, i.e., integrity, confidentiality, availability, and freshness to avoid attacks and threats as much as possible. These goals can be summarized as the following [19]: • Integrity: Data must not be changed in transit, and steps must be taken to ensure that data cannot be altered by unauthorized party. To insure that data reaches to the intended receiver A secure clustering algorithm must be as robust as possible. The degree of robustness is measured by the count of attacks that the algorithm prevents. It also depends on the kind of attack, whether it is active or passive. The previous list of attacks is used through this paper to evaluate the robustness of each of the secure clustering algorithms. D. Efficiency Secure clustering algorithm must take into consideration the WSN resource limitations, i.e., sensor memory size, energy, and computation powers. That is refers to preventing the complex security procedures that may decrease the network lifetime. It must balance between the security issue and the network performance. This is refers to the efficiency of the secure clustering algorithm. We will evaluate efficiency the secure clustering 403 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 algorithms using three criteria: required memory (M), energy consumption (E), and the processing time (P ). E. Dynamic Clustering Dynamic clustering process aims to reforming the network structure after each round according to the updated status and characteristics of the sensor nodes, i.e., the remaining energy of each sensor. On the other side, the static clustering algorithm allows only the CH change after each round. It forms the network structure to a fixed set of clusters at the initial round and makes it unchangeable until the network become unavailable, i.e., all nodes consume their energy. Therefore, we have to find simple solution that allow securing the dynamic cluster network while consuming as little energy as possible and is adapted to a low computing power. This report discuss the existing schemes for secure clustering according to the previous criteria and proposes a complete security schema for routing data between sensors nodes, CHs, and the base station in cluster-based model for WSN. Table 1: The Common Types of Attacks of WSN and the Source of Each. Code Name Description Active Passive Inside Outside A1 Denial of Service It sends unnecessary packets and utilizes more network √ √ √ bandwidth to prevents the user from accessing the service or A2 Selective It tries to put a malicious node to act as normal node and √ √ Forwarding drop the messages as soon as they receive it. A3 Sinkhole A4 Sybil A5 Wormhole A6 HELLO Flood This attack adds a node to the network to capture all data as √ √ if it was the base station. The malicious node claims multiple identities to be able √ √ communicate with many nodes. This attack records the messages to another location and √ √ √ may retransmit them or a selective part of them. This attack sends the HELLO packet to the nodes, the node √ √ may assumes the attacked device as a neighbor that tries to connect with it. It aims to consume the network resources A7 Spoofed, altered or This is the most direct attack. By spoofing, altering or √ √ √ replayed routingreplaying routing information the attacker can complicate information the network through some actions like create routing loops or generating false error messages. A8 Black-Hole The malicious node communicates the destination node √ √ with false route information to enforce it to send the reply to A9 Node Destruction This attack aims eeither to make the node unavailable to √ √ replace it with a malicious one with the same identifier, or to prevent it from collecting data A10 Monitor andThis attack aims to gather information about the network. √ √ Eavesdropping A11 Traffic Analysis This attack aims to intercept and examine messages in order √ √ to deduce information from patterns in communication. Its danger comes from its ability to work even when the A12 Node Replication This attack ecreates tduplicate nodes and built up various √ √ attacks using them nodes. A13 Message Corruption A14 Jamming This attack performs three main actions: receives message, √ √ √ modifies it to be not understandable, and then forwarding it Jamming interferes with the radio frequencies of the sensor √ √ nodes to make them unavailable. A15 Node Malfunction This attack generates inaccurate part of data that could √ √ expose the integrity of the data-aggregating process at the CH. 404 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 Table 2 shows the notations for the previous criteria that we use to evaluate the secure clustering algorithms. Table 2: The Notations of the Evaluation Criteria for a Secure Routing Protocol Notation Meaning 4 THE EXISTING SCHEMES FOR SECURE S-CH CLUSTERING IN WSN S-CF S-DA Secure Cluster Head Selection Secure Cluster Formation Secure Data Aggregation In order to apply security for clustering model, many security procedures such as the data partitioning, using key management, intruder detection by location or trust management [20] have been proposed. Cryptographic techniques, such as encryption and hashing, are useful in addressing these concerns. However, the use of these schemes greatly increases the energy consumption of sensor nodes and thus shortens their lifetime [21] as they need Key management specially in case of using asymmetric key schema. In addition, most of the traditional key management schemes assume the relationship between nodes is fixed, while clusters as well as the relationship between nodes in hierarchical protocol are dynamic, so these schemes designed for flat networks need modifications to be applied for cluster-based WSNs [22]. Furthermore, in asymmetric key schema a larger sensor memory size is required for key storage. On the other side, Key management scheme (specially symmetric key schema) has two main advantages: it is safer by realization of node-to-node authentication, and it saves energy which is a challenge for any secure protocol [23]. In order to make use of these advantages in clustering model, many dynamic key management techniques were proposed [10], [24]–[27]. In these new schemes a Key is created for each cluster and it will be common among the cluster nodes to guarantee the confidential communication between them. After each round, the cluster key will changed with the changing of the CH. The main problem of these methods is its need for more computation and require more memory size to store the encryption Keys. These requirements affect directly the network lifetime. In the remaining of this section we discuss a list of the existing security solutions, they advantages and their limitations as the following: S-DR Secure Data Routing Ai Attack Identifier, i.e., A1 means DOS M The required memory size E The energy consumption ratio P The required processing time. D Dynamic Clustering S Static Clustering A. Data Partitioning/ Multi-path Routing In this type of security schemes, the aim is to divide the information into several parts. If a sensor tries to send information, it cuts the data into several packets of fixed size. Each packet is sent on a different route. Packets pass in different nodes. When the packets are received by the sink, it brings them together to regenerate the original message. The main advantage of this method is that: the attacker has to catch all packets of a message if it wants to know the information. In order to do it, it has to be able to listen the entire network. It is more complicated for an attacker to have the information. On the other hand, this solution requires additional computations to collect the different packets to regenerate the message. In addition, it is not suitable for all cases of clustering model. It is also appropriate to the multi-hope clustering model in which a CH communicates with the base station through another CH. In most cases, data partitioning requires an additional security mechanism, i.e., cryptography, to protect the packets during transmission. B. Hashing Hash functions have a very simple purpose, they take a long message and generate a unique output value (called message digest) derived from the content of the message. Message digest can be generated by the sender and transmitted with the 405 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 message to the receiver which uses the same hash function to recompute the digest. We can exploit the unique properties of hash function as: the input can be of any length, the output has a fixed length, the hash function is one-way, and the hash function is collision free to prevent the active attack that modifies and retransmits the message. In addition, most hash functions produce a 128-bit message digest which represents a solution of the memory size of the sensor nodes. decrypts information to re-encrypt with the key corresponding to the following receiver. This solution increases considerably the security of the network, because if an attacker discovers a key, this key is just able to communicate with two nodes, and limits the power of this attack. The attacker has to find all pair wise key to listen the entire network. However, this technique is not energy-efficient especially in C. Cryptography time of calculation, since each pair of nodes which transmits information has to encrypt and Due to the resource constraints of wireless sensors, public- key based cryptographic algorithms, i.e., RSA, are too complicated and energy-consuming for WSNs. However the symmetric cryptographic technique has its own qualities that always make it favorable as compared to public key cryptography for WSNs [28]. As a result, most of cryptography solutions in WSN use symmetric key for securing the network, which are more adapted, quicker to perform, and not consume more energy. Although the cryptography allows us to secure the confidentiality of data, its main problem is the key distribution, and we need to find an appropriate key management schema for the network. According to [11], there are four types of key management techniques which can be used: decrypt a message. The lifetime of the network and its rate is going to be reduced. So, we think it may be inefficient solution in case of clustering model because it will consume more energy from the CH in order to decrypt all messages from all sensors inside the cluster. Also, it requires additional memory size for the cluster head to store all keys of all nodes which will be impossible in case of dynamic model. 3. Pair wise key group: Each group or cluster has a key to communicate between nodes in the cluster. This solution offers a compromise between security and energy efficiency. It may limit the number of encryption in communications. However it increases the 1. Global key: In this method, one key is shared by the entire network. To send a message, information is encrypted with this key. Once the message is received, it can be decrypted with the same key. This solution is an energy-efficient solution of cryptography. The information is encrypted once by the sender and decrypted only once by the receiver. However, its the solution with a limited security. If an attacker could find the key, he is able to hear the entire network which communicates with this unique key. To know this key also allows the possibility to insert a malicious node in the network. 2. Pair wise key node: Each node has a different key to communicate with a neighboring node which shares this key. So if one node has ”n” neighbors, it has ”n” key stored to communicate with its neighbors. In this solution, a node that sends a message has to encrypt the message with key neighbor who receives the information. The neighboring work of clusters heads, which have to decrypt and encrypt the information. To be effective, we have to ensure that CHs change regularly in order not to consume all the energy of the CH. The main advantage of this method is that it can be applied to the dynamic clustering model. 4. Individual key: In this solution, each node has its own key to encrypt data. This key is only known by the sink. As a consequence, a message sent by this node goes around hidden on the network until it reaches the sink. This solution is one of the better way to limit the consumption of the network. Nevertheless, this solution secures only communication between a node and the sink. In cluster model, this technique may consume the CH energy rapidly in case of many malicious node attached themselves to the cluster and sent unwanted messages to the CH. In such case, the CH will forward the data automatically to the base station without know its meaning. However, if 406 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 we find a method to guarantee that the CH will know the source of the message, i.e. we can use the Node Coordinates as an identifier; this method can be used and represents a good solution. D. Generation Another key distribution solution is to use a key generation. Each round or generation, the sink sends a new key to the whole network. This key is used as a certificate for each node, to prove it belongs to the network. If an unidentified node tries to come into the wireless sensor network and if it does not have this key generation, the network will refuse its integration. Another benefit of this technique is that it limits substitution attacks of a sensor and the reprogramming of the sensor to be reused in the network. This technique is energy-efficient and easy to apply. However it directed only closed networks, which cannot accept new nodes. Moreover, there is the problem of a node, which cannot receive a key to progress time. E. Localization The work of this method is to use a technique for locating a node. For this solution, the wireless sensor network needs specific sensors called beacon node, which are sensors that knowing their geographical position. For example they can use a GPS equipment. The problem is that it cannot work on any other type of sensors. F. Intrusion Detection System (IDS) Intrusion is an unauthorized (unwanted) activity in a net- work that is either achieved passively (e.g., information gathering, eavesdropping) or actively (e.g., harmful packet forwarding, packet dropping, hole attacks). In a security system, if the first line of defense, Intrusion Prevention, does not prevent intrusions, then the second line of defense, Intrusion Detection, comes into play. It is the detection of any suspicious behavior in a network performed by the network members [6]. An IDS is also referred to as a second line of defense, which is used for intrusion detection only; that is, IDS can detect attacks but cannot prevent or respond. Once the attack is detected, the IDSs raise an alarm to inform the controller to take action [7]. election have gained modest attention so far. On the one hand, there are many papers that survey the security solutions applied in wireless sensor networks, e.g. [6], [8], [11], [17], [26], [29], [30]. These papers detail the common security issues in sensor networks, like authentication, intrusion detection, secure routing, secure data aggregation, etc. However, none of these papers address the issue of secure building and data transmission in particular. On the other side, some papers, e.g. [31], [32], tackle the problem of secure clustering and secure CH election in sensor networks focusing on issues like dynamic key change, complexity, cluster head election criteria, and so on. Regrettably, the latter papers do not consider the security routing aspects of clustering [29]. In this section, we focus on the existing secure clustering algorithms for WSN as general to evaluate them according to the proposed criteria. 5.1 SLEACH SLEACH protocol is the first attempt to build a secure version of the well-known LEACH protocol. It is prevents sinkhole, selective forwarding and HELLO flooding attacks. SLEACH prevents an intruder node to send falsified data messages. But it doesn’t guarantee confidentiality and availability. This algorithm works with homogeneous WSNs in which all nodes have the same characteristics, i.e., initial energy, and processing power. This algorithm makes use of cryptography as the security mechanism by using symmetric-key methods. It can protect the network from outsider attack but it decreased the network efficiency and performance. 5.2 SS-LEACH SS-LEACH [33] is another protocol that offers security while being energy efficient. For that, it works with multi- path CHs to communicate with the base station. To ensure security, it employs key pre-distribution and self-localization techniques. SS-LEACH is protected from selective forwarding, Hello flooding and Sybil attacks, but it controls neither data integrity nor freshness [26]. SSLEACH improves the network efficiency by improving the method of selecting CHs and forms dynamic multi-paths CHs chains to transfer data to the base station. 5 SECURE CLUSTERING ALGORITHMS 5.3 ESODR In all clustering methods, security and reliability aspects of clustering and cluster head In ESODR [34] method, each cluster is made up of a CH and multiple gateways (GWs) and other 407 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 cluster members. ESODR combines hash function, symmetric key cryptographic algorithm, and public key cryptographic algorithm together. In ESODR, the computational complexity is low and has got good efficiency and scalability but it suffer from the dynamic clustering nature of the network. In addition, it requires more memory size to store both the encryption key and the hash digest. 5.4 SecLEACH SecLEACH [24] is an improvement of SLEACH. It is a protocol for securing node-to-node communication in LEACH- based networks. It introduced symmetric key and one-way hash chain to provide different performance numbers on efficiency and security depending on its various parameter values. Although it provides authenticity, confidentiality, integrity and freshness for node-to-node communication, SecLEACH did not provide a solution for the compromised CH attack. This is because SecLEACH is vulnerable to key collision attacks and do not provide full connectivity. 5.5 RLEACH RLEACH protocol attempts to apply random pairwise key (RPK) scheme onto LEACH. AS in LEACH, RLEACH operation is round based. It has three basic phases: shared-key discovery phase, cluster set-up phase and data transmission phase. RLEACH has the ability to resist to several attacks such as selective forwarding, sybil and hello flooding. Nevertheless, it is possible that an insider exercises sinkhole attack to be CH. Compromised node can also corrupt BS by the falsified data messages it sends [26]. 5.6 ORLEACH The same idea of RLEACH was applied by adding IDS mechanism as a new phase and produced a new method called ORLEACH [4]. ORLEACH operation is, therefore, divided into the following phases: Shared-key discovery phase, Cluster set-up phase, isolation of previously detected, attackers and MNs selection, Data transmission phase and Intrusion detection and alerting phase. Although this algorithm solved the problems of RLEACH specially whose are related to the active attacks, it is complexity increases the processing time and the consumed energy of the network which directly affect its efficiency. 5.7 NSKM NSKM [35] is a secure clustering method that tries to solve the problems related to key management. It provides an efficient key distribution and establishment way by using three categories of keys; pre-deployed keys, network generated keys and the BS broadcasted keys. It works well against replay and node capture attacks. The selection of CH among nodes is based on its location and its distance to base station. NSKM also ensures that the whole network is never compromised even if there has been an attack in the network by providing a secure data routing from CHs to the base station. Its main problem is it cannot work with dynamic clustering environment and suffers from active attacks, i.e., sinkhole and wormhole. 5.8 EECBKM EECBKM [28] is a cluster based technique for key management which the clusters are formed in the network and the CHs are selected based on the energy cost, coverage and processing capacity. An EBS key set is assigned by the base station to every CH and cluster key to every cluster this proposed technique reduces node-capture attacks and efficiently increases packet delivery ratio with reduced energy consumption. But the problem of this protocol is that it works well in the environment with low density of sensors. In addition, it suffers many kinds of active attack. Another method is the SAC which is successful in preventing attacks caused by adversary like hello flooding and provides resilience to sensor nodes captured by adversary [22]. PIKE uses probabilistic techniques to establish pair wise keys between neighboring nodes in the network. However, in this approach, each node has to store a large number of keys. 5.9 SCMRP Another secure clustering algorithm is SCMRP [36] which is based on multipath technique. SCMRP collects the benefits of both cluster based routing and multipath routing. It provides security against various attacks like altering the routing information, selective forwarding attack, sinkhole attack, wormhole attack, Sybil attack etc. In addition, it uses cryptography as a security mechanism to protect message after portioning it to packets. SCMRP consists of five phase; neighbor detection and topology construction, pairwise key distribution, cluster formation, data transmission, and re-clustering and rerouting. The Base station collects all the neighbor list from sensor node and apply an algorithm called DFS for finding multiple path. The BS generates the pairwise key and unicast to all nodes. The CH 408 M. Elhoseny et. al / International Journal of Computer Networks and Communications Security, 2 (11), November 2014 selection is based on the remaining energy of the node. 5.10 SHEER SHEER [37] aims to create a secure clustering schema with energy-efficient and secure communication on the network layer. SHEER uses the cryptography as the security mechanism. It proposed a schema for key distribution based on the Hierarchical Key Establishment System (HKES). SHEER proposed also a probabilistic transmission mechanism to re- duce energy consumption and extend the network lifetime. This method works effectively against HELLO flood attack, sybil attack and sinkhole attack. Its main drawback is that it is not able to protect the network from selective forwarding attacks.