CC BY-NC-ND
0""'CS Research 4
Available online at www.sciencedirect.com
ScienceDirect
Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Two-phase hybrid cryptography algorithm for wireless
sensor networks
Rawya Rizk *, Yasmin Alkady
Electrical Engineering Department, Port Said University, Port Said, Egypt Received 25 May 2015; received in revised form 25 November 2015; accepted 25 November 2015
6 Abstract
7 For achieving security in wireless sensor networks (WSNs), cryptography plays an important role. In this paper, a new security
8 algorithm using combination of both symmetric and asymmetric cryptographic techniques is proposed to provide high security with
9 minimized key maintenance. It guarantees three cryptographic primitives, integrity, confidentiality and authentication. Elliptical
10 Curve Cryptography (ECC) and Advanced Encryption Standard (AES) are combined to provide encryption. XOR-DUAL RSA
11 algorithm is considered for authentication and Message Digest-5 (MD5) for integrity. The results show that the proposed hybrid
12 algorithm gives better performance in terms of computation time, the size of cipher text, and the energy consumption in WSN. It is
13 also robust against different types of attacks in the case of image encryption.
14 © 2015 Electronics Research Institute (ERI). Production and hosting by Elsevier B.V. This is an open access article under the CC
15 BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
16 Keywords: Advanced Encryption Standard; Cryptography; Elliptic Curve; Message Digest-5; XOR-Dual RSA
1. Introduction
20 21 22
Wireless sensor networks (WSNs) have a great vulnerability due to the broadcast nature and dangerous environment. Correspondingly, there are many solutions for the security issues such as routing security (Fouchal et al., 2014; Hayajneh et al., 2013; Lasla et al., 2014; Farouk et al., 2014), secure localization (Yu et al., 2013), and key management and cryptography (Mary Anita et al., 2015). Cryptographic algorithms are an essential part of the security architecture of WSNs.
WSNs suffer from many constraints such as low battery life and small memory. Due to these limitations, WSN is not able to deal with traditional cryptographic algorithms. Two main problems related to security algorithms arise in WSNs. First, the overload that security algorithms introduce in messages should be reduced at a minimum; every bit
* Corresponding author. Tel.: +20 1009067030. E-mail addresses: r.rizk@eng.psu.edu.eg (R. Rizk), engyasminalkady@yahoo.com (Y. Alkady). Peer review under the responsibility of Electronics Research Institute (ERI).
http://dx.doi.Org/10.1016/j.jesit.2015.11.005
2314-7172/© 2015 Electronics Research Institute (ERI). Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
the sensor sends consumes energy and, consequently, reduces the life of the device. Second, the memory size which refers to size of an encrypted message and the key size should also be reduced (Faye and Myoupo, 2013).
Various cryptographic algorithms have been proposed to achieve the security requirements such as Authentication, Confidentiality, and Integrity. Authentication means preventing unauthorized parties from participating in the network. Confidentiality means keeping information secret from unauthorized parties. Integrity ensures the receiver that the received data is not altered in transit by an adversary. Data authentication can provide data integrity also.
Encryption is the process of encoding information in such a way that hackers cannot read it. There are two types of encryption techniques; symmetric and asymmetric. Symmetric cryptography, also called private-key cryptography uses only one key for encryption and decryption. Common symmetric encryption algorithms include Data Encryption Standard (DES) (Singh and Supriya, 2013) and Advanced Encryption Standard (AES) (Burr, 2003). Asymmetric key cryptography, also called public-key cryptography requires special keys to encrypt and decrypt messages. Common asymmetric encryption algorithms include RSA (Frunza and Asachi, 2007) and Elliptic Curve Cryptography (ECC) (Kodali and Sarma, 2013). ECDSA -Elliptic Curve Digital Signature Algorithm (Balitanas, 2009) and ECDH-Elliptic Curve Diffie Hellman (Johnson et al., 2001) are based on ECC.
Both symmetric and asymmetric cryptographic techniques offer advantages and disadvantages. Symmetric encryption techniques provide cost-effective and efficient methods of securing data without compromising security however; sharing the secret key is a problem. On the other hand, asymmetric techniques solve the problem of distributing the key for encryption however; they are slow compared to symmetric encryption and consume more computer resources. Therefore, the best possible solution for encryption is the complementary use of both symmetric and asymmetric encryption techniques. Hybrid encryption attempts to exploit the advantages of both kinds of techniques while avoiding their disadvantages. Hashing creates a unique, fixed-length signature for a message or data set. It is commonly used to check data integrity. Message Digest-5 (MD5) (Hossain et al., 2012) algorithm is a widely used cryptographic hash function that produces a 128-bit (16-byte) hash value. It has been utilized in a wide variety of security applications.
In this paper, a hybrid cryptography algorithm is proposed and presented. It is designed to provide data security and users authenticity. It includes two phases work at the same time. In Phase I, it takes the advantages of the combination of both symmetric and asymmetric cryptographic techniques using both AES and ECC algorithms. In Phase II, XOR-DUAL RSA is used since it is more robust and cannot be easily attacked. In addition, Hashing is also used for data integrity using MD5 to be ensured that the original text is not being altered in the communication medium. The proposed algorithm has high operation speed, high security performance and strong usability.
The organization of this paper is as follows: Brief overviews of related works of some existing protocols are presented in Section 2. The proposed hybrid encryption algorithm is introduced in Section 3. Sections 4 and 5 present the numerical results and the simulation results of the proposed algorithm in WSNs; respectively. Section 6 presents the implementation of the proposed algorithm in the image protection application. Finally, the main conclusion is presented in Section 7.
2. Related work
To date, many cryptography algorithms have been proposed but a lot of them are not complete suitable for WSNs. In Subasree and Sakthivel (2010), a security algorithm architecture is proposed by Subasree. In this algorithm, the given plain text is encrypted using ECC and the derived cipher text is communicated to the destination through secured channel. Simultaneously, the Hash value is calculated through MD5 for the same plain text, and then encrypted with DUAL RSA. The encrypted message of this hash value is also sent to the destination. In this algorithm, it is difficult to extract the plain text from the cipher text, because the hash value is encrypted with DUAL RSA and the plain text is encrypted with ECC. The new hash value is calculated with MD5 and then it is compared with decrypted hash message for its integrity. By which, it is ensured that either the original text being altered or not in the communication medium. This is the primitive feature of this algorithm however, there are two disadvantages. First, the message is encrypted by asymmetric encryption algorithms (ECC and DUAL RSA) that are slow compared to symmetric encryption. Second, if an attacker determines a person's private key, his or her entire messages can be read.
In Dubal security algorithm architecture (Dubal et al., 2011), the given plain text is encrypted with a key that is generated by ECDH. The encryption algorithm used is DUAL RSA. The derived cipher text is appended with the digital signature for more authentications, generated by the ECDSA algorithm. Simultaneously, the Hash value of this encrypted cipher text is taken through the MD5 algorithm. Then, the generated cipher text and the signature are
HT 621-18
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
communicated to the destination through secured channel. On the other side, i.e., on decryption end, the hash value is first evaluated and integrated. This is compared with the signature, for the verification of the digital signature appended at the end of message. Thereafter, the decryption of cipher text is done by DUAL RSA (Sun et al., 2007). Hence, the plaintext can be derived. In this algorithm, the intruder may be trapped by both the encryption by the DUAL RSA with the key generated by ECDH algorithm and the appended signature. Hence, the message can be communicated to the destination with highly secured manner. However, the used asymmetric encryption algorithms (DUAL RSA and ECDH) are slow compared to symmetric encryption. In addition, the attacker may read the messages if he/she can determine the private key.
A hybrid algorithm architecture is proposed by Kumar (2012). In this algorithm, the given plain text is encrypted first with AES algorithm and then with ECC algorithm. The hash value of this encrypted cipher text is taken through the MD5. On the other side, the Hash value is first evaluated and integrated. Thereafter, the decryption of cipher text is done by AES and ECC decryption algorithms. Hence, the plaintext can be derived. This algorithm is a combination of both symmetric and asymmetric cryptographic techniques. However, the execution time of this algorithm is long because the plaintext is encrypted sequentially by both AES and ECC.
In Ren and Miao (2010), a hybrid algorithm is proposed. In this algorithm, DES algorithm is used for data transmission because of its higher efficiency in block encryption, and RSA algorithm is used for the encryption of the key of the DES because of its management advantages in key cipher. During the process of sending encrypted information, the random number generator uses 64-bit DES session key only once. It encrypts the plaintext to produce the cipher text. On the other hand, the sender gets public key from public key management center, and then uses RSA to encrypt session key. Finally, the combination of the session key from RSA encryption (Boneh and Durfee, 2000) and the cipher text from DES encryption are sent out. This algorithm is considered weak since using DES with RSA affects the security level.
In Zhu (2011), a hybrid algorithm architecture is presented. The plaintext is encrypted with symmetric cipher algorithm, and the key and digital signature belonged to the symmetric encryption algorithm are encrypted with asymmetric key algorithm. The sender encrypts the plaintext with the key belonged to the AES algorithm. To ensure the security of the cipher algorithm and simplify the key management, the sender uses the key only once. The receiver obtains the original information after signature verification. This algorithm suffers from low security level since that the message is encrypted in a single phase which leads to less complexity.
From the previous studies it's shown that the security algorithms that depend on asymmetric encryption algorithms such as Subasree and Dubal have critical weakness points since they are slow compared to symmetric encryption algorithms and consume huge power to encrypt all plaintext by public key. In addition, if an attacker determines a person's private key, his or her entire messages can be read. Although Kumar security algorithm is a hybrid algorithm, it has a critical weakness point. It wastes a lot of time for encryption and decryption processes since it encrypts the plaintext sequentially first by AES and again by ECC. Ren and Zhu security algorithms are suffering from low security level since using DES with RSA in Ren affects the security level and using a single phase in Zhu leads to less complexity. Due to the hard constrains of WSNs, they cannot be able to deal with the majority of such security algorithms.
3. The proposed two-phase hybrid cryptography algorithm (THCA)
In this section, the proposed THCA is presented. It introduces a new method of merging both symmetric and asymmetric techniques by performing two parallel phases. These phases avoid the disadvantages of the existing hybrid algorithms by achieving high security level without increasing the execution time.
3.1. Encryption process
In the Encryption, the plaintext is divided into n blocks Bi. Each block consists of 128 bits. Then, it is divided into two parts mi (0: n/2-1) blocks, and Mi (n/2: n — 1) blocks. If n is not an integer number and has a fraction, THCA algorithm uses padding with null for the last block to be 128 bits. The encryption process is divided into two phases.
In Phase I, The first n/2 blocks are encrypted using (AES and ECC) hybrid encryption algorithm. ECC algorithm is used for protecting secret key since it is the highest secure public key algorithm. Moreover, according to the mathematical problem on which ECC can be solved by fully exponential rather than sub-exponential for other public key systems, ECC needs smaller key size than other algorithms and that refers to less memory size (Kodali and Sarma,
4 R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
2013). It allows the communication nodes to handle a larger number of requests with the smallest number of dropped packet. Since that ECC consumes more power than symmetric algorithm, using AES algorithm reduces the power consumption and raises the system performance (Lenstra, 2001). When using AES with ECC, we are able to save power, and achieve speed up to 25% for encryption and nearly 20% for decryption (Tillich and GroBschadl, 2005). The first n/2 blocks are encrypted as the following:
mi is encrypted using AES by the key ki which is the secret key of AES encryption algorithm with size 128 bits. ki is encrypted by ECC to produce Kj with length L.
i=n/2-1
Mi = Y^ (Bi) 0 <i <n/2 - 1 (1)
Kj = ECCenc(TCPK, ki-!) 0 <j <L - 1 (2)
where ECCenc is Elliptic Curve encryption function. It ciphers the input with trust center public key (TCpK) which is used as a function to authenticate the key.
Ci = Eaes(Kj, Bi) (3)
where E AES is the AES encryption function.
Phase II is performed in parallel of Phase I in order to increase the security level without increasing the execution time. In Phase II, the remaining n/2 blocks are encrypted using XOR-DUAL RSA algorithm. DUAL RSA allows for extremely fast encryption and decryption that is at most four times faster than standard RSA. The XOR Encryption algorithm is a symmetric encryption algorithm that uses the same key for both encryption and decryption. XOR-DUAL RSA algorithm guarantees developing a stronger algorithm, as follows:
Mi = ys (Bi) n/2 <i <n - 1 (4)
In this algorithm, two large prime numbers are chosen randomly; p and q. Then, x=p x q, $(x) = (p - 1) x (q - 1). A number relatively prime to $ is chosen; d. Then, e is calculated such that e x d = 1 mod $(x), and Public key (e, x) is used for encryption.
Ri = (Bi)emod x (5)
ASCII for (Bi) is get and converted to binary
Li = ASCII (Bi) (6)
where Li is a function used to convert message block to ASCII. Ri is a ciphered text using DUAL RSA.
Ci = (Ri) XOR (Li) (7)
MD5 is applied to the cipher texts ci and Ci. It is the best performance of hashing function security (Tillich and GroBschadl, 2005).
di = MD5(ci) (8)
Di = MD5(Ci) (9)
At the final stage of the encryption process, the two n/2 blocks are integrated to generate cipher text of n blocks and it is sent to the sink node. The corresponding hash values (di and A) with size 128 bits for each one are concatenated and sent to the sink node at the same time.
Q = ci + Ci (10)
H = di + Di (11)
The Proposed Encryption Algorithm.
Input: P (Plain text), k (secret key of AES encryption), s (128 bit size of block);
Output: Q (Cipher text), c; (encrypted text using AES with ECC), Ci (encrypted text using XOR DUAL RSA), H (hashing value of cipher text);
1. n = P/s;
2. let i = 0;
3. do{
4. mi = 'i-^2 1(Bi) first part of plaintext;
5. for(j = 0;j <=n -1;j++)
7. Kj = ECCenc (TCpk, ki-1);
9. c; = E AEs(Kj,Bi);
10. di=MD5(ct);
11. i++; 12. }
13. while(i<n/2);
14. i = (n/2)
15. Let p and q two large prime numbers
16. x=p x q
17. t(x) = (p -1) x (q -1)
18. Let d a relatively prime number to tfr
19. e x d = 1 mod t(x)
20. Let (e, x) public key of DUAL RSA.
21. do{
22. Mi = ^2 i_n^2 (Bi) sec ond part of plaintext;
23. Ri = (Bi) e mod x;
24. Li = ASCII (Bi);
25. Ci = (Ri) XOR (Li);
26. Di=MD5(Ci);
27. i++;
29. while(i<n);
30. Q = ci + Ci;
31. H = di +Di;
163 3.2. Decryption process
164 In the decryption, the cipher text Q is divided into n blocks each block consists of 128 bits, Then, it is divided
165 into two parts ci (0: n/2 - 1) blocks and Ci (n/2: n - 1) blocks. Hashing is used in order to identify whether the sink
166 node receives the same cipher text or not. The hash values in both phases are compared. If they are the same, then the
167 algorithm proceeds the decryption process. Else, it discards the message.
168 In the case of the hash values are the same at the source and sink nodes, the first n/2 blocks are decrypted using
169 AES and ECC algorithms as follows:
i=n/2-1
170 ci = Y (Bi) 0 <i < n/2 -1 (12)
ki = ECCdec(TCpk, Kj-1) 0 <i <n/2 - 1
171 (13)
0< j< L- 1
6 R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
The key of AES kj with length L of bits is decrypted by ECC to produce ki which has used to decrypt the cipher text using AES decryption scheme by DAES (AES decryption function).
mi = DAES(Kj,ci)
mi is the first part of the plain text. The remaining n/2 blocks are decrypted using XNOR-DUAL RSA algorithm as follows:
i=n—1
Ci = Y^ (Bi) n/2 <i <n - 1
Private Key (d, p, q) is used for decryption. To make decryption, first some parameters are computed dp = d mod
(p — 1), dq = d mod (q - 1), Rpi = Ri dp mod p, Rqi = Ri dq mod q,
50 = (Rqi — Cpi)p—1mod q
51 = Rpi + S0 P
ASCII for (Ci) is converted to binary. Wi = ASCII (Ci) where Li is a function used to convert block of cipher text to ASCII. Mi = Si XNOR Wi
(16) (17)
Mi is the second part of the plain text. At the final stage of the decryption process, the two n/2 blocks are integrated to produce plain text of n blocks.
P = mi + Mi
The decryption algorithm is described in Algorithm 2.
Algorithm 2.
The Proposed Decryption Algorithm.
Input: Q (Cipher text), H (Hashing value of cipher text), s (128 bit size of block), L (key length), di, Di, K (encrypted key using ECC); Output: P (Plain text);
1. n = C/s;
2. let i = 0;
i=n/2-1
Ci = ^ ^ (Bi) first part of cipher text;
df =MD5(ci); Df = MD5 (Ci);
if(di = d/, )&(Di=Df, )
for(j = 0;j<=L -l;j++)
ki=ECCdec (TCpk, Kj-i);
mi =DAEs(Kj, ci);
10. ll. l2.
18. i = n/2;
while(i < n/2);
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx 7
Algorithm 2 (Continued )
19. Give (d, p, q);
20. dp = d mod (p - 1);
21. dq = d mod (q - 1);
22. do{ i_n-1
23. Ci = ^ ^ (Bi) second part of cipher text;
24. Rpi = Ridp mod p;
25. Rqi = Ridq mod q;
26. Sq = (Rqi - Cpi). p-1 mod q;
27. Si v=Rpi +Sq. P;
28. Wi= ASCII (Ci);
29. Mi = Si XNOR Wi
30. i++;
32. while(i < n);
33. P = mi +Mi;
3.3. Strength of THCA
The strength of any cryptographic algorithm is based on several factors: the computational methods and the used key are two of them. In normal cryptographic approach the intruders may be able to identify cipher text patterns that are transmitted to the destination side. By analyzing the sequence of bit patterns; it is possible for the intruder to identify which type of encryption algorithm is used or they will identify the key used for encryption/decryption process.
In THCA, splitting the plain text improves the strength of the proposed algorithm. The intruder will not be able to identify which type of specific algorithm is applied to generate the cipher text. Thus, it is impossible to decrypt the cipher text. In addition, the two halves of the plain text are encrypted in parallel at the same time which reduces the time of both encryption and decryption.
When mixing AES with ECC in the first half of the plain text, the encryption process is done by symmetric algorithm (AES) which is faster than asymmetric algorithm. The secret key of AES is encrypted by ECC which is more complicated and then more secure. So that we obtain time reduction and power saving that are the advantages of symmetric encryption techniques in addition to the complexity which is the main advantage of asymmetric encryption techniques. Using XOR-DUAL RSA in the second half of the plain text allows our hybrid algorithm to be more robust and cannot be easily attacked.
In addition, Hashing is also used for data integrity using MD5 to be ensured that the original text is not being altered in the communication medium. Then, the proposed algorithm has high operation speed, high security performance and strong usability. So, we can say that THCA is the true meaning of hybrid security algorithm.
4. Numerical results
In this section, the performance of the proposed THCA is measured in terms of size of the cipher text, time of encryption and decryption processes, and time complexity. The proposed algorithm is compared with the already existing algorithms that are presented in Section 2, Subasree (Subasree and Sakthivel, 2010), Dubal (Dubal et al., 2011), Kumar (Kumar, 2012), Ren (Ren and Miao, 2010), and Zhu (Zhu, 2011).
4.1. Size of cipher text
Table 1 describes the output of the encryption process. It shows the size of the cipher text in bytes. It is shown that Kumar algorithm has a largest size of cipher text whereas the other algorithms give a cipher text sizes that are equal or very close to the size of the plain text.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Table 1
Size of cipher text (byte).
Size of plain text (bytes)
Subasree
25,615
35,080
61,386
184,162
25,615
35,080
61,386
184,162
25,645
35,192
61,486
184,262
35,142
48,226
84,340
253,008
25,610
35,070
61,369
184,143
25,615
35,080
61,386
184,162
25,647
35,112
61,418
184,194
218 4.2. Time of encryption and decryption processes
219 The encryption time is the time that an encryption algorithm takes to produce a cipher text from a plaintext. The
220 decryption time is the time that an decryption algorithm takes to produce a plaintext from a cipher text.
221 Tables 2 and 3 show the time of encryption and decryption processes for different sizes of plain text; respectively.
222 It is clear that, THCA achieves the least time for both encryption and decryption. This is due to that the plain text of
223 the proposed algorithm is split into two different parts and these parts are encrypted and decrypted simeltaneously.
224 The times shown in the tables are the maximum time of processing the two parts. It is shown that Zhu algorithm has
225 the same time of THCA since that the message in this algorithm is encrypted in a single phase which leads to less
226 security level. Achieving less encryption time results in highest throughput since that the throughput of encryption can
227 be calculated as the total plain text over the encryption time. Then, THCA can realize the highest throughput.
228 4.3. Time complexity
229 In THCA, the time complexity of encryption process is calculated as follows: Max[O(log2(n +1) + 2n + *Jn +
230 8) AND O(log(n2) + log(n) + 3n + 6)] that is equal to O(log(n2) + log(n) + 3n + 6) which is the integration of
231 O(log(n2) + log(n) + n + 2n + 2 +1 + 3). It consists of seven terms. The first term (log(n2)) denotes time complex-
232 ity of DUAL RSA (Sun et al., 2007). The second and third terms (log(n) + n) refer to time complexity of XOR. The
233 forth and fifth terms (2n + 2) refer to time complexity of two for loops. The sixth term (1) refers to time complexity
234 of MD5. It is a constant value since it is a probabilistic comparison algorithm (Erickson, 2008). The seventh term (3)
235 refers to the variables. It can be neglected since it is very small with respect to n. Then, the total time complexity of
236 encryption can be summerized to O(log(n2) + log(n) + 3n) that can be reduced further to O(n).
Table 2
Time of encryption (ms).
Size of plain text (bytes) Subasree Dubal Kumar Ren Zhu THCA
609 2063 2032 1500 1432 998 998
25,615 3683 6305 1518 1490 1022 1022
35,080 5651 15,643 1526 1468 1059 1059
61,386 15,351 120,608 4219 3019 3143 3143
184,162 105,889 198,700 5752 4970 3814 3814
Table 3
Time of decryption (ms).
Size of plain text (bytes) Subasree Dubal Kumar Ren Zhu THCA
609 1078 1016 966 756 562 562
25,615 1085 4053 972 821 713 713
35,080 1082 13,227 980 953 824 824
61,386 1197 13,227 991 864 891 891
184,162 2087 18,578 1099 1075 907 907
Algorithm
Encryption process
Decryption process
Subasree
O(log(n2) + 4n)
O(log(n2) + log 2(n) + 4n + 4n) O(log 2(n + 1) + 4n + 4n) O(log(n2) + V" + 4n) O(log 2(2n + 1) + Vn + 4n) O(log(n2) + log(n) + 3n)
O(log(2n3) + 4n)
O(log(2n3) + log 2(n) + Vn + 4n)) O(log 2(n + 1) + Vn + 5n) O(log(n3) + Vn +4n) O(log 2(2n + 1) + Vn + 4n) O(log(n) + log(2n3 ) + 2n)
The time complexity of decryption process of THCA is calculated as follows: Max[O(log2(n +1) + 2n + ^¡n + 10) AND O(log(n) + log(2n3) + 3n + 7)] that is equal to O(log(n) + log(2n3) +2n +6) which is the integration of O(log(2n3) + log(n) + n + 2n + 2 +1 + 4). It consists of seven terms. The first term (log(2n3)) denotes time complexity of decryption of DUAL RSA, the second and third terms (log(n) + n) refer to time complexity of XNOR, the forth and fifth terms (2n + 2) refers to time complexity of two for loops, the sixth term (1) refers to time complexity of MD5, the seventh term (7) refers to time complexity of the variables. Then, it can be summerized to O(log(n) + log(2n3) + 2n) that can also be reduced further to O(n).
Table 4 shows the time complexity of THCA compared with the existing algorithms for encryption and decryption process. Note that the time complexity shown in the table before final abbreviation that yields to O(n) in all algorithms. However, it is shown that THCA has the least amount of processing time because of that both the two phases of the algorithm are encrypted and decrypted simultaneously.
5. Simulation Results of WSN
In order to prove the results of the proposed protocol, it is tested as the security protocol in WSN. The simulation is done using the network simulator NS2.
5.1. Simulation environment
It is assumed in the topology of the WSN that it consists of twenty nodes. The nodes are located randomly in the network. Different scenarios are assumed for transmission of data between different nodes. Each node must have the information about the other nodes present in the WSN. This information is first transmitted in the form of small packet. This packet contains the information about the source address. If any intermediate node receives a packet, it forwards this packet to the next neighboring node. When this packet reaches the final node it checks all the address present in this packet and then transmits reply back to the source node. The size of the packet increases gradually as the intermediate nodes add their address to the packet. After transmission of packet, every sensor node has the idea of the location of every other sensor node in the network. Therefore, the communication can be done from one node to the other node.
In some situations, the links present between the sensors nodes fail or the sensor nodes move from their actual location and thereby resulting in breakage of the link. In some other cases, improper packets may be propagated over the link between any two nodes. In addition, some packets may be dropped due to delay of execution time (time out). When such insecure packets are dropped, the link will not be used for a certain time and the network uses an alternate path.
5.2. Energy consumption
Because of the power source limitation of WSNs, all processes and communication protocols regarding sensor networks must minimize energy consumption so that sensor lifetime may be maximized. The evaluation of energy consumption considers both the energy consumed during the execution of cryptographic algorithms and the energy of communication. The energy required for the calculation of cryptography algorithm is simply the product of the average power consumption and the execution time of this algorithm. The execution time was determined through simulations.
0 2 4 6 8 10 12 14 16 18
Size of plain text (byte) 4
Fig. 2. Rate of dropped packets of THCA and the existing algorithms.
Communication energy depends on the distance between sending and receiving nodes and the time required for sending the cipher text, which it is proportional to the size of the plain text.
Fig. 1 shows the energy consumption of THCA compared to the other algorithms. It consumes about 10 (mJ) at 184,162 plain text size (byte) as opposed to 63, 31, 544, 290 (mJ) for Ren, Kumar, Dubal, and Subasree; respectively. It is shown that the THCA achieves the least energy consumption which is the demand to guarantee the lifetime of sensor networks.
277 5.3. Rate of dropped packets
278 Fig. 2 shows the rate of dropped packets. It is shown that the THCA achieves the least rate of packet dropping
279 compared to the other protocols. This is due to that, the proposed THCA checks authentication using DUAL RSA
280 and then protects the network from unsecured nodes. In addition, the number of dropped packets due to time out is
281 decreased in the THCA since it has the least execution.
282 6. Implementation of THCA on images
283 In this section, the proposed THCA is tested on image encryption to prove its robustness against different types of
284 attacks. It is applied on image protection application using magic cube theory. The magic cube (Bashir et al., 2012)
285 divides the original image into six sub-images and these sub-images are divided into a number of blocks and attached
286 to the faces of a magic cube as shown in Fig. 3. Then the attached image is fed to THCA which is applied to the pixels
287 of the image to encrypt the scrambled image.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx 11
Fig. 3. Mapping the six sub-images on the magic cube faces.
6.1. Description of the cube mapping
1. The original image is resized to a size of M x N so that the resized image can be divided into six sub-images of the same size and with no overlapping.
2. The sub-images have the size (M/3) x (N/2). The six faces are marked as Up (U), Front (F), Right (R), Left (L), Down (D) and Back (B).
3. The six sub-images are divided into a number of blocks with the same number of pixels.
4. According to the implementation of THCA, The sub-images (U, D, F) are encrypted by first part of THCA which uses (AES with ECC) for encryption while sub-images (B, L, R) are encrypted by the second part which uses (XOR-DUAL RSA).
5. At the receiver side, the original image is retrieved by mapping of the six sub-images on the magic cube faces.
6.2. Security analysis
A good encryption procedure should be robust against all kinds of attacks. Some security analysis was performed on the proposed THCA in the case of using image, including the most important like statistical analysis and key space analysis.
6.2.1. Statistical analysis
An ideal cipher should be robust against any statistical attack. To prove the robustness of the proposed algorithm, we performed statistical analysis by calculating the histograms and the correlations of two adjacent pixels in the plain image/cipher image.
6.2.1.1. Histograms analysis. The histogram of the image shows how pixels in the original images are distributed by graphing the number of pixels at each gray level (Abderrahim et al., 2012; Ahmed et al., 2007). A color histogram represents the number of pixels that have colors in each of a fixed list of color ranges that used for three-dimensional space like RGB channels. We calculated and analyzed the histograms of the several original and encrypted images that have widely different contents.
Fig. 4 shows the histogram of Man plain image and the Man encrypted image. It is shown that they are significantly different. Fig. 5 shows the histogram of Lena plain image. Fig. 5 (e), (g) and (i) illustrate the Histogram of Lena's image of Red, Green and Blue channels respectively. The histograms of the cipher images are shown in Fig. 5 (d), (f), (h) and (j). It is also shown that they are significantly different from that of the original images and have no statistical resemblance to the plain images.
It is clear that the histogram of the encrypted images are significantly different from the respective histogram of the original images and hence does not provide any clue to employ any statistical attack on the proposed THCA in the case of using image.
6.2.1.2. Correlation coefficient analysis. In addition to the histogram analysis, we also analyzed the correlation between two vertically adjacent pixels, two horizontally adjacent pixels and two diagonally adjacent pixels in plain
R. Rizk, Y. Alkady / Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
O lOO 200
d) Histogram of encrypted image
Fig. 4. Histograms of the plain image and the corresponding cipher image.
image/cipher image; respectively. First, we randomly selected 2000 pairs of two adjacent pixels from an image. Then, we calculated their correlation coefficient (rxy) using the following formula (Abderrahim et al., 2012):
Cov(x, y)
v/D(x)^D(y)
Cov(x, y) = Y(xi - x')(y - y')
y = NT.y'
D(x) = ]T(xi - x )
D(y) = y - y')
where xi is the intensity of the ith pixel in original image, yi is the intensity of the ith pixel in ciphered image, x' is the mean intensity of original image (the sum values divided by number of selected pixels), y' is the mean intensity of ciphered image, and N is the number of selected pixels.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Fig. 5. Histograms of the colored plain image and the corresponding cipher image in RGB channels.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Table 5
Correlation analysis in man plain image/man cipher image.
Algorithm Direction of adjacent pixels Plain image Cipher image
Horizontal -0.0468 0.2951
Vertical -0.0488 0.09935
Diagonal -0.0353 0.09614
Table 6
Correlation analysis in lena plain image/lena cipher image.
Algorithm direction of adjacent pixels
Plain image
Cipher image
Horizontal
Vertical
Diagonal
0.9898 0.9805 0.9769
0.0303 0.0302 0.0311
The correlation coefficient has the value 1 if the two images are absolutely identical, 0 or very near to zero if they are completely uncorrelated, or -1 if they are completely anti-correlated, for example if one image is the negative of the other. It is clear from Table 5 that the cipher Man image is highly independent of the Man plain image.
In Table 6, the correlation coefficients of two horizontally adjacent pixels are 0.9898 and 0.0303; respectively for both Lena plain image/Lena cipher image of THCA. Similar results for vertical and diagonal directions are obtained. It is clear from Table 6 that there is a negligible correlation between the two adjacent pixels in the cipher image. However, the two adjacent pixels in the plain image are highly correlated as shown in the Fig. 6.
Fig. 6. Correlation of two adjacent pixels: (a) distribution of two horizontally adjacent pixels in the Man plain image, (b) distribution of two horizontally adjacent pixels in the Man encrypted image, (c) distribution of two horizontally adjacent pixels in Lena Plain image, (d) distribution of two horizontally adjacent pixels in Lena encrypted image.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Fig. 7. Key sensitive Test (1) with THCA.
342 6.3. Key space analysis
343 A good image encryption algorithm should be sensitive to the cipher keys, and the key space should be large enough
344 to make brute-force attacks infeasible. THCA has two types of keys, the first key is the secure key which is used to
345 encrypt the first half of plain image and the second key is the public key which used to encrypt the second half of
346 plain image. Then, for the proposed THCA, key space analysis and testing have to be carefully performed.The change
347 of a single bit in the secret key or private key should produce a completely different encrypted image, which means
348 that the cipher image cannot be decrypted correctly although there is only a slight difference between encryption and
349 decryption keys. This guarantees the security of THCA against brute-force attacks to some extent. For testing the key
350 sensitivity of THCA, the following steps were performed in Test (1) shown in Fig. 7:
351 1. First, the original image (Fig. 7(a)) is encrypted by using the test secure key "1551917990046475381" which is
352 equivalent to "1589853085422475" (in hexadecimal) and public key (3, 33). The resultant image is referred as
353 encrypted image A as shown in Fig. 7(b).
354 2. Then, the most significant bit of the secret key (in hexadecimal) is changed, so that the original secret key becomes,
355 say "2704839494653322357" which is equivalent to "2589853085422475" (in hexadecimal) and the same public
356 key (3, 33). The resultant image is referred as encrypted image B as shown in Fig. 7(c).
357 3. Then, the least significant bit of the secret key (in hexadecimal) is changed, so that the original secret key becomes,
358 say "1551917990046475380" which is equivalent to "1589853085422474" (in hexadecimal) and the same public
359 key (3, 33). The resultant image is referred as encrypted image C as shown in Fig. 7(d).
360 4. Finally, the three ciphered images A, B and C encrypted by the three slightly different keys are compared.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
Table 7
Correlation coefficients between the corresponding pixels of the three different encrypted images obtained by using slightly different secret keys.
Image 1 Image 2 Correlation coefficient
Encrypted image A Fig. 7(b) Encrypted image B Fig. 7(c) 0.0309
Encrypted image B Fig. 7(c) Encrypted image C Fig. 7(d) 0.0358
Encrypted image C Fig. 7(d) Encrypted image A Fig. 7(b) 0.0342
Fig. 7 shows the original image as well as the three encrypted images produced in the above steps. It is not easy to compare the encrypted images by simply observing these images. So for comparison, the correlation coefficients between the corresponding pixels of the three encrypted images have to be compared. For this calculation, the formula in (21) is used except that in this case x and y are the values of corresponding pixels in the two encrypted images to be compared. Table 7 shows the results of the correlation coefficients between the corresponding pixels of the three encrypted images A, B and C. It is clear from the table that there is no correlation exists among the three encrypted images even though these have been produced by using slightly different secret keys. Key sensitivity analysis shows that changing one bit in encryption key will result in a completely different cipher image.
Moreover, Fig. 8 shows the results of Test (2) that presents some attempts to decrypt an encrypted image with slightly different secret keys than the one used for the encryption of the original image. Fig. 8(a) and (b) shows the original image and the encrypted image produced using the secret key "1589853085422475" (in hexadecimal) and private key (7,3,11), respectively. Whereas Fig. 8(c) and (d) shows the images after the decryption of the encrypted image with the same secret key "1589853085422475" (in hexadecimal) and the slightly different secret key "1589853085422474" (in
Fig. 8. Key sensitive Test (2) with THCA.
R. Rizk, Y. Alkady /Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx 17
hexadecimal); respectively. It is clear that the decryption with a slightly different key fails completely and hence the proposed THCA is highly key sensitive.
7. Conclusion
In this paper, a robust hybrid security algorithm for WSNs is proposed. It is designed in order to solve several problems as practical implementation, short response time, efficient computation and the strength of cryptosystem. The proposed THCA tries to trap the intruder by splitting the plain text and then applies two different techniques. First, it takes the advantages of the combination of both symmetric and asymmetric cryptographic techniques using both AES and ECC algorithms. Second, XOR-DUAL RSA is used since it is more robust and cannot be easily attacked. In addition, Hashing is also used for data integrity using MD5 to be ensured that the original text is not being altered in the communication medium. The performance of THCA is compared with other existing security algorithms. It offers better security for a shorter encryption and decryption time and smallest cipher text size. Thereby, it decreases the processing overhead and achieves lower energy consumption that is appropriate for all WSN applications. The proposed THCA is implemented in the case of image encryption. It is shown that it is robust against different types of attacks.
References
Abderrahim, N., Benmansour, F., Seddiki, O., 2012. Integration of chaotic sequences uniformly distributed in a new image encryption algorithm. Int. J. Comput. Sci. (IJCSI) 9 (2).
Ahmed, H., Kalash, H., Farag Allah, O., 2007. An Efficient Chaos-Based Feedback Stream Cipher (ECBFSC) for image encryption and decryption.
Informatica 31 (1), 121-129. Balitanas, M., 2009. WiFi protected access-pre-shared key hybrid algorithm. Int. J. Adv. Sci. Technol. 12.
Bashir, A., Basari, A., Almangush, H., 2012. Novel Image Encryption using an Integration Technique of Blocks Rotation based on the Magic cube
and the AES Algorithm. Int J. Comput. Sci. 9 (4). Boneh, D., Durfee, G., 2000. Cryptanalysis of RSA with private key d less than N. IEEE Trans. Inf. Theory 46 (4), 1339-1349. Burr, W., 2003. Selecting the advanced encryption standard. IEEE Secur. Priv. 1 (2), 43-52.
Dubal, M.J., Mahesh, T.R., Ghosh, P.A., 2011. Design of a new security protocol using hybrid cryptography architecture. In: Proceedings of 3rd
International Conference on Electronics Computer Technology (ICECT), vol. 5, India. Erickson, J., 2008. Hacking: The Art of Exploitation, ch. 7, 2nd ed. W. Pollock of Publisher, USA, pp. 393-450.
Farouk, F., Rizk, R., Zaki, F.W., 2014. Multi-level stable and energy efficient clustering protocol in heterogeneous wireless sensor network. IET
Wirel. Sens. Syst. 4 (4), 159-169, http://dx.doi.org/10.1049/iet-wss.2014.0051. Faye, S., Myoupo, J.F., 2013. Secure and energy-efficient geocast protocols for wireless sensor networks based on a hierarchical clustered structure. Int. J. Netw. Secur. 15 (1), 121-130.
Fouchal, H., Hunel, P., Ramassamy, C., 2014. Towards efficient deployment of wireless sensor networks. Secur. Commun. Netw., http://dx.doi.org/10.1002/sec.1059.
Frunza, M., Asachi, Gh., 2007. Improved RSA encryption algorithm for increased security of wireless networks. In: ISSCS International Symposium, vol. 2.
Hayajneh, T., Doomun, R., Almashaqbeh, G., Mohd, B.J., 2013. An energy-efficient and security aware route selection protocol for wireless sensor
networks. Secur. Commun. Netw., http://dx.doi.org/10.1002/sec.915. Hossain, Md.A., Islam, Md.K., Das, S.K., Nashiry, Md.A., 2012. Cryptanalyzing of message digest algorithms MD4 and MD5. Int. J. Cryptogr. Inf. Secur. (IJCIS) 2 (1).
Johnson, D., Menezes, A., Vanstone, S., 2001. The elliptic curve digital signature algorithm (ECDSA). Int. J. Inf. Secur. 1 (1), 36-63.
Kodali, R., Sarma, N., 2013. Energy efficient ECC encryption using ECDH. In: Emerging Research in Electronics, Computer Science and Technology,
Lecture Notes in Electrical Engineering, vol. 248. Springer, pp. 471-478. Kumar, N., 2012. A Secure Communication Wireless Sensor Networks Through Hybrid (AES+ECC) Algorithm, vol. 386. von LAP LAMBERT Academic Publishing.
Lasla, N., Derhab, A., Ouadjaout, A., Bagaa, M., Challal, Y., 2014. SMART: secure mlti-paths routing for wireless sensor networks. In: Ad-hoc,
Mobile, and Wireless Networks, Lecture Notes in Computer Science, vol. 8487, pp. 332-345. Lenstra, A., 2001. Unbelievable security matching AES security using public key systems. Adv. Cryptol. ASIACRYPT 2248, 67-86. Mary Anita, E.A., Geetha, R., Kannan, E., 2015. A novel hybrid key management scheme for establishing secure communication in wireless sensor
networks. Wirel. Pers. Commun., http://dx.doi.org/10.1007/s11277-015-2290-9. Ren, W., Miao, Z., 2010. A hybrid encryption algorithm based on DES and RSA in bluetooth communication. In: Proceedings of the 2nd International
Conference on Modeling, Simulation and Visualization Methods, China, pp. 221-225. Singh, G., Supriya, 2013. A study of encryption algorithms (RSA, DES, 3DES and AES) for information security. Int. J. Comput. Appl. 67 (19), 33-38.
Subasree, S., Sakthivel, N.K., 2010. Design of a new security protocol using hybrid cryptography algorithms. IJRRAS 2 (2), 95-103.
18 R. Rizk, Y. Alkady / Journal of Electrical Systems and Information Technology xxx (2015) xxx-xxx
428 Sun, H., et al., 2007. Dual RSA and its security analysis. IEEE Trans. Inf. Theory, 2922-2933.
429 Tillich, S., GroBschâdl, J., 2005. Accelerating AES using instruction set extensions for Elliptic Curve cryptography. In: Computational Science and
430 Its Applications - ICCSA, vol. 3481., pp. 665-675.
431 Yu, N., Zhang, L., Ren, Y., 2013. A novel D-S based secure localization algorithm for wireless sensor networks. Secur. Commun. Netw.,
432 http://dx.doi.org/10.1002/sec.909.
433 Zhu, S., 2011. Research of hybrid cipher algorithm application to hydraulic information transmission. In: Proceedings of International Conference
434 on Electronics, Communications and Control (ICECC), China.
