Hindawi Publishing Corporation

International Journal of Distributed Sensor Networks

Volume 2015, Article ID 691498, 7 pages
http://dx.doi.org/10.1155/2015/691498

Research Article
An Efficient Key Management Scheme Based on ECC and AVL
Tree for Large Scale Wireless Sensor Networks

Zhongyuan Qin,1 Xinshuai Zhang,1 Kerong Feng,1 Qunfang Zhang,2 and Jie Huang1
1
School of Information Science and Engineering, Southeast University, Nanjing 210096, China
2
Computer Department, Nanjing Institute of Artillery Corps, Nanjing 211132, China

Correspondence should be addressed to Zhongyuan Qin; [email protected]

Received 22 May 2015; Revised 17 August 2015; Accepted 20 August 2015

Academic Editor: Shusen Yang

Copyright © 2015 Zhongyuan Qin et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

With the wide adoption of wireless sensor network (WSN), security problems emerge as a challenging issue because of the limited
computational power and energy of the sensor nodes. Furthermore, the deployment of WSN in the hostile area with a large
number of nodes also poses a threat. In this paper, we proposed a new efficient key management scheme based on Elliptic Curve
Cryptography (ECC) and AVL tree for large scale WSNs. In our scheme the Elliptic Curve Paillier Encryption (ECPE) cryptography
is adopted for communication and the AVL tree is used to store the neighbors’ ID and public key. The number of keys used in our
scheme is smaller than the proposed schemes as we store these keys using AVL tree to shorten the search time. Regular key updates
are also designed to further improve the security of the whole network. Our scheme has a good scalability where the node addition
and deletion are supported. Experimental results and analysis show our scheme can significantly reduce the energy consumed by
the node and reduce the memory and computational overhead.

1. Introduction than the flat homogeneous model. One of the most famous
hierarchical algorithms is Low-Energy Adaptive Clustering
Nowadays, wireless sensor networks (WSNs) have a critical
Hierarchy (LEACH) [4, 5], which was proposed by Chan-
application in military, medical, and industrial sectors. WSNs
drakasan to balance the energy consuming among nodes
consist of a large number of tiny, cheap sensor nodes which
are computational and energy-constrained. The security in by randomly rotating cluster head memberships among all
WSNs is gaining importance as a large number of nodes nodes.
may be exposed in hostile environments. If only one node is In this paper, we propose an efficient key management
captured by the attacker, the whole network can be compro- scheme built on LEACH protocol. Furthermore, Elliptic
mised. Curve Paillier Encryption (ECPE) algorithm rather than
Because of the wireless connectivity, the absence of the general public key algorithms is adopted to achieve the key
physical protection, and the unattended deployment, the agreement with fewer keys and meet the needs of storage
security in WSNs becomes a challenging research hotspot. while catering to computing limited sensor nodes. We also
Key management protocols are the core of the secure commu- use AVL tree (Georgy Adelson-Velsky and Evgenii Landis’
nications [1–3]. The goal of the key management in WSNs is tree, named after the inventors) [6], which is a self-balancing
to establish secure links between neighbor sensors at network binary search tree, to store the nodes’ ID and public keys
formation phase. which can greatly reduce the key search time and thus reduce
Basically, there are two architectures available for WSNs. the energy consumption. Combined with the energy-saving
One is a distributed flat architecture and the other is a hierar- property of LEACH, our scheme can significantly reduce
chical architecture. Considering the limitations of WSNs, the the energy consumed by the node and extend the network
hierarchical network model has more operational advantages lifetime.
2 International Journal of Distributed Sensor Networks

The contributions of this paper are as follows: they did not take into account security issues confronted
by each node when the share transfers in secret. A key
(1) We are the first to use the AVL tree to store the management scheme without secure channel was proposed
neighbors’ ID and public keys; thus less search time in [13], which is a combination of the discrete logarithm
and computing energy are consumed in large scale problem on elliptic curve and the threshold key theory, but
WSN. has left out the security issues that arise in the exchange of the
(2) A combined scheme including LEACH and ECPE is secret share between nodes. In [14], the network was divided
proposed in which fewer keys are needed to be stored into three levels: central authority (CA), server nodes (SN),
in the node. Storage overhead is reduced for each and ordinary nodes (ON). SN generate session private/public
sensor node and the network lifetime is extended. keys for each node in the network, which makes them the
(3) Quantitative memory and computation overhead most vulnerable part in the whole network when captured.
with a particular security analysis are provided. The Rajendiran et al. proposed a key predistribution technique
obtained results show that our scheme can signifi- using the ECC for WSNs. They choose elliptic curve points
cantly reduce the energy consumed by the node and as the key pool to achieve better connectivity. But they still
reduce the memory and computational overhead. cannot resist the node capture attack [15]. Azarderakhsh et
al. proposed a key management for heterogeneous sensor net-
As for the rest of this paper, Section 2 introduces the works using a hybrid technique of public key and symmetric
related works; we give some preliminaries about ECPE key cryptography, while they assume that the CHs cannot be
algorithm and AVL tree in Section 3; Section 4 describes the captured and nodes know their own location [16]. In [17],
scheme we proposed in detail; the performance analysis to the VEGK is proposed for key management in heterogeneous
proposed scheme is elaborated in Section 5; finally this paper cluster based WSNs using hybrid key management technique
is concluded in Section 6. between public key cryptography ECC and pairwise symmet-
ric keys.
2. Related Works Altogether, state-of-the-art public key cryptography
schemes do not meet the strict limitation of resource-con-
Key management problems in WSNs have been extensively strained sensor nodes, because a large computational over-
studied in the literature and various solutions have been head is introduced especially with large scale networks,
presented. In this work, we mainly classify these schemes which is the common case for WSNs. In this paper, we
into two categories: symmetric schemes and asymmetric proposed a novel efficient key management scheme based on
ones. In symmetric key schemes, a preinstalled system-wide ECPE public key cryptography for WSNs. We also use AVL
symmetric key or pairwise keys are usually stored on the tree to store the ID and public key in our scheme so as to
devices. Symmetric key schemes are used in most of WSNs shorten the search time. Meanwhile, we design regular key
because they consume less computation time. Asymmetric updates to further improve the security of the network and
key schemes adopt public key technology, such as Elliptic our scheme also provides perfect scalability to allow nodes
Curve Cryptography (ECC), to realize key distribution. addition and removal.
Though public key technology was thought to be too com- It must be stated that in [18] the AVL tree is also used to
putationally expensive for WSNs, recent studies [7, 8] have store the ECC keys, but they store the keys in a whole WSN.
successfully implemented it in wireless sensor networks. In their method, each sensor node is associated with a leaf,
Since there are lots of symmetric key management and all keys located along the path from the leaf to the root of
schemes in WSNs, we will not give them a review. Useful sur-
the AVL tree belong to that node. If two nodes need a shared
veys on symmetric key management schemes can be found in
key, they find the common ECC key which is at the highest
[3, 9, 10].
level and closest to the leaves in the AVL tree. However, in our
The key management schemes based on public key
cryptography (PKC) are convenient for WSNs compared with scheme, the AVL tree is used to store the neighbors’ public
those based on symmetric cryptographic algorithm. Usually keys in one node so as to reduce search time. When there
the PKC schemes are considered resource-rich so that they are many neighbors in a large scale WSN, there is no need to
are not suitable for WSN; but recently many researchers generate the shared communication key in our scheme with
put their eyes on the adoption of PKC in WSN. Wandert the AVL tree available.
et al. made a comparison between RSA (the Rivest-Shamir-
Adleman cryptosystem, named after the inventors) and ECC 3. Preliminaries
in the nodes of WSNs and gave the conclusion that ECC
behaves better on storage, computation, and communication 3.1. LEACH. LEACH was proposed by Heinzelman et al. to
overhead [11]. Ren et al. come up with several effective PKC balance the energy consumption among nodes by randomly
based schemes upon the integration of several cryptographic rotating cluster head memberships among all nodes [19], and
techniques, including the Bloom filter, the partial message- it can extend the network lifetime by 15%. The operation of
recovery signature scheme, and the Merkle hash tree [12]. LEACH is broken up into rounds. Every round of LEACH has
In terms of the elliptic curve discrete logarithm problem two phases: one is the setup phase, where nodes elect CH, and
difficulty, they propose a key management scheme without the other is the steady state phase, where nodes communicate
secure channel by adopting the key threshold theory, but with their CHs. During the setup phase, node 𝑛 chooses a
International Journal of Distributed Sensor Networks 3

random number between 0 and 1. If the number is less than

54
the threshold 𝑇(𝑛), the node becomes a cluster head for the
current round. The threshold is set as
𝑃 17 89
{ if 𝑛 ∈ 𝐺
𝑇 (𝑛) = { 1 − 𝑃 ∗ {𝑟 mod (1/𝑃)} (1)
{0 otherwise,
4 28 68 96
where 𝑃 is the desirable percentage to be CH, 𝑟 is the current
round, and 𝐺 is a set of nodes that have not been selected as
CH in previous rounds. 16 33 71 84

3.2. Elliptic Curve Paillier Encryption (ECPE). ECPE [20] Figure 1: An example of the AVL tree.
was first proposed by Paillier in 2000. It is a probabilistic
encryption scheme employing elliptic curves over rings based
on the use of twists of anomalous curves.
It is known that curves 𝐸𝑝 (𝑎, 𝑏) over 𝐹𝑝 which have trace are not needed in our model, which makes it more practical
of Frobenius one (they are said to be anomalous) present the for WSNs. We assume the node in our network is capable
property that computing discrete logarithms on them is very of running the ECPE algorithm. The notations used in this
easy [21]. Okamato extended this property to 𝑝-subgroup of paper are listed in “Notations.”
𝐸𝑝2 (𝑎, 𝑏) so that the projection onto 𝐹𝑝 gives the twist of an Our scheme can be divided into two phases: (1) network
anomalous curve. formation phase, where nodes elect CHs and generate the
The details of this cryptosystem are as follows. session key with each other, and (2) network steady state
phase, where nodes communicate with each other with the
Initialization. 𝑛 = 𝑝𝑞 is generated which is a product of two session key. Node addition, node deletion, and key updating
large primes, 𝑝 and 𝑞, while 𝑝 ≡ 𝑞 ≡ 2 (mod 3). Choose an are also allowed during the second phase.
elliptic curve 𝐸𝑛2 (𝑎, 𝑏) and a base point 𝐺 ∈ 𝐸𝑛2 on this curve.
The details of our scheme are as follows.
Public Key. Consider 𝑃𝑈 = (𝑛, 𝐸𝑛2 , 𝐺).
4.1. Network Formation Phase
Private Key. 𝑠 = lcm(𝑝 + 2, 𝑞 + 2) is the lowest common
multiple of 𝑝 + 2 and 𝑞 + 2. 4.1.1. Key Predeployment. Following the rules of ECPE, firstly,
base station chooses a large integer 𝑄 = 𝑝𝑞 as a product of
Encryption. Suppose the plaintext is 𝑀 ∈ 𝑍𝑛 , pick random two odd primes, 𝑝 and 𝑞, while 𝑝 ≡ 𝑞 ≡ 2 (mod 3). Then
𝑟 < 𝑛, and return the ciphertext: 𝐶 = (𝑀 + 𝑛𝑟) ⋅ 𝐺. base station chooses a safe elliptic curve 𝐸𝑛2 and a point 𝐺 on
this curve. Afterwards base station calculates the parameters
Decryption. Compute 𝑀 = Ψ𝑛 (𝑠 ⋅ 𝐶)/Ψ𝑛 (𝑠 ⋅ 𝐺) mod 𝑛, where of ECPE for the nodes: the public key 𝑄 = 𝑝𝑞, the private
Ψ𝑛 (𝑥, 𝑦) = −(𝑥/𝑦) mod 𝑛2 . key 𝑆 = lcm(𝑝 + 2, 𝑞 + 2), the identity of the node ID, and
an initial shared symmetric key ikey. Finally, the base station
3.3. The AVL Tree. An AVL tree is a self-balancing binary stores these parameters, that is, 𝑄𝑖 , 𝑆𝑖 , ID𝑖 , and ikey, into the
search tree. For each node of the tree, the height difference of nodes.
its subtrees is at most 1; therefore, it is also height-balanced.
𝑂(log 𝑛) time is needed for search, insertion, and deletion 4.1.2. Cluster Head Election. During this phase, all nodes are
operations in both the average and worst cases. able to compete for the CHs for the current round using (1).
Figure 1 shows an example of the AVL tree. The values of The elected CHs send their own ID and public key to the base
left subtree are always smaller than the ones of the root node, station to get registered. The base station makes an AVL tree
while the values of the right subtree are always larger than the list of all the CHs’ public keys and IDs and sends this list to the
ones of the root node. In order to find a particular element in nodes encrypted by the ikey. This phase is shown as follows:
the tree, for example, 28, we firstly compare the element with
the root node 54. We turn left because 28 is smaller than 54, ID𝑖 ‖𝑄𝑖
and then we find 28 larger than 17 so we turn right and find CH𝑖 󳨀󳨀󳨀󳨀󳨀→ BS,
(2)
28. 𝐸ikey (AVLCH )
BS 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ All Nodes.
4. The Proposed Scheme
4.1.3. Cluster Formation. After the last step, each node gets
Our scheme is based on the network model of LEACH which the list of the CHs’ ID and the public key. In order to form the
includes a base station and a large number of sensor nodes. clusters, each CH broadcasts its own ID and public key to its
The base station is assumed to be trusted and capable of neighbors. The neighboring nodes can verify their identifiers
computation and has storage ability. Special cluster heads according to the list received from the base station. Once
4 International Journal of Distributed Sensor Networks

4.2. Network Steady State Phase

ID 4
4.2.1. New Node Addition. It is essential to add new nodes due
to the node energy depletion or capture. The new node has to
ID 2 ID 5 be registered at BS before being added to the network and BS
will store the current CH list to it. After the deployment of the
··· new node, it sends a “join-in” request message including its
ID1 ID 3 ID and public key encrypted by its private key to the nearby
···
CH. The CH will reencrypt the message with its own private
key and then send it to BS. BS verifies the identity of the new
Figure 2: The AVL tree formation process. node and replies with a confirming message to CH. Then CH
broadcasts the new node’s ID and public key to the nodes in
its cluster. Now the new node can exchange public keys with
its neighbor nodes and negotiate session keys with them. This
a sensor node 𝑗 decides to join cluster 𝑖, it replies its ID and above process is shown as follows:
public key to CH𝑖 . CH stores the nodes’ ID and public key in
𝐸𝑄new (AVLCH )
a form of AVL tree. This phase is shown as follows: BS 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ Nodenew ,
ID𝑖 ‖𝑄𝑖
CH𝑖 󳨀󳨀󳨀󳨀󳨀→ Neighbor Nodes, 𝐸𝑆new (IDnew ‖𝑄new )
Nodenew 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ CH𝑖 ,
(3)
𝐸𝑄𝑖 (ID𝑗 ‖𝑄𝑗 ) 𝐸𝑆𝑖 (𝐸𝑆new (IDnew ‖𝑄new ))
Node𝑗 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ CH𝑖 . CH𝑖 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ BS, (6)

𝐸𝑄𝑖 (𝑌/𝑁)
4.1.4. The AVL Tree Formation of Nodes’ Information. After BS 󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ CH𝑖 ,
the clusters are formed, each node broadcasts its own ID
𝐸𝑄 (IDnew ‖𝑄new )
and public key to its neighboring nodes. The node gathers its CH𝑖 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ ALL Nodes.
neighbors’ IDs and public keys and then stores them in AVL
tree. The creation of AVL tree is shown in Figure 2; the IDs 4.2.2. Node Deletion. If a sensor node does not send data
of five nodes in this AVL tree are listed from small to large: for a long time, it will be considered as a dead node and
ID1 < ID2 < ID3 < ID4 < ID5 . We put smaller nodes in left will be removed from the network. CH broadcasts a message
subtrees and larger nodes in right subtrees. Then the rotation including the node’s ID and public key to all nodes in this
operation of AVL tree is used to keep the balance so that the cluster. CH and the live node will update their AVL tree at
height difference of the subtrees is 1 at most. Consider the same time.
𝐸ikey (ID𝑗 ‖𝑄𝑗 )
Node𝑗 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ Nodeneighbor . (4)
4.2.3. Key Update. In LEACH algorithm reclustering is
needed after a certain period of time. New CH will be
4.1.5. The Session Key Agreement between Nodes. Suppose elected according to formula (1). With the establishment of
node 𝐴 wants to communicate with node 𝐵; then node 𝐴 the new clusters, the nodes will update the session key with
needs to generate a session key with node 𝐵. Firstly, 𝐴 queries its neighbors. Therefore, the energy consumption can be
𝐵’s public key in its AVL tree and then generates a random balanced, which will extend the lifetime of the whole network.
number sKey as the session key. 𝐴 then encrypts sKey and Meanwhile security is also enhanced by the updated session
the timestamp 𝑇1 with its own private key and reencrypts key.
the ciphertext plus 𝐴’s public key and ID using 𝐵’s public
key. After 𝐵 decrypts the message, 𝐵 verifies the legitimacy 5. Performance Evaluation
of 𝐴’s public key using its AVL tree. If authentication passed
𝐵 uses 𝐴’s public key to obtain the session key sKey and the 5.1. Security Analysis. Prior to the deployment of sensor
timestamp and checks the timestamp 𝑇1 to prevent the replay nodes in our scheme, a shared key is preset to encrypt
attack. 𝐵 then encrypts sKey and the timestamp 𝑇2 with its the information exchanged during the network formation
own private key and reencrypts them together with 𝐵’s public phase; therefore the adversary is unable to acquire the nodes’
key and ID using 𝐴’s public key. 𝐴 takes similar procedure to identity. What is more, each node mutually verifies the other’s
verify the identity of 𝐵. In this way, mutual authentication is identity by AVL tree when they try to agree on a session key.
finished and a session key is generated. The above process is Hence it is impossible for the adversary to launch identity
shown as follows: related attacks, for instance, the Sybil attack. The usage of
ECPE cryptography makes our network more secure than
𝐸𝑄𝐵 (𝐸𝑆𝐴 (sKey‖𝑇1 )‖ID𝐴 ‖𝑄𝐴 )
Node𝐴 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ Node𝐵 , either symmetric cryptography or the traditional public key
(5) cryptography like RSA. Public key cryptography prevents
𝐸𝑄𝐴 (𝐸𝑆𝐵 (sKey‖𝑇2 )‖ID𝐵 ‖𝑄𝐵 ) plenty of frequent attacks on the network like selective
Node𝐵 󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀󳨀→ Node𝐴. forwarding, Flooding, and Sinkhole attack. In addition,
International Journal of Distributed Sensor Networks 5

Table 1: Comparison of memory overhead. Table 2: Comparison of computational overhead (ECC operations).

Rajendiran Azarderakhsh El-Din et Azarderakhsh El-Din et Ours (times)

Ours
et al.’s et al.’s al.’s et al. (times) al. (times)
Storage needed Cluster head 100 40 10
52480 144480 8000 16000
for CH (bit) Sensor node 3 5 5
Storage needed
for sensor 52480 16320 7040 5120
node (bit) in CH. In Azarderakhsh et al.’s scheme, the session key
shared between two nodes is distributed by CH. For the
10 nodes cluster, there are altogether 45 keys between each
signature verification can be achieved by the private key two nodes; then CH encrypts the session key with the two
encryption. A timestamp is also adopted in our scheme to
node public keys, respectively, so it needs 90 ECC based
avoid the replay attack.
encryption operations plus the 10 session keys between itself
We use public key cryptography to generate a session
and the 10 nodes which means 100 ECC based encryption
key, which will be updated after a certain period of time,
between two nodes. In our scheme, different pairs of nodes operations in total. In El-Din et al.’s scheme, CH needs 4
share different session keys. Even if the adversary captures a ECC based encryption and decryption operations with one
few sensor nodes he cannot get the session key between other node, which equals 400 ECC operations needed by CH for
nodes and cannot recover the keys used before, so that the 10 nodes. However, in our scheme, only 1 ECC operation is
forward secrecy of the network is ensured. On the contrary, needed for one node and 10 for 10 nodes. Then we compare
in Rajendiran et al.’s scheme, if an adversary compromises the ECC operations for one normal node during the period
one more node every time, he can obtain more information it joins in the cluster and shares a session key with its
about the ECC key pool. When the number of captured nodes neighbor. In Azarderakhsh et al.’s scheme, a node needs 3 ECC
exceeds some threshold, the adversary will be able to recover operations, while 5 are needed for the node in El-Din et al.’s
the entire ECC key pool and thus compromise the whole scheme. In our scheme, the nodes need to authenticate each
network. other through verifying signature, so the node needs 5 ECC
operations too. Comparisons of the number of ECC based
5.2. Memory Overhead. For the convenience of comparison encryption and decryption operations executed by CH and
on memory overhead, we follow the network size of Azarder- nodes are shown in Table 2. We can see from the table that our
akhsh et al.’s [16], 1000 nodes and 10 clusters every round, scheme has superiority in overall computational overhead.
while each node has 31 neighbor nodes. As stated previously, a
sensor node in our scheme only needs to store the public keys 5.4. Energy Consumption. Given the large number of nodes,
of its cluster head and neighbors, and the CH needs to store WSNs usually take a long time and lot of energy to find the
the nodes’ public keys in its cluster. In El-Din et al.’s scheme, public key of a certain node. In our scheme, we use the AVL
CH only stores its neighbors’ keys, which makes it hard for tree to store the data so that search time and the energy cost
the CH to manage the cluster. However, in Azarderakhsh et can also be significantly reduced. For instance, it needs (𝑛 +
al.’s scheme, a node needs to store its neighbors’ public keys 1)/2 operations to search one node among 𝑛 nodes in direct
plus the public keys of all CHs. In Rajendiran et al.’s scheme, search; namely, the time complexity is 𝑂(𝑛). It is only 𝑂(log 𝑛)
they combined ECC with E-G scheme to distribute keys. The when using AVL tree, while the schemes of Rajendiran et
probability of a sensor node sharing one key with its neighbor al., Azarderakhsh et al., and El-Din et al. are direct search.
is 𝑃󸀠 = 1 − {(𝑝 − 𝑟)!/[(𝑝 − 2 × 𝑟)! × 𝑝𝑟 ]}, where 𝑝 is the Figure 3 shows the comparison of search time between the
size of key pool and 𝑟 is the size of key chain. For Rajendiran direct search and the AVL tree. We also implement our
et al.’s scheme, they need 𝑝 = 5000 and a chain of 120 keys proposed scheme in Microsoft Visual C++ 6.0. The operating
to achieve a 95% connectivity, while 328 keys are needed to system is Windows 7 Ultimate. The computer configuration
have 99.999% connectivity. Suppose the length of ECC key is as follows: CPU, Intel Core i5 3.2 GHz; memory, 4 GB; hard
is 160 bits; Table 1 shows the superiority of our proposed disc, 500 GB. We test our AVL tree method in 3 scenarios:
scheme compared to those proposed by El-Din et al. [22], 100 nodes, 2000 nodes, and 10000 nodes. Ten searches are
Azarderakhsh et al. [16], and Rajendiran et al. [15]. executed for every scenario and the average time cost is listed
in Table 3. The result shows that, with the increasing nodes
5.3. Computational Overhead. Our scheme uses along with number, the AVL tree method demonstrates its superiority
El-Din et al.’s and Azarderakhsh et al.’s schemes, ECC based in search time. In other words, the energy consumption of
algorithm for encryption and decryption. Since ECC based searching in our scheme is much less than in the schemes
encryption and decryption are the most energy consuming using direct search method.
operations in the schemes, we make a comparison on the
times of ECC based encryption and decryption to show the 6. Conclusion
computational efficiency. In order to facilitate the compari-
son, we assume a small cluster with one cluster head (CH) In this paper, an efficient key management scheme based on
and 10 normal nodes. We first compare the ECC operations ECPE and AVL tree for large scale wireless sensor networks is
6 International Journal of Distributed Sensor Networks

Table 3: Comparison of average search time. sKey: Shared secret key between two nodes
𝑇: Timestamp.
Rajendiran et
al.’s,
Number of nodes Ours (𝜇s) Azarderakhsh et Conflict of Interests
al.’s, and El-Din
et al.’s (𝜇s) The authors declare that there is no conflict of interests
100 259 4798 regarding the publication of this paper.
2000 271 4929
10000 296 5240 Acknowledgment
This work is supported by the National High Technology
400 Research and Development Program of China (863 program)
350 under Grant no. 2013AA014001.
300
References
Total searches

250
200 [1] S. H. Erfani, H. H. Javadi, and A. M. Rahmani, “A dynamic key
150 management scheme for dynamic wireless sensor networks,”
Security and Communication Networks, vol. 8, no. 6, pp. 1040–
100 1049, 2015.
50 [2] S.-H. Seo, J. Won, S. Sultana, and E. Bertino, “Effective key
management in dynamic wireless sensor networks,” IEEE Trans-
0
0 100 200 300 400 500 600 700 800 900 actions on Information Forensics and Security, vol. 10, no. 2, pp.
The number of nodes 371–383, 2015.
[3] C.-Y. Chen and H.-C. Chao, “A survey of key distribution
Ordinary search
in wireless sensor networks,” Security and Communication
AVL
Networks, vol. 7, no. 12, pp. 2495–2508, 2014.
Figure 3: Search time comparison. [4] S. Cho, L. Han, B. Joo, and S. Han, “P-LEACH: an efficient
cluster-based technique to track mobile sinks in wireless sensor
networks,” International Journal of Distributed Sensor Networks,
vol. 2014, Article ID 803656, 10 pages, 2014.
proposed. Our scheme follows the network model of LEACH.
[5] A. Salim, W. Osamy, and A. M. Khedr, “IBLEACH: intra-
We use ECPE to generate session keys between nodes since
balanced LEACH protocol for wireless sensor networks,” Wire-
ECPE can achieve the same security as normal public key less Networks, vol. 20, no. 6, pp. 1515–1525, 2014.
cryptography but with shorter keys, thus saving storage. In
[6] AVL tree, 2015, https://en.wikipedia.org/wiki/AVL tree.
addition, our approach adopts AVL tree for storing public
keys which can significantly reduce the search time in large [7] S. K. Sahoo and M. N. Sahoo, “An elliptic-curve-based hierar-
scale WSN. Experimental results and theory analysis show chical cluster key management in wireless sensor network,” in
Intelligent Computing, Networking, and Informatics: Proceedings
the memory and computational overhead are reduced and
of the International Conference on Advanced Computing, Net-
the energy consumption is also cut down. Furthermore, our working, and Informatics, India, June 2013, vol. 243 of Advances
scheme has a good scalability that supports the node addition in Intelligent Systems and Computing, pp. 397–408, Springer,
and deletion; key updates also support guaranteeing the New Delhi, India, 2014.
security of WSN.
[8] D. J. Malan, M. Welsh, and M. D. Smith, “Implementing public-
key infrastructure for sensor networks,” ACM Transactions on
Notations Sensor Networks, vol. 4, no. 4, article 22, 2008.
[9] S. Bala, G. Sharma, and A. K. Verma, “A survey and taxonomy
𝑛: Public key of symmetric key management schemes for wireless sensor
𝑢: Private key networks,” in International Information Technology Conference
𝑝, 𝑞: Large prime numbers (CUBE ’12), pp. 585–592, Maharashtra, India, September 2012.
𝐸𝑛2 : A random elliptic curve
[10] A. Selva Reegan and E. Baburaj, “Key management schemes in
𝐺: A point on 𝐸 wireless sensor networks: a survey,” in Proceedings of the IEEE
𝐸𝑘 (⋅): Symmetric encryption with key 𝑘 International Conference on Circuit, Power and Computing Tech-
BS: Base station nologies (ICCPCT ’13), pp. 813–820, IEEE Computer Society,
CH: Cluster head Nagercoil, India, 2013.
Node: Normal node [11] A. S. Wandert, N. Gura, H. Eberle, V. Gupta, and S. C. Shantz,
IDCH : Identity of cluster head CH “Energy analysis of public-key cryptography for wireless sensor
ID𝑖 , ID𝑗 : Identity of node 𝑖 and node 𝑗 networks,” in Proceedings of the 3rd IEEE International Con-
𝑛𝑖 , 𝑢𝑖 : Public key and private key of node 𝑖 ference on Pervasive Computing and Communications (PerCom
AVL𝑖 : The AVL tree stored by node 𝑖 ’05), pp. 324–328, March 2005.
International Journal of Distributed Sensor Networks 7

[12] K. Ren, S. C. Yu, W. J. Lou, and Y. Zhang, “Multi-user broadcast

authentication in wireless sensor networks,” IEEE Transactions
on Vehicular Technology, vol. 58, no. 8, pp. 4554–4564, 2009.
[13] H.-X. Li, L.-J. Pang, and Y.-M. Wang, “Key management
scheme without secure channel for ad hoc networks,” Journal
on Communications, vol. 31, no. 1, pp. 112–117, 2010.
[14] H. Dahshan and J. Irvine, “An elliptic curve distributed key
management for mobile ad hoc networks,” in Proceedings of the
71st IEEE Vehicular Technology Conference (VTC ’10), pp. 1–5,
IEEE, Taipei, Taiwan, May 2010.
[15] K. Rajendiran, R. Sankararajan, and R. Palaniappan, “A secure
key predistribution scheme for WSN using elliptic curve cryp-
tography,” ETRI Journal, vol. 33, no. 5, pp. 791–801, 2011.
[16] R. Azarderakhsh, A. Reyhani-Masoleh, and Z.-E. Abid, “A
key management scheme for cluster based wireless sensor
networks,” in Proceedings of the 5th International Conference on
Embedded and Ubiquitous Computing (EUC ’08), pp. 222–227,
December 2008.
[17] A. E. El-Din, R. A. Ramadan, and M. B. Fayek, “VEGK: virtual
ECC group key for wireless sensor networks,” in Proceedings
of the International Conference on Computing, Networking and
Communications (ICNC ’13), pp. 364–368, IEEE Computer
Society, San Diego, Calif, USA, January 2013.
[18] H. Boumerzoug, B. A. Bensaber, and I. Biskri, “A key man-
agement method based on an AVL tree and ECC cryptogra-
phy for wireless sensor networks,” in Proceedings of the 7th
ACM Symposium on QoS and Security for Wireless and Mobile
Networks Co-Located with MSWiM’11 (Q2SWinet ’11), pp. 57–
62, Association for Computing Machinery, Miami, Fla, USA,
October-November 2011.
[19] W. R. Heinzelman, A. Chandrakasan, and H. Balakrish-
nan, “Energy-efficient communication protocol for wireless
microsensor networks,” in Proceedings of the 33rd Annual
Hawaii International Conference on System Siences, vol. 2,
January 2000.
[20] P. Paillier, “Trapdooring discrete logarithms on elliptic curves
over rings,” in Advances in Cryptology—ASIACRYPT 2000, T.
Okamoto, Ed., vol. 1976 of Lecture Notes in Computer Science,
pp. 573–584, Springer, Berlin, Germany, 2000.
[21] N. P. Smart, “The discrete logarithm problem on elliptic curves
of trace one,” Journal of Cryptology, vol. 12, no. 3, pp. 193–196,
1999.
[22] A. E. El-Din, R. A. Ramadan, and M. B. Fayek, “VEGK: virtual
ECC group key for wireless sensor networks,” in Proceedings
of the International Conference on Computing, Networking and
Communications (ICNC ’13), pp. 364–368, January 2013.
 International Journal of

Rotating
Machinery

International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of

Hindawi Publishing Corporation

World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design
Advances in
OptoElectronics
International Journal of

International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2010
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014

International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014