Fe25959964

V.U.K. Sastry, K. Anup Kumar / International Journal of Engineering Research and
Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 2, Issue 5, September- October 2012, pp.959-964
A Block Cipher Obtained By Blending Modified Feistel Cipher
And Advanced Hill Cipher Involving A Pair Of Key Matrices
1
V.U.K. Sastry, 2K. Anup Kumar
1
Director School of Computer Science and Informatics, Dean(R & D), Dean (Admin),
Department of Computer Science and Engineering,
Sreenidhi Institute of Science and Technology, Ghatkesar, Hyderabad, 501301, Andhra Pradesh, India.
2
Associate Professor, Department of Computer Science and Engineering,
Sreenidhi Institute of Science and Technology, Ghatkesar, Hyderabad, 501301, Andhra Pradesh, India.

Abstract
In this investigation, we have developed a block Pi = ( K Qi-1 L ) mod N (1.1)
cipher by blending a modified Feistel cipher and Qi= Pi - 1 Pi
i = 1 to n,
the advanced Hill cipher. In this analysis, we and
have made use of a pair of involutory matrices, Qi-1 = ( K Pi L ) mod N, (1.2)
say A and B which include the keys K and L. Pi -1= Qi Pi
i = n to 1,
Here the size of the plaintext is 1024 binary bits
and the size of the keys (both put together) is
256 binary bits. The involutory matrices A and Here, Pi and Qi are the plaintext matrices
B, the modular arithmetic inverse, the XOR at the ith stage of the iteration, K and L are the
operation, and the functions Shift ( ) and Mix ( ) involutory key matrices and N is a positive integer
are playing a vital role in strengthening the chosen appropriately. Here n denotes the number of
cipher. The cryptanalysis carried out in this iterations. In the development of this cipher, we
investigation clearly shows that the strength of have utilized the functions Mix ( ) and Shift ( ), and
the cipher is quite considerable, and the cipher XOR operation.
cannot be broken by any attack.
We now present the plan of the paper. In
Key words: Encryption, Decryption, Key matrix, section 2, we discuss the development of a pair of
Shift, Mix, XOR Operation. involutory matrices. In section 3, we discuss the
development of the cipher and present the
Introduction flowcharts and the algorithms describing the
In the recent investigations [1], we have cipher. Section 4 is devoted to an illustration of the
developed a block ciphers by generalizing the cipher, and in this we have determined the
classical Feistel cipher [2], wherein we have avalanche effect. We investigate the cryptanalysis
considered a plaintext which can be represented in in section 5. Finally in section 6, we have given the
the form of a pair of matrices instead of a pair of details about the computations and conclusions.
binary strings that was used in the case of classical
Feistel cipher. 1. Development of the Involutory Matrix
An involutory matrix is a square matrix
In this development, we have made use of whose inverse is same as the original matrix.
an involutory matrix to multiply both the sides of Let A = [aij] and B = [bij] be two square matrices
the plaintext matrix. In addition to this, we have of size n.
made use of the operations Mix ( ) and Shift ( ) and Let the involutory matrices of matrices A and B be
XOR operation. The avalanche effect and the denoted as
cryptanalysis discussed in this analysis effectively A11 A12
indicate that the cipher is a strong one. A= (2.1)
A21 A22
In the present investigation, our objective and
is to develop the modified Feistel cipher using the
features of Advanced Hill cipher [3] with multiple B11 B12
keys. B= (2.2)
B21 B22
The basic equations governing the
encryption and the decryption of this cipher are Where all the sub matrices A11, A12, A21
given by and A22 B11,B12,B21 and B22 are square matrices of
size n/2.

959 | P a g e

As the modular arithmetic inverse of an involutory In the development of this cipher, encryption and
matrix given in (2.1) is governed by the relations decryption are governed by the relations (1.1) and
(1.2) respectively. We now present the flow charts
( A A-1 ) mod N = I, (2.3) and the algorithms describing the encryption and
the decryption processes.
and A = A-1 (2.4)
Read K , L , P
Thus we have, A2 mod N = I (2.5)

From (2.1) and (2.5) we get, P0 Q0

A22 mod N = - A11 mod N, (2.5)
for i = 1 to n
A12 = [ d (I – A11)] mod N (2.6)

(2.7) Pi-1 Qi-1
A21 = [ w ( I + A11 ) ] mod N

where (dw) mod N = 1 (2.8)

Given A11, on selecting d, where d lies in Pi = ( K Qi-1 L ) mod N
the interval 0 < d < N, firstly we obtain w from
(2.8), then we determine A22, A12 and A21 by Pi
satisfying the relations (2.5) to (2.7). Qi
Pi = M ( Pi ) Pi = M ( Pi )
If A11 is the key represented by square
matrix A of size n/2, then we get the involutory
matrix K of size n. Pi = S ( Pi ) Pi = S ( Pi )

Pi Qi
Now let us consider the modular
arithmetic inverse of an involutory matrix given in Pi , Qi
(2.2) governed by the relations C = Pn || Qn
(2.9)
( B B-1 ) mod N = I,
and B = B-1 Fig 1. The process of Encryption
(2.10)
Thus we have, B2 mod N = I (2.11)

From (2.2) and (2.11) we get, In the matrices Pi and Qi, all the numbers
are lying in the interval [0-255]. Here K and L are
B22 mod N = - B11 mod N, (2.12) two involutory matrices of the key matrices and A
and B respectively. In this analysis, we have taken
B12 = [ s (I – B11)] mod N (2.13) n equal to 16.
(2.14)
B21 = [ t ( I + B11 ) ] mod N In the flowchart, for the sake of elegance,
where (st) mod N = 1 (2.15) we have written the functions Mix ( ) and Shift ( ),
arising in encryption, as M ( ) and S ( )
Given B11, on selecting s, where s lies in the respectively. The reverse processes represented by
interval 0 < s < N, firstly we obtain t from (2.15), IMix ( ) and IShift ( ), arising in decryption, are
then we determine B22, B12 and B21 by satisfying denoted by IM ( ) and IS ( ).
the relations (2.12) to (2.14). If B11 is the key
represented by square matrix B of size n/2, then we
get the involutory matrix L of size n.

2. Development of the Cipher
Let P be the plaintext consisting of 2m2
characters. On employing the EBCIDIC code, the
plaintext can be written in the form of a pair of
square matrices P0 and Q0, wherein, each one is of
size m. Let us consider a key matrices K and L,
whose size is mxm.

960 | P a g e

Read C, K, L Qi = S (Qi)
end

Pn Qn
6. C = Pn Qn /* represents concatenation */

7. Write(C)

for i = n to 1 Algorithm for Decryption
Pi Qi
1.Read C, A, n and N.
Pi = IS (Pi) Qi = IS (Qi)
2.K = Involute (A,d)
Pi = IM ( Pi ) 3.L = Involute (B,s)
Qi = IM ( QiQi
)
4. Pn = Left half of C
Pi
Qn = Right half of C
Qi
Qi-1 = (K Pi L ) mod N
5. for i = n to 1
Qi-1
begin
Pi = IS (Pi)
Pi-1 Qi-1
Pi = IM (Pi)

Pi-1 , Qi-1 Qi = IS (Qi)
P = P0 || Q0
Qi = IM (Qi)
Fig 2. The process of Decryption
Qi-1 = ( K Pi L ) mod N

The processes of encryption and Pi-1 = Qi Pi
decryption depicted in the flow charts are described end
by the algorithms given below. 6. P = P0 Q0 /* represents concatenation */
Algorithm for Encryption
7. Write (P)
1.Read P, A, n and N. For the sake of elegance, in the iteration
process when we come across the function Mix ( )
2.K = Involute (A,d) and Shift ( ) during encryption, they are denoted
as M ( ) and S ( ) respectively, and the functions
3.L = Involute (B,s) IMix ( ) and IShift ( ) during decryption are
denoted as IM ( ) and IS ( ). For a detailed
4. P0 = Left half of P. explanation of these functions see [1].

Q0 = Right half of P. 3. Illustration of Cipher
Consider the plaintext given below.
5. for i = 1 to n Dear brother! India is well known for
mines. There are many mines from which gold iron
begin and silver can be obtained very easily. This is the
Pi = ( K Qi-1 L ) mod N basic reason for some Indians to become rich. Do
remember in India right from the ancient times
Qi = Pi-1 Pi even upto today Indians fight among themselves.
Even brothers cannot remain united! Right from the
Pi = M (Pi) time of mahabharata even upto today same is the
story of brothers. They do not think about their
Pi = S (Pi) progress in a conjoint manner. Each one is a selfish
fellow. Tell our brothers; one day or the other day
Qi = M (Qi)

961 | P a g e

we have to win over this country and rule again.
(4.1) (4.5)
Let us take a pair of key matrices A and B as
Consider the first 128 characters of the plaintext.
Thus we have
Dear brother! India is well known for
mines. There are many mines from which gold iron 69 124 27 167
and silver can be obtained very easily. T (4.2)
On converting each character of the 135 79 99 111
plaintext in (4.2) into its EBCIDIC code form, we A= (4.6)
get the plaintext matrix P as Q0 248 199 209 75
=
68 101 97 114 32 98 114 111 116 104 101 239 45 255 92
114 33 32 73 110

100 105 97 32 105 115 32 119 101 108 108 and
32 107 110 111 119
215 113 19 147
110 32 102 111 114 32 109 105 110 101 115
46 32 84 104 101 223 109 254 12
B= (4.7)
P= 114 101 32 97 114 101 32 109 97 110 121 56 01 127 174
32 109 105 110 101
59 146 189 81
115 32 102 114 111 109 32 119 104 105 99
104 32 103 111 108

100 32 105 114 111 110 32 97 110 100 32 On using (2.5 ) to (2.7 ) and taking d = 99, we get
115 105 108 118 101 the involutory key matrix K as

114 32 99 97 110 32 98 101 32 111 98 69 124 27 167 180 12 143 107
116 97 105 110 101 135 79 99 111 203 214 183 19
248 199 209 75 24 11 144 255
100 32 118 101 114 121 32 101 97 115 105 239 45 255 92 147 153 99 207
108 121 46 32 84 K=
130 84 233 237 187 132 229 (4.8)
89
(4.3) 141 112 1 133 121 177 157 145
168 77 134 249 8 57 47 181
Consider P0 and Q0 as two matrices obtained from 5 47 181 63 17 211 1 164
(4.3), as the left halve and the right halve
respectively. Thus we have

68 101 97 114 32 98 114 111 On using ( ) to ( ) and taking s = 189 , we get the
100 105 97 32 105 115 32 119 involutory key matrix L as
110 32 102 111 114 32 109 105
P0 114 101 32 97 114 101 32 109 (4.4) 215 113 19 147 02 147 249 121
= 115 32 102 114 111 109 32 119 223 109 254 12 93 68 122 36
100 32 105 114 111 110 32 97 56 01 127 174 168 67 250 138
114 32 99 97 110 32 98 101 59 146 189 81 113 54 119 240
L=
100 32 118 101 114 121 32 101 184 197 15 143 41 143 237 109
203 06 214 252 33 147 02 244
152 149 128 70 200 255 129 82
116 104 101 114 33 32 73 110 87 250 01 186 197 110 67 175
101 108 108 32 107 110 111 119
110 101 115 46 32 84 104 101 (4.9)
97 110 121 32 109 105 110 101
104 105 99 104 32 103 111 108
110 100 32 115 105 108 118 101 On using (4.4) , (4.5) , (4.8) , (4.9) and the
32 111 98 116 97 105 110 101 encryption algorithm given in section 3 , we get the
97 115 105 108 121 46 32 84 cipher text C as

962 | P a g e

On comparing (4.10) and (4.11) in their
binary form, we readily notice that these two
28 188 242 196 4 152 196 240 246 144 220 ciphers differ by 513 bits out of 1024 bits this
218 18 68 132 224 shows that the cipher is a potential one.
58 110 90 220 182 222 22 146 102 108 30
8 44 84 154 100 Let us now consider a one bit change in
200 58 122 230 136 10 48 140 172 236 106 the key K, this can be done by changing the first
C= 176 182 172 138 74 row, first column element from 69 to 67.
140 230 146 62 190 54 42 54 210 230 44
254 210 88 170 52 On using the modified key and the
36 176 86 104 144 118 246 206 16 22 98 encryption algorithm given in section 3 , and by
18 194 218 70 114 keeping the plaintext as it is , we get the cipher text
110 94 150 150 48 236 164 108 6 206 70 C as
114 96 194 166 120
182 216 250 66 4 56 202 118 184 38 248 238 218 42 222 254 174 116 244 154 190 206
110 168 182 100 138 46 156 166 178 236
122 206 126 250 150 62 124 54 110 102 100 16 146 104 90 188 164 136 6 186 108 152
124 152 128 122 186 44 170 126 252 64
248 14 242 218 110 164 10 38 88 146 204
(4.10) 220 152 212 174 166
64 212 198 48 204 138 186 28 204 190 12
Now on applying the decryption algorithm 156 150 60 48 16 (4.12)
given in section 3 with necessary inputs, we get 10 218 210 228 164 172 22 198 146 24 108
back the original plaintext given in (4.3). 130 182 204 224 236
244 184 58 120 14 212 176 146 182 142
Let us now consider the avalanche effect 58 108 102 56 44 144
which shows the strength of the cipher in a 60 144 244 214 44 146 88 228 184 188
qualitative manner. 100 212 60 184 184 20
176 190 138 48 140 170 114 110 164 98 234
In order to carry out this one, firstly, let us 214 146 190 120 202
consider a one bit change in plaintext P. This can
be done by changing the first row , first column On comparing (4.10) and (4.12) in their binary
element of ( 4.3) from 68 to 69. form we notice that they differ by 517 bits out of
1024 bits. This shows that the cipher is a strong
On using the modified plaintext and one.
without altering the keys K and L and by adopting
the encryption algorithm given in section 3, we get 4. Cryptanalysis
the ciphertext as In the literature of cryptography, the conventional
cryptanalytic attacks used are
1. Cipher text only (Brute Force) attack.
124 38 238 178 202 230 126 216 94 246 244 2. Known Plaintext attack.
122 156 166 178 236 3. Chosen Plaintext attack.
16 146 104 90 188 164 136 6 186 108 152 4. Chosen Cipher text attack.
44 170 126 252 64 The primary goal of all these cryptanalytic
248 14 242 200 150 36 236 182 210 2 204 attacks is to break the cipher by obtaining the key
220 152 212 174 166 used for encryption.
C 64 212 198 48 204 138 186 28 204 190 12 Let us now consider the brute force attack
=
156 150 60 48 16 first, in the development of this cipher, as we have
24 108 130 182 204 132 236 232 114 82 86 used a pair of key matrices A and B, each one
138 98 72 224 236 having size 4x4. There are 32 decimal number in
244 184 58 120 14 212 176 146 182 142 58 the key. As each element of the key can be
108 102 56 44 144 represented with 8 binary bits in its EBCIDIC code
60 144 244 214 44 90 154 198 76 66 72 98 form, the total length of the key is 256 bit.
188 184 184 20 Hence the size of the key space is
176 190 138 48 140 170 114 110 164 98 234
214 146 190 120 202
25.6 25.6
(4.11) 2256 = (210) ≈ (103) = 1076.8

963 | P a g e

If the time required to get the plaintext by The programs required for encryption and
using one value of the key in the key space is10 -7 decryption are written in C Language.
seconds, then the time required to execute the
cipher with all possible keys in key space is 5. References
1076.8 x 10-7 [1] V. U. K. Sastry, K. Anup Kumar, “A
= 3.171x 1061.8 years. Block Cipher Obtained by Blending
365x24x60x60 Modified Feistel Cipher and Advanced
As this number is very large, it is impractical to Hill Cipher Involving a Single Key
break the cipher by using this attack. Matrix” ( Sent for Publication ) IJERA,
Let us examine the known plaintext Aug 2012
attack, to carry out this attack, assume that the [2] William Stallings, Cryptography and
attacker knows the plaintext P ciphertext C pairs as Network Security, Principles and Practice,
many as required. Third Edition, Pearson, 2003.
If we confine our attention to only one [3] Bhibhudendra Acharya, Saroj Kumar
round of the iteration process i.e r = 1. Then the Pahigrahy, Sarat Kumar Parta, and
relations governing the encryption process for one Ganapati Panda, “Image encryption using
round can be written as Advanced Hill cipher Algorithm”,
P1 = ( K Q0 L ) mod N (5.1) International Journal of Recent Trends in
Q1 = P0 P1 (5.2) Engineering, Vol . 1(1), May 2009.
P1 = M (P1) (5.3)
P1 = S (P1) (5.4) Authors profile:
Q1 = M (Q1) (5.6)
Q1 = S (Q1) (5.7)
C = P1 Q1 (5.8)
Here we know P0, Q0 and C and the
encryption algorithm. As the functions Shift ( ) and
Mix ( ) are known, using (5.8) to (5.3), we get P1
and Q1 occurring on the left hand side of (5.1) and Dr. V. U. K. Sastry is presently
(5.2) As P0 and the Q0 occurring on the right hand working as Professor in the Dept. of Computer
side of (5.1) and (5.2) are known to the attacker, Science and Engineering (CSE), Director (SCSI),
the key K or the key L can not be obtained due to Dean (R & D), SreeNidhi Institute of Science and
the modulo of N used in (5.1). Thus the cipher Technology (SNIST), Hyderabad, India. He was
cannot be broken by the known plaintext attack Formerly Professor in IIT, Kharagpur, India
even if we confine our attention only to the first and Worked in IIT, Kharagpur during
round of the iteration process. This shows that it is 1963 – 1998. He guided 12 PhDs, and published
impossible to break the final cipher obtained after more than 40 research papers in various
sixteen rounds of the iteration process by using the international journals. His research interests are
known plaintext attack. Network Security & Cryptography, Image
Intuitively choosing a plaintext or Processing, Data Mining and Genetic Algorithms.
ciphertext and determining the key or a function of
the key is a formidable task in the case of this
cipher.
Thus from the above discussion we
conclude that this cipher is not breakable by all the
possible attacks that are available in the
cryptography. Mr. K. Anup Kumar is presently
working as an Associate Professor in the
5. Computations and Conclusions Department of Computer Science and Engineering,
In present paper, we have developed the SNIST, Hyderabad India. He obtained his B.Tech
modified Fesitel cipher. To introduce confusion (CSE) degree from JNTU Hyderabad, and done his
and diffusion, we have used a pair of functions M.Tech (CSE) from Osmania university,
namely Shift ( ) and Mix( ), and the modulo Hyderabad. He is now pursuing his PhD from
operation in every round of the iteration. As these JNTU, Hyderabad, India, under the supervision of
features thoroughly mix the plaintext at every stage Dr. V.U.K. Sastry in the area of
of the iteration process, the strength of the cipher is Information Security and Cryptography. He
good which is proved by the avalanche effect has 10 years of teaching experience and his interest
discussed in section 4. Further, as we have in research area includes, Cryptography,
increased the key space by considering a pair of Steganography and Parallel Processing Systems.
keys, the strength of the cipher has increased
enormously.

964 | P a g e

Fe25959964

More Related Content

What's hot

Viewers also liked

Similar to Fe25959964

Fe25959964