Page 1


INTERNATIONAL JOURNAL OF COMPUTER NETWORK AND INFORMATION SECURITY (IJCNIS) ISSN Print: 2074-9090, ISSN Online: 2074-9104 Editor-in-Chief Prof. I.A. Dychka, National Technical University of Ukraine "KPI", Ukraine

Associate Editors Prof. Bernard Cousin, University of Rennes 1, France Prof. N. Ch. S. N. Iyengar, VIT University, India Prof. Stanisław Deniziak, Kielce University of Technology, Poland Prof. S.S. Tyagi, Manav Rachna International University, India

Members of Editorial and Reviewer Board Prof. Belabbes Yagoubi University of Oran1 Ahmed Benbella, Algeria Prof. M. Afshar Alam Jamia Hamdard University, India Dr. A. V. Petrashenko National Technical University of Ukraine "KPI", Ukraine Dr. Seema Verma Banasthali University, India Prof. G.R.Sinha Shri Shankaracharya Technical Campus Bhilai, India Prof. V. Mukhin National Technical University of Ukraine "KPI", Ukraine Prof. D.P. Acharjya VIT University, India

Dr. E. George Dharma Prakash Raj Bharathidasan University, India

Dr. Mehdi Bateni Sheikhbahaei University, Iran

Prof. B.Ananda Krishna Gudlavalleru Engineering College, India

Dr. Mohammed M. Nasef Minufiya University, Egypt

Dr. Rakesh K. Jha Shri Mata Vaishno Devi University, India

Dr. Ajay Koul SMVD University, India

Dr. Piyush Kumar Shukla University Institute of Technology, India

Dr. C. Shoba Bindu JNTUA College of Engineering, India

Dr. Shailender Gupta Y.M.C.A. University of Science and Technology, India

Dr. Saied M. Abd El-atty Salman Bin Abulaziz University, Saudi Arabia

Dr. Krishan Kumar Shaheed Bhagat Singh State Technical Campus, India

Dr. Muhammed A. Aydin Istanbul University, Turkey

Dr. Salekul Islam United International Bangladesh

Dr. Mohamed Koubàa Tunis UniversitéTunis El Manar, Tunisia University,

Dr. Qinghai Gao Farmingdale State College, USA

Prof. Yong-Jin Lee Korea National University of Education, Korea

Prof. Kishori Lal Bansal Himachal Pradesh University, India

Prof. Shikhar Kumar Sarma Gauhati University, India

Dr. Kamel Mohamed Faraoun UDL- University of Sidi Bel Abbes, Algeria

Prof. Satish Chand Netaji Subhas Institute of Technology, India

Prof. Jamshed Siddiqui Aligarh Muslim Universit, A.M.U., India Dr. Mohammad Rasmi H. AL-Mousa Zarqa University, Jordan

International Journal of Computer Network and Information Security(IJCNIS, ISSN Print: 2074-9090, ISSN Online: 2074-9104) is published monthly by the MECS Publisher, Unit B 13/F PRAT COMM’L BLDG, 17-19 PRAT AVENUE, TSIMSHATSUI KLN, Hong Kong, E-mail: ijcnis@mecs-press.org, Website: www.mecs-press.org. The current and past issues are made available on-line at www.mecs-press.org/ijcnis. Opinions expressed in the papers are those of the author(s) and do not necessarily express the opinions of the editors or the MECS publisher. The papers are published as presented and without change, in the interests of timely dissemination. Copyright © by MECS Publisher. All rights reserved. No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner.


International Journal of Computer Network and Information Security (IJCNIS) ISSN Print: 2074-9090, ISSN Online: 2074-9104 Volume 8, Number 12, December 2016

Contents REGULAR PAPERS Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial Sura M. Searan, Ali M. Sagheer

1

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3 Lakshman Naik.L, R.U.Khan, R.B.Mishra

9

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes Nisar Ahmed, Hafiz Muhammad Shahzad Asif, Gulshan Saleem

18

Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection Jaspreet Kaur, Parminder Singh

30

An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks Reza MotahariNasab, Ali Bohlooli, Neda Moghim

36

A Novel Technique to Prevent PUE Attack in Cognitive Radio Network Poonam, Ekta gupta, C.K. Nagpal

44

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator Richa Thakur, K.L. Bansal

51

Energy Efficient Clustering Protocol for Sensor Network Prachi, Shikha Sharma

59


I. J. Computer Network and Information Security, 2016, 12, 1-8 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.01

Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial Sura M. Searan University of Anbar/ Department of Computer Science, Baghdad, 10012, Iraq. E-mail: Surasms917@gmail.com

Ali M. Sagheer University of Anbar/ Department of Computer Science, Baghdad, 10012, Iraq. E-mail: ali.m.sagheer@gmail.com

Abstract—RC4 algorithm is one of the most significant stream and symmetric cryptographic algorithms, it is simple and used in various commercial products, it has many weaknesses such as a bias in the key stream that some key bytes are biased toward some values. In this paper, a new algorithm is proposed by using initial state factorial to solve the correlation issue between public known outputs of the internal state by using an additional state table with the same length as that of the state to contain the factorial of initial state elements. The analysis of RC4 and developed RC4 algorithm is done based on their single bias and double byte bias and shows that many keystream output bytes of RC4 are produced key stream bytes that are biased to many linear combinations while developed RC4 key bytes have no single and double biases. The results show that the series that is generated by developed RC4 is more random than that generated by RC4 and the developed algorithm is faster than RC4 execution time and requires less time. Additionally, the developed algorithm is robust against many attacks such as distinguishing attack. Index Terms—RC4, KSA (Key Scheduling Algorithm), PRGA (Pseudo-Random Generation Algorithm), Single Bias, Double Bias.

I.

INTRODUCTION

Encryption is a process that involves transforming plaintext into ciphertext in order to hide its meaning and to prevent unauthorized parties from retrieving plaintext [1]. The cryptographic algorithms are designed to provide lower size, high speed of implementation, less complexity, and a larger degree of security for resourceconstrained devices [2]. The strength of stream ciphers is the random key stream that guarantees secure computation of the cipher [3]. The cryptanalysis of stream cipher essentially focuses on identifying nonrandom process [4]. When the key size is small, it must be very efficient and encryption time be very fast, many encryptions that are used in wireless devices are based on symmetric key encryption such as RC4 algorithm [5]. RC4 is an effective stream cipher algorithm that is most popular. It is used in Oracle, SQL, Secure Sockets Layer, Copyright © 2016 MECS

and Wired Equivalent Privacy Protocol [6]. The attack on this algorithm was presented by Fluhrer, Mantin, and Shamir, it is an algorithm to use the symmetric key and it is an important one of the encryption algorithms [7]. This algorithm includes two main components to generate the key, the first is (KSA) Key Scheduling Algorithm and the other is (PRGA) Pseudo-Random Generation Algorithm [8]. RC4 starts with the permutation and uses the secret key with a variable length from 1 to 256 bits against a 256-bit state table [9]. The key is limited to 40 bits because of missing of restrictions, but it is sometimes used as 128 key bits [10]. Symmetric encryption can be classified into stream and block ciphers [11]. RC4 is analyzed by different people and there are different weaknesses detected [12]. KSA is more problematic and it is prepared to be simple [13]. At the beginning, few bytes of the output of PRNG are biased or related to some key bytes [14]. There are different types of attack that are classified by the amounts of information available to the adversary for cryptanalysis based on available resources [15]. The aim of this work is to solve interconnection between public known outputs of the internal state of RC4. The rest of the paper is arranged as follows. Section 2 gives the related works. Section 3 shows a brief depiction of RC4 algorithm. Several weaknesses of RC4 are determined in section 4. Section 5 illustrates the Modified RC4 by using Two State Tables and Initial State Factorial. Sections 6 and 7 show the implementation, results and discussion. In the results, section A shows a comparison between analyzing of RC4 and developed RC4 with factorial based on single byte bias. And section B shows a comparison between analyzing of RC4 and developed RC4 with factorial based on double byte bias. Conclusion is shown in section 8.

II.

RELATED WORKS

Mantin I. and Shamir A. (2001) showed an essential statistical weakness in the RC4 keystream by analyzing RC4 algorithm. This weakness makes it insignificant to discriminate between random strings and short outputs of RC4 by analyzing the second bytes. It is observed that

I.J. Computer Network and Information Security, 2016, 12, 1-8


2

Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

the second output byte of RC4 has a very strong bias that takes the value 0 with twice the expected likelihood (1/128 instead of 1/256 for n = 8). The main result is the detection of a slight distinguisher between the RC4 and random ciphers, that needs only two output words under many hundred unrelated and unknown keys to make robust decision [1]. Sepehrdad P. et al. (2011) discovered new biases in RC4 and proposed a mechanism to focus on linear attachments in the RC4 and detected linear attachment in the PRGA of the RC4 algorithm, and presented the elements within one trial of PRGA. Then this way is popularized to RC4 as a black crate with confident keywords as an input and words of the keystream as an output. These mechanisms lead to the detection of 57 attachments in the RC4. Some of these can be immediately utilized in the present key retrieval attacks on the RC4, WEP, and WPA [3]. Al-Fardan N. J. et al. (2013) measured the security of RC4 in TLS and WPA and analyzed RC4 based on its single and double byte bias and attacked it based on its two types of bias by using plaintext recovery attack. Their results show that there are biases in the first 256 bytes of the RC4 keystream that can be exploited by passive attacks to retrieve the plaintext by using 2 44 random keys. They focused on the multi-session setting, where the same plaintext is repeatedly encrypted with different keys and could recover full plain text from cipher text by using single byte bias attack and double byte bias attack [12]. Hammood M. M. et al. (2015) presented enhancing RC4 security and speed. Many algorithms were proposed as development for RC4. The first is RRC4 (RC4-Random initial state) which is to make RC4 more secure by increasing its randomness. The second suggestion is RC4 with two state tables to increase the randomness in the key sequence and the execution time of RC4-2State is faster than that of RC4. The last suggestion is RC42State + which is to produce 4 keys in every cycle to improve the randomness in the key sequence. The output sequences of all suggested algorithms provide more randomness [13].

key is limited to 40 bits but it is sometimes used as a 128 bit key [8] .It has the receptivity to use 1 to 2048 key bits .In general, the RC4 key size is “5 to 16 bytes” (40 to 128 bits) and the size of the typical state is 256 bytes [7]. It is highly utilized on the internet. It is used popularly as a default cipher for “Secure Socket/Transport Layer Security” (SSL/TLS) connections. It is very fast and simple in design and it is a set of stream systems represented by n that shows the word size in bits. The inner state is array (State) of (2n words) [8]. Algorithm 1. KSA

INPUT: Key OUTPUT: State 1. For (i = 0 to 255) 1.1 State[i] = i 2. Set j = 0 3. For (i = 0 to 255) 3.1 j = (j + State[i] + Key [i mod key-length]) mod 256 3.2 Swap(State[i], State[j]) 4. Output: State

The second step is pseudo-random algorithm. It generates the output keystream

generation

Algorithm 2. PRGA

INPUT: State, Plaintext i OUTPUT: Key sequence (K sequence) 1. Initialization: 1.1 i=0 1.2 j=0 2. For (i = 0 to Plaintext length) 2.1 i = (i + 1) mod N 2.2 j = (j + State[i]) mod N 2.3 Swap(State[i], State[j]) 2.4 K sequence = State [State[i] + State[j]] mod N 3. Output: K sequence

The output sequence of key K is XOR-ed with the Plaintext III. RC4 DESCRIPTION The RC4 algorithm was proposed by Ron Rivest in1987 and kept secret as a trade secret until it was leaked in 1994. It is very fast and simple in design [16]. The internal state is an array S of (2n words) [17]. RC4 has a variable length of the key that ranges between (0 255) bytes for initializing an array of 256 bytes in the initial state (State [0] to State [255]) [2].RC4 is carried out in two phases: The first is the key scheduling algorithm (KSA). It initializes the internal state [7]. RC4 starts the replacements and uses private key to get a random replacement with the KSA .Based on confidential key, the other phase is PRGA that produces key bytes that are XOR-ed with original bytes to get the cipher [5] .The state array is used to produce pseudorandom bits .These are done in the KSA .The operation which is performed between key and plain text is equivalent in some regard to the Vernam cipher [6] .The Copyright © 2016 MECS

Ci = Ki ⊕Plaintext i [4].

IV. THE WEAKNESS OF RC4 There are several weaknesses found in RC4 algorithm. Some of these weaknesses are easy and can be resolved, but other weakness is dangerous because attackers can exploit it. Another weakness in initialization state is a statistical bias which occurs in distributing words of the first output [7]. The key stream which begins the algorithm swaps the entry of the s-box exactly one time (identical to the pointer i that points to an entry) for low values of i, it is probable that Sj = j during the initialization [18]. Roos found weaknesses in RC4 that has serious correlation between generated value and the first few values of the state table. The first byte of the generated key is highly correlated with a few key bytes.

I.J. Computer Network and Information Security, 2016, 12, 1-8


Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

So, the keys allow precursor of first bytes from the output of the PRGA [19]. The goal of the attack is to retrieve the original key, the internal state, or the output keystream to have an access to the original messages. From the previous studies based on KSA and PRGA, there are some weaknesses of RC4 such as the biased bytes, distinguishers, key collisions, and key recovery from the state [20]. PRGA is reversible in nature. Then it is very simple to retrieve the secret key from the state [21]. Mantin and Shamir detected the major weakness in the second round the probability of zero output bytes. Fluher found a large weakness, if anyone knows the portion of the private key then its potential is to attack the RC4. Paul and Maitra found the secret key by using initial state table [22]. There are various methods of applying a brute force attack to the RC4 that is composed of two types: KSA attacks and PRGA attacks [23]. The vulnerabilities include A broadcast attack by using various unique keys for the encryption on similar plaintext offers the redundancy of the cipher text. The vulnerabilities include that PRGA is utilized as a seed for the RC4 which can be weak by analyzing a large collection of cipher text and as a result got duplicated encryption of the same plaintext with the same key [24]. For making RC4 secure and capable of standing against the attack, lot of researches was done over RC4 to enhance the security of RC4.

V.

Algorithm 3. KSA of developed RC4 with Factorial

INPUT: Key[i]. OUTPUT: State [i]. 1. For (i = 0 to 255) State[i] = i. 2. State_Fact[i] = Factorial(i) mod 256 3. j=0 4. For (i = 0 to 255) 4.1 j = (State_Fact[i] + State[i] + Key [i mod key-length]) mod 256 4.2 Swap (State[i], State[j]) 5. Output: State [i].

The second is PRGA which generates the output keystream: Algorithm 4. PRGA of developed RC4 with Factorial

INPUT: State [i], Plain i. OUTPUT: Key sequence (Key seq.) 1. Set i = 0, j = 0 2. Output Generation loop 2.1 i = (i + 1) mod 256 2.2 j = (State [(j + state[i] ) mod 256 ]) mod 256 2.3 Swap (State[i], (State-Fact [j] mod 256)) 2.4 Z = (State[(i + j) mod 256] + State[(j + State[state[i]]) mod 256]) mod 256 2.5 Key sequence = State [Z] 3. Output: Key sequence.

Cipher i = Key seq. ⊕ Plain i.

THE MODIFIED RC4 BY USING TWO STATE TABLES AND INITIAL STATE FACTORIAL

RC4 has various weaknesses in the KSA and PRGA that cause vulnerablity to this algorithm. This section determines the new insertion to the RC4 algorithm to improve it and to solve the weak key problem by using two state tables, one of them contains the factorial of other state contents with the same length to reduce the weakness that is exploited by the attacks. This algorithm consists of initialization step (KSA) and another step (PRGA) as shown in Algorithms (3) and (4). All addition operations are implemented in mod state length (N). The first step (KSA) takes a secret key k with a variable length between 1 and 256 n-bit words. In the first step of the KSA, one of the state tables is filled by the factorial of the contents of the other state table that is generated by the sender and filled with numbers from 0 to N-1. The input is the secret key used as a state table seed. After the KSA, the state becomes input to the next step (PRGA). In the PRGA step, additional operations are used as permutation to the state table. This phase generates the keystream that is XOR-ed with the plaintext to get the ciphertext.

Copyright Š 2016 MECS

3

VI. IMPLEMENTATION This algorithm is executed by using C# language. The inputs to this algorithm are an initial state that is filled with the values from 0 to 255 and secret key with a length between 1 and 256, and another state table with 256 bytes to contain the factorial of initial state elements. The implementation of the proposed algorithm required less time than that required for implementation of RC4 when implemented on the same size of secret keys and showed that the proposed algorithm is faster than RC4 as shown below. The table 1 and figure 1 below show the time of key generation for RC4 and the proposed algorithm Table 1. Key Generation Time for RC4 and Developed RC4 with Initial State Factorial. Key size

RC4 Time (m. s.)

1 kilobytes 2 kilobytes 3 kilobytes 5 kilobytes

4185 4237 4711 6899

RC4 with Factorial Time (m.s.) 4091 4191 4213 6372

I.J. Computer Network and Information Security, 2016, 12, 1-8


4

Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

other test accepted both large and small sizes .In our program, a large size (2,000,000 bits) is generated from each secret key .these sequences are tested, and p-values average which resulted from these tests are calculated as shown in table 2, in the test, P-value is compared to 0.01, the p-values are passed when they are greater than 0.01, and the produced series is random, and uniformly distributed .If the tests give p-value equal to 1, then the series is taken to have complete randomness. a p-value of zero means that the sequence is fully nonrandom. The SUCCESS means that the sequence is acceptable and it has good randomness, where FAILURE indicates that it is not acceptable and not-random. In general, these sixteen tests are composed of two collections. The first is called non-parameterized test and include Frequency Test, Cumulative Sums Test (forward and reverse), Discrete Fourier Transform (Spectral) Test, Lempel-Ziv compression Test, test for Longest Run of Ones in a Block, Rank Test, Runs Test, Random Excursions Test, and Random Excursions Variant Test. The second collection is called a parameterized test, it includes Serial Test, Linear Complexity Test, Overlapping Template of All One's Test, Non-overlapping Template Matching Test, Approximate Entropy Test, Block Frequency Test, and Universal Statistical Test.

Fig.1. Implementation Time of RC4 and Developed RC4.

VII. RESULTS AND DISCUSSION The generated key stream is examined by NIST (National Institute of Standards and Technology) Test Suite that is a statistical combination for random number generator test that includes 16 statistical tests for measuring the output series randomness of pseudorandom number or true random number generators .The tests of this PRNG were done by using NIST STS1.6 .The likelihood of a good random number generator is represented by P-value .Some tests accepted large sequence sizes and failed in the small sequence size, and

Table 2. Result of Running NIST on the Generated Key by RC4 and the Proposed RC4. RC4 Test No.

RC4 with Factorial

Statistical Test Name P-VALUE

Conclusion

P-VALUE

Conclusion

1

Approximate Entropy

0.805578

SUCCESS

0.195979

SUCCESS

2

Block Frequency

0.742455

SUCCESS

0.990906

SUCCESS

3

Cumulative Sums (Forward)

0.739164

SUCCESS

0.829138

SUCCESS

4

Cumulative Sum (Reverse)

0.854066

SUCCESS

0.716066

SUCCESS

5

FFT

0.279715

SUCCESS

0.556777

SUCCESS

6

Frequency

0.898580

SUCCESS

0.582269

SUCCESS

7

Lempel-Ziv compression

0.889521

SUCCESS

0.730735

SUCCESS

8

Linear Complexity

0.407918

SUCCESS

0.828157

SUCCESS

9

Longest Runs

0.767817

SUCCESS

0.985925

SUCCESS

10

Non periodic Templates

0.5407084

SUCCESS

0.527034

SUCCESS

11

Overlapping Template

0.497550

SUCCESS

0.614743

SUCCESS

12

Random Excursions

395805.0

SUCCESS

395.05.0

SUCCESS

13

Random Excursion Variant

395855.5

SUCCESS

39028350

SUCCESS

14

Rank

0.610871

SUCCESS

0.144541

SUCCESS

15

Runs

0.115965

SUCCESS

0.574394

SUCCESS

16

Serial

0.646168

SUCCESS

0.583851

SUCCESS

17

Universal Statistical

39003020

SUCCESS

39503500

SUCCESS

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 1-8


Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

5

This algorithm is implemented in C# language. Several biases were identified in the previous researches .RC4 is successfully reproduced and proved these bias in the first 32 bytes of keystream while developed RC4 has no bias in the first 32 positions of keystream byte. The algorithm of measuring key distribution bytes is implemented with 234 secret key as shown in figures 3,4,5 and 6. Measuring distributions of keystream bytes Z1 for 2 34 for RC4 and the proposed algorithms 0.0318 RC4 RC4 with factorial

0.0316 0.0314

Frequents

0.0312 0.031 0.0308

Fig.2. NIST Statistical Test for RC4 and Developed RC4.

0.0306

A. Comparison Between Analyses of RC4 and Developed RC4 with Factorial Based on Single Byte Bias.

0.0302

0

15 20 Positions 0 ,1,2,....31

25

30

Measuring distributions of keystream bytes Z2 for 2 34 for RC4 and the proposed algorithms RC4 RC4 with factorial

0.06 0.055 0.05 0.045 0.04 0.035 0.03 0.025

Copyright Š 2016 MECS

10

0.065

Algorithm 5. Measuring Distributions of RC4 Keystream Bytes Based on Its Single Bias

0

5

10

15 20 Values 0 ,1,2,....31

25

30

Fig.4. Key Distribution in the 2nd Position with 221 for RC4 and Developed RC4. Measuring distributions of keystream bytes Z16 for 2 34 for RC4 and the proposed algorithms 0.0322 RC4 0.032

RC4 with factorial

0.0318 0.0316 Frequents

INPUT: K [k1, k2, ‌., k16]. OUTPUT: Key position (Kp), key value (Kv), and the number of frequencies (Kf) for each position of key stream bytes. 1. For (x = 1 to 234) Do 1.1 i = 0, j = 0 1.2 Call Algorithm 1: KSA. 1.3 Call Algorithm 2: PRGA. 1.4 Deducting new key with a length of 16 bytes from each generated key are to be new secret key. 2. For (col = 0 to key Length) 2.1 For (row = 0 to 234) Set key [row] [col] as string 2.2 For (i = 1 to values. Count) 2.2.1 If (values [i] = value) 2.2.2 Increment count by 1 2.2.3 Key position = col 2.2.4 Key value = value 2.2.5 Number of frequencies = (count / (234 * 16)) 3. Output: Kp, Kv, and Kf for each position of key stream bytes.

5

Fig.3. Key Distribution in the 1st Position with 221 for RC4 and Developed RC4.

Frequents

RC4 has many weaknesses in the generated key stream, the key stream bytes are biased. In 2002 Mantin and Shamir [1] found that in the second round, the key is biased toward zero with high probability. AL-Fardan et al. also Hamood M.M. et al. analyzed RC4 based on its bias and proved that the first 256 bytes are biased. In this work, RC4 and the new algorithm were also analyzed and proved that developed RC4 has no bias while RC4 key stream is biased and shows the same bias that is proved in the literature, this work reduces the search space and uses state with length 32 and 230 random secret keys each one with length 16 and proved bias in standard time (less than one hour) as shown below. Expected biases start appearing for runtime 221 key generation. No bias is statistically identified for the proposed algorithm. Singlebias is calculated by using the following algorithm:

0.0304

0.0314 0.0312 0.031 0.0308 0.0306 0.0304

0

5

10

15 Values 0 ,1,2,....31

20

25

30

Fig.5. Key Distribution in the 16th Position with 221 for RC4 and Developed RC4.

I.J. Computer Network and Information Security, 2016, 12, 1-8


6

Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

Algorithm 6. Measuring Distributions of RC4 Keystream Bytes Based on Its Double Bias

Measuring distributions of keystream bytes Z32 for 2 34 for RC4 and the proposed algorithms 0.0314 0.0312

INPUT: K [k1, k2, …., k16]. OUTPUT: 3-Dimentions array. 1. i = j = i1 = k = 0 2. For (x = 1 to 210) 2.1 Call Algorithm 1. KSA 2.2 For (x = 1 to 232) 2.2.1 i = (i + 1) mod 256 2.2.2 j = (j + State[i]) mod 256 2.2.3 Swap (State[i], State[j]) 2.2.4 Generated Key = State[(State[i] + State[j]) mod 256] 2.2.5 A[k][Generated Key][i1] = A[k][Generated Key][i1] +1 2.2.6 Deducting new key with 16 bytes from each generated key to be new secret key. 2.2.7 k = Generated Key 2.2.8 i1 = (i1 + 1) mod 256 3. Output: A[k][Generated Key][i1].

0.031

Frequents

0.0308 0.0306 0.0304 0.0302 0.03 0.0298 RC4 0.0296

RC4 with factorial 0

5

10

15 20 Values 0 ,1,2,....31

25

30

Fig.6. Key Distribution in the 32nd Position with 221 for RC4 and Developed RC4.

B. Comparison Between Analyses of RC4 and Developed RC4 with Factorial Based on Double Byte Bias.

The figure below shows the distribution of (Zr, Zr+1) for all the first 32 bytes where Zr = i and Zr+1 = i for RC4. Double-byte biases (Zr, Zr+1) where Zr=i and Zr+1=i for RC4

0.0345 0.0335

Bias

0.0325 0.0315 0.0305 0.0295 0.0285 1000

800

600

400

200

0

0

5

Second byte values 0,1,2,.....,1023

15

10

20

25

30

Positions 0 ,1,2,....31

Fig.7. Double-Byte Biases (Zr, Zr+1) for RC4 where Zr=i and Zr+1=i.

The figure below shows the double-byte biases (Zr, Zr+1) for modified RC4 with factorial where Zr=i and Zr+1=i for the first 32 bytes. Double-byte biases (Zr, Zr+1) where Zr=0 and Zr+1=0 for RC4 with Factorial

1

0.5

Bias

After explaining single-byte biases that are of great benefit to the cryptographic society, the attack simply can be avoided by ignoring the initial bytes. Thus, RC4 with additional configuration can still resist single-byte bias attack. However, the researchers have studied and investigated biases beyond initial bytes and different multi-byte biases have been discovered in the key stream of RC4. Fluhrer and McGrew [25] were the first researchers that discovered the biases in a consecutive pair of bytes (Ki, Ki+1) and detected long-term biases of RC4. Hamood et al. [26] estimated the probability of the cipher for generating each pair of byte values through each 256-byte cycles and got a complete view of the distributions of every pair of byte values at the positions (i, i + 1). They replicated biases of Fluhrer and McGrew and their work was endorsed by AlFardan et al. [12] They found two new positive biases not mentioned in [25] by Fluhrer and McGrew. This work reproduced the Fluhrer and McGrew biases and Hammood bias with 1024 keys of 16 bytes to generate 232 keystream bytes after discarding the first 1024 bytes. Each key from the 1024 keys generates 2 32; therefore, the whole amount of generated keys is 2 42. The proposed algorithms does not generate any statistical bias and its output in the range only ±24 from the predicted occurrences. Algorithm 6 below is designed to determine the measure of double byte bias. The main idea of this algorithm is to measure the appearance of the consecutive pair (Zi, Zi+1) in each position of the output of RC4. The measure of double byte bias is illustrated in the algorithm below:

0

-0.5

-1 1000 800

25

600 10

200 0 Second byte valyes 0,1,2,.....,1023

30

20

400 0

15

5 Positions 0 ,1,2,....31

Fig.8. Double-Byte Biases (Zr, Zr+1) for Modified RC4 with Factorial where Zr=i and Zr+1=i.

Copyright © 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 1-8


Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

VIII. CONCLUSION RC4 stream cipher is a significant encryption algorithm and it is one of the widely used cryptosystems on the Internet that is used to keep information privacy. RC4 implementation is simple and fast compared with other encryption algorithms, but its key bytes are biased, this weakness makes RC4 vulnerable to attack. The proposed algorithm uses factorial of the state table contents and addition operations in KSA and PRGA to increase the randomness of the generated key while the key generation time of suggested algorithm is faster than key generation time of RC4. The key stream of the proposed algorithm has no single or double bias in the first 32 position as we reduce the search space and measured distribution bytes in standard time while RC4 key stream is biased in different positions. The generated key stream of the proposed RC4 has passed the NIST suite of statistical tests. Thus, it can be executed in the software or hardware.

[13]

[14]

[15]

[16]

[17]

[18]

REFERENCES I. Mantin and A. Shamir, “A Practical Attack on Broadcast RC4”. Springer, Lecture Notes in Computer Science. 2002, (2355), pp 152-164. [2] M. M. Hammood, K. Yoshigoe, and A. M. Sagheer, “RC4-2S: RC4 Stream Cipher with Two State Tables”. Springer, Lecture Notes in Electrical Engineering, 2013, 1, pp 13-20. [3] P. Sepehrdad, S. Vaudenay, and M. Vuagnoux, M. “Discovery and Exploitation of New Biases in RC4”, Springer, In Selected Areas in Cryptography, 2011, pp. 74-91. [4] M. E. McKague, “Design and analysis of RC4-like stream ciphers”, MS.C. Thesis, University of Waterloo, Canada/ Ontario, 2005. [5] P. Prasithsangaree, and P. Krishnamurthy, “Analysis of energy consumption of RC4 and AES algorithms in wireless LANs”, In Global Telecommunications Conference, 2003. GLOBECOM'03. IEEE, 2003, 3, pp 1445-1449. [6] A. M. S. Rahma, A. M. Sagheer, and A. A. Salih, “Development of RC4 Stream Ciphers Using Boolean Functions”, Journal of Baghdad College of Economic Sciences University, 2012, 29. [7] L. Stosic, and M. Bogdanovic, “RC4 stream cipher and possible attacks on WEP”, Editorial Preface, 2012, 3(3). [8] S. Maitra, and G. Paul, “New Form of Permutation Bias and Secret Key Leakage in Keystream Bytes of RC4”, In Fast Software Encryption, Springer, 2008, pp. 253-269. [9] S. Maitra, and G. Paul, “Analysis of RC4 and Proposal of Additional Layers for Better Security Margin”, Springer, Lecture Notes in Computer Science. 2008, pp 27-39. [10] C. Garman, K. G. Paterson, and T. Van der Merwe, “Attacks Only Get Better: Password Recovery Attacks Against RC4 in TLS”, In Presented as Part of the 24th USENIX Security Symposium (USENIX Security 15), 2015. [11] S. Paul, and B. Preneel, “Analysis of Non-Fortuitous Predictive States of the RC4 Keystream Generator”, Springer, Lecture Notes in Computer Science. 2003, pp 52-67. [12] N. J. Al-Fardan, D. J. Bernstein, K. G. Paterson, B. Poettering, and J. C. Schuldt, “On the Security of RC4 in [1]

Copyright © 2016 MECS

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

7

TLS and WPA”, In Presented as part of the 22nd USENIX Security Symposium. USENIX, 2013, 13, pp 305-320. M. M. Hammood, K. Yoshigoe, and A. M. Sagheer, “Enhancing Security and Speed of RC4”, International Journal of Computing and Network Technology, 2015, 3(2). L. L. Khine, “A New Variant of RC4 Stream Cipher”, Mandalay, Myanmar: World Academy of Science, Engineering and Technology, 2009. M. U. Bokhari, S. Alam, F. S. Masoodi, “Cryptanalysis Techniques for Stream Cipher: a survey”, International Journal of Computer Applications, 2012, 60(9), pp 29-33. K. K. H. Wong, G. Carter, and E. Dawson, “An analysis of the RC4 Family of Stream Ciphers Against Algebraic Attacks”, In Proceedings of the Eighth Australasian Conference on Information Security, 2010, 105, pp 67-74. M. A. Orumiehchiha, J. Pieprzyk, E. Shakour, and R. Steinfeld, “Cryptanalysis of RC4 (n, m) Stream Cipher”, In Proceedings of the 6th International Conference on Security of Information and Networks, 2013, pp 165-172. S. Mister, and S. Tavares, “Cryptanalysis of RC4-like Ciphers”, In Selected Areas in Cryptography, 1999, pp 632-632. Springer. M. M. Hammood, K. Yoshigoe, and A. M. Sagheer, “RC4 Stream Cipher with a Random Initial State”, Springer, Lecture Notes in Electrical Engineering, 2013, 1, pp 407416. P. Jindal, and B. Singh, “A Survey on RC4 Stream Cipher”, International Journal Computer Network and Information Security, 2015, 7, pp 37-45. M. Robshaw, and O. Billet, “New Stream Cipher Designs: The eSTREAM Finalists”, Springer, Lecture Notes in Computer Science, 2008. P. Pardeep, and P. K. Pateriya, “PC 1-RC4 and PC 2-RC4 Algorithms: Pragmatic Enrichment Algorithms to Enhance RC4 Stream Cipher Algorithm”, International Journal of Computer Science and Network, 2012, 1(3). M. Omari, and H. S. Soliman, “Exponential Brute-Force Complexity of a Permutation Based Stream Cipher”, International Journal Computer Network and Information Security, 2013, 1, pp 1-13. V. K. Keerthi, and R. P. Arun, “Taxonomy of SSL/TLS Attacks”, International Journal Computer Network and Information Security, 2016, 2, pp 15-24. S. R. Fluhrer, and D. A. McGrew, “Statistical Analysis of the Alleged RC4 Keystream Generator”, Springer, Lecture notes in computer science, 2001, (1978), pp 1930. M. M. Hammood, and K. Yoshigoe, “Previously Overlooked Bias Signatures for RC4”, International Symposium on Digital, Forensic Security, 2016, 101-106. doi:10.1109.

Authors’ Profiles Sura M. Searan has received her B.Sc. in Computer Science (2013) from the University of Anbar, Iraq. She is a master student (2014, till now) in the Computer Science Department, College of Computer Sciences and Information Technology at Al-Anbar University. She is interested in the following fields; Cryptology, Information Security, Coding Systems.

I.J. Computer Network and Information Security, 2016, 12, 1-8


8

Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial

Ali M. Sagheer is a Professor in the Computer College at Al-Anbar University. He received his B.Sc. in Information System (2001), M.Sc. in Data Security (2004), and his Ph.D. in Computer Science (2007) from the University of Technology, Baghdad, Iraq. He is interested in the following fields; Cryptology, Information Security, Number Theory, Multimedia Compression, Image Processing, Coding Systems, and Artificial Intelligence. He has published many papers in different scientific journals.

How to cite this paper: Sura M. Searan, Ali M. Sagheer,"Modification of RC4 Algorithm by using Two State Tables and Initial State Factorial", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.1-8, 2016.DOI: 10.5815/ijcnis.2016.12.01

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 1-8


I. J. Computer Network and Information Security, 2016, 12, 9-17 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.02

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3 Lakshman Naik.L Indian Institute of Technology (Banaras Hindu University)/Department of Electronics Engg., Varanasi, 221005, India E-mail: lnaikl@bhu.ac.in

R.U.Khan1 and R.B.Mishra2 Indian Institute of Technology (Banaras Hindu University)/Dept. of Electronics Engg.1, Dept. of Computer Science and Engg.2, Varanasi 221005, India E-mail: {rukhan.ece, mishravi.cse}@iitbhu.ac.in

Abstract—Networks which function without having any centralized fixed infrastructure or central administration are called MANETs (Mobile Ad hoc Networks). These networks are formed by small or large set of mobile nodes and communicate through the wireless links. Such Networks requires best routing protocols to establish error-free and efficient communication links. MANETs has the property of dynamically changing topology due to their mobile nodes, which move from one place to another. Overall performance of MANET routing protocols depends upon various network and protocol parameters. Mobile ad hoc networks have the characteristics of self-forming and self-healing. The routing algorithms of the routing protocols ensure selection of routes and connectivity between the mobile nodes. This paper presents analysis of three well known routing protocols of MANETs, namely; AODV (Ad hoc On Demand Distance Vector), DSDV (Destination Sequenced Distance Vector) and OLSR (Optimized Link State Routing). Analyses of these routing protocols have been carried out using NS-3 (Network Simulator-3) by varying node density and node pause time. Different performance metrics such as throughput, packet delivery ratio, end to end delay, packet loss and normalized routing load have been considered for this analysis. This analysis concludes better performance of the OLSR routing protocol. Index Terms—Throughput, Packet delivery ratio, End to end delay, Packet loss Routing, Simulation, NS-3 (Network Simulator – 3).

I. INTRODUCTION A mobile ad-hoc network often referred as MANET is a set of wireless nodes which move freely from one point to another without having any fixed infrastructure [1, 2]. These are self-forming and self-healing networks, nodes of such networks are mobile in nature, hence they acquire dynamic network topologies. In MANETs, any node can openly establish connection with the other nodes of the network with in the transmission range of the other nodes. Mobile ad hoc networks are lively subject of popular Copyright © 2016 MECS

researches because of their application in Wi-Fi/802.11 supported portable devices that become extensive [3]. The aim behind all such new research is to improve performances of MANETs by improving performances of their routing protocols with the help of routing algorithms. MANET survivability varies with different routing protocols. Their survivability is also depends on factors like; node density, node pause time, varied transmission power and node mobility speed etc. MANET routing protocols are designed to regulate efficient and error free route links between the mobile nodes. Advancement in technology have achieved performance improvements in small, mobile-wireless devices like laptops, mobile phones etc. [4]. Based on procedure of route discovery, these routing protocols are classified into three types; proactive or table-driven, reactive or on-demand and hybrid. Hybrid routing protocols are developed by combining the features of proactive and reactive routing protocols [5].

Fig.1. Mobile Ad-hoc Network

In mobile ad-hoc networks, mobile nodes communicate each other using multi hop wireless links. These networks are usually deployed for various diverse applications like; military networks, conference rooms and in commercial applications like vehicle ad-hoc networks [6]. Due to mobility nature of the nodes, the physical network topology of these networks often changes randomly. MANETs do not possesses any stationary infrastructure like access points or base stations, thus, every node acts as router. These routers forward the progressing data packets to all their neighboring nodes. The well-known MANET routing protocols are; AODV (Ad hoc On

I.J. Computer Network and Information Security, 2016, 12, 9-17


10

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

Demand Distance Vector), DSDV (Destination Sequenced Distance Vector), OLSR (Optimized Link State Routing), DSR (Dynamic Source Routing), and TORA (Temporally Ordered Routing Algorithm) [7, 8, 9, 10, 11, 12, 13]. Performance of these routing protocols depends upon various factors like; complex interplay of the protocol mechanisms and their specific parameter settings with traffic intensity, mobility, node density and conduct of the mobile wireless nodes. This paper presents simulation based analysis of AODV, DSDV and OLSR routing protocols with varied number of nodes and their pause time in different scenarios. Over the last few years, various MANET routing protocols have been developed and presented by the researchers. These researches improved their performance with respect to establishing error free and efficient routes between the nodes. We have carried out this analysis to study node density and node pause time effects in AODV, DSDV and OLSR. Our upcoming research will focus; presenting improved version of these routing protocols. This paper presents related works in section II, MANET routing protocols in section III, performance calculating matrices in section IV, simulation tools in section V, results and discussions in section VI and conclusions in section VII.

II. RELATED WORKS Many researchers have studied comparative analysis of MANET routing protocols and presented their results to the research community for further research. Most of them used NS-2 (Network Simulator-2) in their analysis. Some of such research works have been discussed here. Teressa Longjam and Neha Bagoria have studied comparative analysis of DSDV and AODV and declared better performance of DSDV as per their simulation scenarios [14]. Rakesh Kumar Jha and Pooja Kharga have studied comparative analysis of OLSR, AODV and DSR, their conclusion states better performance of AODV [15]. D.Kumar & S.C.Gupta have studied transmission range, density & speed based performance analysis of OLSR, DSR and ZRP (Zone Routing Protocol). Based on their analysis and network scenarios, they concluded better performance of DSR [16]. Rajneesh Kumar, Jitender Grover and Anjali have studied analysis of network survivability with variable transmission range and mobility on AODV over MANET, their conclusion declares that the AODV has better performance with highest node mobility speed and higher transmission ranges; in terms of network survival and QoS parameters [17]. Researchers Ali Khosrozadeh, Abolfazle Akbari, Maryam Bagheri and Neda Beikmahdavi have studied conventional AODV routing protocol, their analysis have presented a new algorithm of AODV to the research community [25]. Some researchers have worked on solution to the energy constraints in mobile ad-hoc network protocols. Researchers D.Loganathan, P.Ramamoorthy have studied DSDV routing protocol, they presented the multicast parameters based DSDV (MPB-DSDV) routing protocol to enhance the energy efficient of ad hoc networks [26]. Researchers Charles E. Copyright Š 2016 MECS

Perkins, T.J. Watson and Pravin Bhagwat have proposed an ad-hoc network which has supportive appointment of set of mobile nodes without the requisite involvement of any centralized base station or access point. Charles E. Perkins presented a state-of-the-art design for the ad-hoc networks operation. The indication behind the design is to make each mobile node as a specialized router. These nodes periodically announce their interconnection topology information with their neighbouring nodes and other mobile nodes within the network. These extents to a new kind of routing protocol in mobile ad-hoc networks. Charles E. Perkins has examined amendments to the basic Bellman-Ford routing mechanism, as quantified by conventional RIP (Routing Information Protocol), to mark it appropriate for a dynamic and self-starting network appliance for userâ€&#x;s choice. These modifications address looping issues of conventional Bellman-Ford techniques, and helpful for the broken links [27].

III. MANET ROUTING PROTOCOLS Here, in this paper, we present simulation based study and analysis of AODV, DSDV and OLSR routing protocols. These routing protocols are liable to establish the correct and efficient routes among the mobile nodes in MANETs. Routing is a process which discovers the error free and efficient routes between a source node and the destination node; it makes sure correct and timely delivery of data packets. A. Ad-hoc on demand distance vector (AODV) AODV is developed for mobile ad-hoc networks and other wireless ad-hoc networks. It is a reactive routing protocol; AODV was developed by C.Perkins, S.Das and E.Belding-Royer during July 2003 [18]. In ad-hoc on demand distance vector routing, discovery of route takes place subjected to route requests received from the neighboring nodes and other nodes in the network [19]. AODV maintains newest routing information by means of route discovery procedures and updated routing tables [20]. In AODV, path discovery takes place when a source node transmits RREQ (Route Request) message throughout the network until required destination reached.

Fig.2. Message Transmission in AODV

Upon receiving RREQ message, the destination node generates RREP (Route Reply) message for the source node to ensure the path. During path breaks, the destination node generates a RERR (Route Error) message and transmits it throughout the network so that

I.J. Computer Network and Information Security, 2016, 12, 9-17


Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

every node receives it. In Fig. 2, the source node „S‟ is transmitting RREQ message, whereas the destination node „D‟ is transmitting RREP message throughout the network of mobile nodes „N‟. The destination node „D‟ generates RERR message when path break occurs between the source node „S‟ and the destination node „D. B. Destination sequenced distance vector (DSDV) DSDV is one of the proactive routing protocols of the MANETs. DSDV is an altered version of DBF (Distributed Bellman Ford) technique. This technique is used to calculate the shortest path. The conventional DBF technique creates certain routing loops during the routing process. The DSDV was developed to suppress this looping problem with the help of DSN (Destination Sequence Number) [21]. DSDV is similar to the RIP (Routing Information Protocol), excepting the DBF technique. In DSDV, mobile nodes transmit updated routing information and incremented sequence number throughout the network. Route selection process is carried out by the distance vector shortest path algorithm. DSDV minimizes its transmission overheads by using two updated packets namely; “full dump” and “incremental dump”. The full dump packet holds the routing data, whereas the incremental dump holds only the changed data successively since the previous full dump.

11

N3, N4, N5, N6 and N7, where N2, N3, and N4 are the neighbor nodes of the mobile node N1.

IV. PERFORMANCE METRICS Various metrics are available to analyze the performances of the MANET routing protocols. Following metrics are taken into account for our simulation based analysis [15]. (1) Throughput Throughput is the total data transmitted from the source node to the destination node in a time unit which is expressed in Kilobits per second (Kbps). (1) Unit of throughput is Kbps. Higher values of the throughput provide better performance. (2) Packet Delivery Ratio (PDR) Packet Delivery Ratio is the fraction of amount of received packets to the amount of sent packets.

(2)

C. Optimized Link State Routing (OLSR) OLSR is one of the proactive protocols of the mobile ad hoc networks. OLSR was developed based on link state routing algorithm; OLSR uses optimized technique to extract information pertaining to the network topology [22]. In optimized link state routing, when there is a change in network topology occurs, flooding of information to all the network nodes happens. These flooding are minimized by the help of MPR (Multi Point Relays). Table-driven nature of the OLSR helps to broadcast updated routing tables to all the mobile nodes of the network. Various control messages are used in OLSR routing protocol; HELLO, TC (Topology Control), HNA (Host and Network Association) and MID (Multiple Interface Declaration). The OLSR broadcasts these control messages periodically that is why OLSR does not require usage of control message delivery. This aids OLSR to have reasonable losses in control messages.

PDR is derived in % (percentage). Higher values of PDR provide better performance. (3) End to End Delay (EED) EED is the average time interval between packets generated at the source and effective delivery of these packets at the destination. EED is the fraction of delay sum to the packets received.

(3) EED is derived in mille seconds (ms). Lower values of EED provide better performance. (4) Packet Loss (PL) Packet loss is the difference of total packets sent and the total packets received.

(4) Packet loss is derived in number of packets. (5) Normalized Routing Load (NRL) NRL is the ratio of the numbers of transmitted routing packets to the number of packets received [23]. Fig.3. Control Message Transmission in OLSR

(5)

Fig.3 illustrates the processing of TC message from the node N1 to the network of seven mobile nodes N1, N2, Copyright © 2016 MECS

Larger values of NRL provide better and enhanced

I.J. Computer Network and Information Security, 2016, 12, 9-17


12

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

performance however, higher values of NRL leads to reduced efficiency in terms of ingestion of bandwidth.

V. NETWORK SIMULATOR -3 (NS -3) We have used NS-3(Network Simulator-3) version 3.13 for our simulation based experiments. NS-3 is an open source discrete-event based network simulating software developed specially for research and educational purposes. NS-3 is licensed under GNU GPLv2 license and it is publicly available for research and development. The NS-3 project builds a solid simulation core, easy to use and debug. “NS-3 core caters the needs of the simulation workflow, from the simulation configuration to trace collection and analysis. The NS-3 simulation core supports research on both IP and non-IP based networks” [24]. Majority of NS-3 users emphases on wireless/IP simulations. NS-3 is developed using C++ high level programming language with the optional python bindings. NS-3 has enhanced simulation reliability. It is not backward adjusted with NS-2 (Network Simulator-2). NS-3 is built from the scratch to replace application program interfaces (APIs) of NS-2. Some modules of NS2 have been ported in it. NS-3 does not support APIs of

NS-2 [15]. NS-3 supports real-time schedulers which simplifies number of “simulations-in-the-loop”. Packets generated by the NS-3 can be emitted and receive on real network devices. NS-3 is aligned with the simulation needs of modern networking research.

VI. RESULTS AND DISCUSSIONS Simulation based experiments and performance comparison of mobile ad-hoc network routing protocols (AODV, DSDV and OLSR) have been carried out in two different scenarios, in first scenario (SS-I), we varied node density and in another (SS-II), varied pause time have been taken into account. The simulation scenarios and obtained results are illustrated in the following tables and graphs. A. Simulation Scenario - I (SS-I) General Network parameters that have been taken into account for simulation scenario-I are mentioned in the Table 1. In SS-I, number of nodes have been varied keeping 10 number of source/sink connection fixed.

Table 1. Network Parameters for SS-I 1

Number of Nodes

30,40,50,60,70,80,90,100

2 3 4 5 6

Simulation Time Pause Time Wi-Fi mode Wi-Fi Rate Transmit Power

150 seconds No pause time Ad-hoc 2Mbps (802.11b) 7.5 dBm,

7

Mobility model

Random Waypoint mobility model

8

No.of Source/Sink

10

9

Sent Data Rate

2048 bits per second (2.048Kbps)

10 11 12 13 14

Packet Size Node Speed Protocols used Region Loss Model

64 Bytes 20 m/s AODV,DSDV and OLSR 300x1500 m Friis loss model

calculate the throughput as per throughput metrics, results so obtained are mentioned in Table 2.

(1) Throughput Captured experimental packet data have been used to

Table 2. Throughput in Kbps No. of Nodes 30 40 50 60 70 80 90 100

AODV 16.04 17.93 14.47 1.87 9.73 11.62 0.68 1.42

DSDV 14.95 14.3 12.64 14.87 14.58 15.6 13.47 13.58

Fig. 4 explores performance of AODV, DSDV and OLSR in terms of average throughput with the increasing Copyright © 2016 MECS

OLSR 18.27 16.93 17.99 18.91 18.99 18.60 17.47 18.45

node density. OLSR has shown better performance as compared to AODV and DSDV. AODV has performed

I.J. Computer Network and Information Security, 2016, 12, 9-17


Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

better than DSDV for smaller number of nodes however, DSDV has shown better performance for large number of nodes.

13

Table 4. End to End Delay in Mille Seconds No. of Nodes

AODV

DSDV

OLSR

30 40 50

6.17 2.89 9.56

8.45 9.96 14.55

2.37 4.53 2.8

60 70

242.38 26.41

8.62 9.30

1.44 1.33

80 90

18.04 706.71

7.06 12.13

1.88 3.62

100

327.94

11.83

2.10

OLSR has shown better results as compare to the rest two routing protocols however, when comparing AODV and DSDV, the AODV has performed fine for smaller number of nodes and the DSDV has shown better performance for higher number of nodes.

Fig.4. Throughput Over No. of Nodes

(2) Packet Delivery Ratio (PDR) Data of Table 3 have been extracted from the packet data obtained from the experiments and the metrics of the packet delivery ratio. Table 3. Packet Delivery Ratio in % No. of Nodes

AODV

DSDV

OLSR

30 40 50 60 70 80 90 100

80.22 89.63 72.33 9.35 48.63 58.08 3.42 7.08

74.73 71.52 63.22 74.35 72.88 77.98 67.33 67.88

91.33 84.67 89.93 94.55 94.97 93.00 87.37 92.25

Fig.6. EED Over No. of Nodes

(4) Packet Loss (PL) Here, OLSR has least number of packet losses as compare to AODV and DSDV. Table 5 explores the PL of all the three routing protocols.

From the results, packet delivery ratio of the OLSR routing protocol is found better than that of AODV and DSDV. Here, AODV has performed well for smaller number of nodes but, DSDV has shown better results for higher number of nodes.

Table 5. Packet Loss in No. of Packets No. of Nodes

AODV

DSDV

OLSR

30 40 50 60 70 80 90 100

1187 622 1660 5439 3082 2515 5795 5575

1516 1709 2207 1539 1627 1321 1960 1927

520 920 604 327 302 420 758 465

When comparing AODV and DSDV, AODV has lesser packet losses for smaller number of nodes and the DSDV has less number of these losses for higher number of nodes. Fig.5. PDR Over No. of Nodes

(3) End to End Delay (EED) Table 4 shows the data sheet of end to end delay calculated from the experimental data and the metrics. Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 9-17


14

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

Table 7. Network Parameters for SS-II 1

Number of Nodes

50

2

Simulation Time

150 seconds

3

Pause Time (Seconds)

5,10,15,20,25,30

4

Wi-Fi mode

Ad-hoc

5

Wi-Fi Rate

2Mbps (802.11b)

6

Transmit Power

7.5 dBm,

7

Mobility model

Random Waypoint mobility model

8

No. of Source/Sink

9

Sent Data Rate

10

Packet Size

11

Node Speed

Fig.7. Packet loss Over No. of Nodes

(5) Normalized Routing Load (NRL) NRL data sheet shows better performance of the OLSR as compare to AODV and DSDV.

10 2048 bits per second (2.048Kbps) 64 Bytes 20 m/s

Table 6. NRL 12

Protocols used

AODV,DSDV and OLSR

No. of Nodes

AODV

DSDV

OLSR

13

Region

300x1500 m

30 40 50 60 70 80 90 100

0.802 0.896 0.723 0.094 0.486 0.581 0.034 0.071

0.747 0.715 0.632 0.744 0.729 0.78 0.673 0.679

0.913 0.847 0.899 0.946 0.95 0.93 0.874 0.923

14

Loss Model

Friis loss model

(1) Throughput Here, average throughput of the OLSR is better found as compare to AODV and DSDV routing protocols. Table 8. Throughput in Kbps

Like in other metrics discussed above, AODV has better values for smaller number of nodes and the DSDV has performed well for higher number of nodes.

Pause Time in Seconds 5 10 15 20 25 30

AODV

DSDV

OLSR

2.88 5.20 13.55 16.27 16.42 14.21

12.96 13.25 14.31 12.75 12.21 12.76

17.91 17.80 17.48 18.74 18.69 18.27

When comparing AODV and DSDV, AODV is found better performing for higher node pause times, whereas DSDV has shown better results for smaller values of the node pause time.

Fig.8. NRL Over No. of Nodes

B. Simulation Scenario - II (SS-II) General Network parameters for SS-II have been chosen as per Table - 3.7. In SS-II, pause times of nodes have been varied while keeping 10 numbers of fixed source/sink connections. Random waypoint mobility model has been considered for this experiment.

Copyright Š 2016 MECS

Fig.9. Throughput Over Pause Time

I.J. Computer Network and Information Security, 2016, 12, 9-17


Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

15

(2) Packet Delivery Ratio (PDR) As per data sheet shown in Table 9, OLSR protocol has shown better results as compare to AODV and DSDV. Table 9. Packet Delivery Ratio in % Pause Time in Seconds 5 10 15 20 25 30

AODV

DSDV

OLSR

14.42 25.98 67.77 81.37 82.10 71.07

64.82 66.23 71.53 63.73 61.03 63.78

89.55 89.02 87.40 93.72 93.45 91.35

Fig.11. EED Over Pause Time

When comparing AODV and DSDV, AODV has shown better performance for higher node pause times, whereas DSDV has shown better results for the lower pause times.

(4) Packet Loss (PL) As compare to AODV and DSDV, OLSR has lesser packet losses for the node pause time variation. Table 11. Packet loss in No. of packets Pause Time in Seconds 5 10 15 20 25 30

AODV

DSDV

OLSR

5135 4441 1934 1118 1074 1736

2111 2026 1708 2176 2338 2173

627 659 756 377 393 519

When Comparing AODV and DSDV, DSDV has lesser packet losses for lesser pause time values and AODV has lesser packet losses for higher node pause times.

Fig.10. PDR Over Pause Time

(3) End to End Delay (EED) Here, among all the three routing protocols, OLSR has shown better performance in terms of delay. Table 10. EED in Mille Seconds Pause Time in Seconds 5 10 15 20 25 30

AODV

DSDV

OLSR

148.410 71.220 11.890 5.730 5.450 10.180

13.570 12.750 9.950 14.230 15.960 14.200

2.920 3.080 3.600 1.680 1.750 2.370

Fig.12. PL Over Pause Time

AODV has larger delays for lesser pause times but it has lesser delay values for higher node pause times. DSDV has lesser delay values for lesser node pause times as compare to AODV.

Copyright Š 2016 MECS

(5) Normalized Routing Load (NRL) NRL values of OLSR protocol have shown better results as compare to AODV and DSDV routing protocols.

I.J. Computer Network and Information Security, 2016, 12, 9-17


16

Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

like, transmit power, no. of source/sink connections, node density, node velocity, transmission region, transmission range, type of load traffic, Wi-Fi rate and packet size etc. Future research can be carried out by varying the network parameters as well as protocol parameters for further improvement in the MANET routing protocols.

Table 12. Normalized Routing Load Pause Time in Seconds 5 10 15 20 25 30

AODV

DSDV

OLSR

0.144 0.260 0.678 0.814 0.821 0.711

0.648 0.662 0.715 0.637 0.610 0.638

0.896 0.890 0.874 0.937 0.935 0.914

ACKNOWLEDGMENT

When comparing AODV and DSDV, AODV has performed better for higher pause times and the DSDV has better results for lesser node pause times. Fig. 13 illustrates the better performance of the OLSR routing protocol as compare to AODV and DSDV. The green line in the graph shows the performance of the OLSR routing protocol, likewise; blue line represents the performance of the AODV protocol and the brown line represents the performance of the DSDV routing protocol. These lines represent performance of all the three routing protocols with respect to increasing node density and the node pause times.

Fig.13. NRL Over Pause Time

VII. CONCLUSIONS As per the simulation based experiments in both the scenarios (Scenario-I & Scenario-II), it is noticed that the performance of the OLSR routing protocol is better in all the metric calculations as compare to the DSDV and AODV routing protocols. In scenario-I, throughput of the OLSR is better, whereas AODV and DSDV are concerned, initially throughput of the AODV is better, but after certain point of time, it decreases. DSDV is performing well as compare to AODV, in terms of throughput. In rest metrics too, OLSR has good results as compare to AODV and DSDV. In scenario-II, OLSR is performing well as compared to rest two routing protocols. As far as AODV and DSDV are concerned, DSDV performed well as compare to AODV in all the metrics we used here. However, in some cases, performance of the DSDV routing protocol is found improved. In some other cases, performance of AODV is found well. These conclusions are totally based on the NS-3, version 3.13, and the network parameters that we set for our analysis. However, performance of the MANET routing protocols depends on various factors Copyright © 2016 MECS

We thank Dr.R.S.Yadava and Dr.Chandan Kumar Rai, Programmers of Computer Centre, Banaras Hindu University, Varanasi, India, for their kind support and encouragement. REFERENCES [1]

Mobile ad hoc networks (MANET). http: //www.ietf.org/html.charters/manet-charter.html, IETF Working Group Charter 1997. [2] Anuj K. Gupta, Harsh Sadawarti, Anil K. Verma, “Review of various Routing Protocols for MANETs”, International Journal of Information & Electrical Engineering, Article No. 40, 1(3): 251-259, November 2011. [3] E.M. Royer, C-K. Toh, “A Review of Current Routing Protocols for Ad Hoc Mobile Wireless Networks”, IEEE Personal Communications Magazine, April 1999, pp. 4655.Kredo, K. and P. Mohapatra, Medium access control in wireless sensor networks. Computer Networks, 2007. 51(4): p. 961-994. [4] H. Tafazolli, “A Survey of QoS Routing Solutions for Mobile Ad Hoc Networks”, IEEE Communications Surveys & Tutorials, Vol. 9, No. 2, pp. 50–70, 2007. [5] Arunima Patel, Sharda Patel, Ashok Verma “A Review of performance Evaluation of AODV Protocol in Manet With and Without Black HoleAttack” International Journal of Emerging Technology and Advanced Engineering ISSN 22502459,Volume 2, Issue 11, November 2012. [6] H. Pucha, S. M. Das, and Y. C. Hu, "The performance impact of traffic patterns on routing protocols in mobile ad hoc networks," Computer Networks, vol. 51, no. 12, pp. 359 5-3616, 2007. [7] Huhtonen, "Comparing AODV and OLSR routing protocols," presented at Telecommunications Software and Multimedia, pp. 1- 9, 2004. [8] M. Benzaid, P. Minet, and K. AI Agha, "Integrating fast mobility in the OLSR routing protocol," presented at the fourth IEEE Conference in Mobile and Wireless Communications, pp. 1- 5, 2002. [9] G. Jayakumar and G. Gopinath, "Performance comparison of two ondemand routing protocols for ad-hoc networks based on random way point mobility model," American Journal of Applied SCiences, vol. 5, no. 6,pp. 659664,2008. [10] N. Taing, S. Thipchaksurat, R. Varakulsiripunth, and H. Ishii, "Performance improvement of dynamic source routing protocol for multimedia services in mobile ad hoc network," presented at the 1st International Symposium on Wireless Pervasive Computing, pp. 1- 5, 2006. [11] F. D. Rango, J.-c. Cano, M. Fotino, C. Calafate, P. Manzoni, and S. Marano, "OLSR vs DSR: A comparative analysis of proactive and reactive mechanisms from an energetic point of view in wireless ad hoc networks," Computer Communications, vol. 31, no. 16, pp. 38433854, 2008.

I.J. Computer Network and Information Security, 2016, 12, 9-17


Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3

[12] S. Giannoulis, C. Antonopoulos, E. Topalis, and S. Koubias, "ZRP versus DSR and TORA: A comprehensive survey on ZRP performance," presented at the 10th IEEE Conference on Emerging Technologies and Factory Automation (ETFA ' 05), pp. 1-8, 2005. [13] F. Yu, Y. Li, F. Fang, and Q. Chen, "A new TORA-based energy aware routing protocol in mobile ad hoc networks," presented at the 3rd IEEE/IFIP International Conference in Central Asia on Internet, ICI ' 07, pp. 1-4, 2007. [14] Teressa Longjam and Neha Bagoria. “Comparative Study of Destination Sequenced Distance Vector and Ad-hoc on-demand Distance Vector Routing Protocol of Mobile Ad-hoc Network”. IJSRP, vol. 3, Issue 2, February 2013. [15] Rakesh Kumar Jha, Pooja Kharga. “A Comparative Performance Analysis of Routing Protocols in MANET using NS3 Simulator”, in IJCNIS – MEC Publications, vol. 4 pp.62-68. March, 2015. [16] D.Kumar & S.C.Gupta. “Transmission Range, Density & Speed based Performance Analysis of Ad Hoc Networks”. African Journal of Computing & ICT. Vol. 8.No. 1-March, 2015. [17] Rajneesh Kumar Gujral, Jitender Grover, Anjali. “An Analysis of Network Survivability with Variable Transmission Range and Mobility on AODV over MANET”. TECHNICA, vol. 5, No.2, Jan, 2013. [18] [Online].Available:https://en.wikipedia.org/wiki/Ad_hoc_ On Demand_Distance_Vector_Routing. [19] Vinay P.Virada “Securing And Preventing Aodv Routing Protocol From Black Hole Attack Using Counter Algorithm” International Journal ofEngineering Research & Technology (IJERT) Vol. 1 Issue 8, October 2012 ISSN: 22780181. [20] AshishBagwari, Raman Jee,Pankaj Joshi and Sourabh Bisht “ Performance of AODV Routing Protocol with increasing the MANET Nodes and its effects on QoS of Mobile Ad hoc Networks ” 2012 International Conference on Communication Systems and Network Technologies 9780769546926/12 © 2012 IEEE. [21] Sreekanth Vakati, Dr.Ch.Balaswamy, “Performance Analysis of Routing Protocols in Mobile Ad Hoc Networks”, July 2013. [22] Dilpreet Kaur, Naresh Kumar, “Comparative Analysis of AODV, OLSR, TORA, DSR and DSDV Routing protocols in Mobile Ad-Hoc Networks,” in IJCNIS journal, vol.5, no.3, pp.39, 2013. [23] Qutaiba Razouqi, Ahmed Boushehri, Mohamed Gaballah, Lina Alsaleh, “Extensive Simulation Performance Analysis for DSDV, DSR, and AODV MANET Routing Protocols. IEEE 2013. [24] [Online].Available: https://www.nsnam.org/overview/what-is-ns-3/ [25] Ali Khosrozadeh, Abolfazle Akbari, Maryam Bagheri and Neda Beikmahdavi, “A New Algorithm AODV Routing Protocol in Mobile ADHOC Networks”, published in International Journal of Latest Trends in computing, vol.2, No.3, September 2011. [26] D.Loganathan, P.Ramamoorthy, “Performance Analysis of Enhanced DSDV Protocol for Efficient Routing In Wireless Ad Hoc Networks”, published in International

17

Journal Of Engineering And Science Vol.2, Issue 10 pp. 01-08, April 2013. [27] Charles E.Perkins, T.J. Watson, Pravin Bhagwat. Proceeding SIGCOMM '94 Proceedings of the conference on Communications architectures, protocols and applications Pages 234-244, ACM New York, NY, USA ©1994.

Authors’ Profiles Lakshman Naik.L received Bachelor of Engineering degree in Electronics and Communication Engineering from Gulbarga University, Gulbarga, India. Currently, he has been pursuing Ph.D., in Indian Institute of Technology (Banaras Hindu University), Varanasi, India. Now, he is working as a Maintenance Engineer at Computer Centre of the Banaras Hindu University, in Varanasi, India. He has more than 10 years of experience in maintaining computer Networks, having 5 years of research experience in Engineering and Educational fields. His research interest includes Computer Networks, Wired and Wireless Communication, Mobile Ad hoc Networks, and Network Routing Protocols. He has published several research papers in various international journals.

Dr. R.U.Khan is an Associate Professor in Department of Electronics Engineering, Indian Institute of Technology (Banaras Hindu University), Varanasi, India. He received B.Tech, M.Tech and Ph.D. degree from Banaras Hindu University, Varanasi, India. He has more than 36 years of experience in teaching. His research interests focus on Microwave solid State Devices, Microelectronics, Opto-electronics devices and Computer Networks. He is guiding for several research scholars under Indian Institute of Technology (Banaras Hindu University), Varanasi, India. He has been published many research articles in National, International Journals and conferences.

Prof. R.B. Mishra is a Professor in Department of Computer Science and Engineering, Indian Institute of Technology (Banaras Hindu University), Varanasi, India. He received B.Sc.( Engg.), M.Tech (Control) and Ph.D., degree from Banaras Hindu University, Varanasi, India. He has more than 35 years of experience in teaching. His research interests focus on Artificial Intelligence, Multi-agent Systems, Semantic Web and Computer Networks. He has guided 16 Ph.D scholars under Indian Institute of Technology (Banaras Hindu University), Varanasi, India. He has been published more than 230 research articles in Journals and conferences.

How to cite this paper: Lakshman Naik.L, R.U.Khan, R.B.Mishra,"Analysis of Node Density and Pause Time Effects in MANET Routing Protocols using NS-3", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.9-17, 2016.DOI: 10.5815/ijcnis.2016.12.02 Copyright © 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 9-17


I. J. Computer Network and Information Security, 2016, 12, 18-29 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.03

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes Nisar Ahmed, Hafiz Muhammad Shahzad Asif Department of Computer Science and Engineering, University of Engineering and Technology Lahore, Pakistan. E-mail: nisarahmedrana@yahoo.com

Gulshan Saleem Department of Computer Engineering, EME College, National University of Science and Technology, Pakistan. E-mail: gulshan.saleem14@ceme.nust.edu.pk

Abstract—Digital images have become part of everyday life by demonstrating its usability in a variety of fields from education to space research. Confidentiality and security of digital images have grown significantly with increasing trend of information interchange over the public channel. Cryptography can be used as a successful technique to prevent image data from unauthorized access. Keeping the nature of image data in mind, several encryption techniques are presented specifically for digital images, in literature during past few years. These cryptographic algorithms lack a benchmark for evaluation of their performance, cryptographic security and quality analysis of recovered images. In this study, we have designed and developed a benchmark based on all the parameters necessary for a good image encryption scheme. Extensive studies have been made to categories all the parameters used by different researchers to evaluate their algorithms and an optimum benchmark for evaluation is formulated. This benchmark is used to evaluate three image encryption schemes. The results of evaluation have highlighted the specific application areas for these image encryption schemes. Index Terms—Image encryption, cryptographic security, encryption performance, cryptographic benchmark. I. INTRODUCTION Advances in technology have affected cheap access to digital storage and multimedia processing and capturing devices. Multimedia capturing devices are not restricted to cameras or camcorder but smartphones, laptop’s, tablets and other devices or everyday use are equipped with digital cameras. Moreover, access to the free or cheap internet, 3G, and 4G cellular networks has caused a large increase in internet users. These unsecured public networks are used frequently for multimedia communication. Wireless communication, on the other Manuscript received July 20, 2016;

Copyright © 2016 MECS

hand, is a big troublemaker in security. Satellite communication or other wireless technologies provide wire-free access to remote terminals through VSAT and other technologies. With the increasing trend of communication over public channel and growth of digital multimedia devices, the need for methods to protect this data from unauthorized access is becoming crucial. Three common methods are used for security of digital images from unauthorized access or copyright violation. Image cryptography is a technique, which allows visual information to be transformed into such a form that encrypted image become unintelligible. Steganography and watermarking, on the other hand, does not encrypt the actual image but hide some other media in an image in such a way that it becomes invisible. Digital steganography encodes a secret message (pictures, text, etc.) that it becomes imperceptible for others. The image may seem to be a simple photograph but it contains an invisible secret message. The discovery of this secret message by an investigator is very difficult and can be recovered by intended recipient having the embedding algorithm and secret key. Digital watermarking is much similar to steganography in working but its application is different. It focuses on authentication of digital media and protection of intellectual property rights. A watermark image is inserted into a cover image, which is later detected or identified for copyright claim or authentication purpose. Watermarking provides the way to ensure intellectual property rights and keep track of the quick and inexpensive distribution of digital media over the internet. Digital image encryption transforms an input plaintext image to an output ciphertext image through the cryptographic algorithm with the help of a secret key. The ciphertext image is not usable unless the decryption algorithm and secret key are available. There are numerous cryptographic algorithms and their categorization is made based on certain parameters. Two categories of the cryptographic algorithm based on secret key are; private key cipher and public key cipher [1]. In private-key cipher, the secret key is same for encryption and decryption processes. Private-key ciphers are also

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

known as symmetric cipher. While in public-key cipher, the secret key is different for encryption and decryption processes and are not related to one another [2]. The secret key used for encryption is made public so anyone can perform encryption but only the intended recipient having the secret key can decrypt the image. This type of cipher is also known as non-symmetric cipher [3-4]. There are three categories of image encryption algorithms based on the mechanism of operation. Transposition based cipher simply work with rearrangement of pixels with a complex regular system. It has been demonstrated that all type of permutation only cipher, which works with rearrangement of pixel position, can be broken [2]. However, it can be combined with other techniques to make it more complex and cryptographically secure. Visual transformation, on the other hand, encrypts images by dividing them into several shares (layers). These shades are positioned mechanically in such a way to reveal the original image (message). This same technique is extended to digital images for visual encryption and decryption process requires all the shares and their exact orientation for decryption. This way of encryption has very limited application and mainly used for binary images. Value transformation based cipher has the diversity of encryption schemes. They work by modifying the gray value of pixels either in transform domain or in the spatial domain. Spatial domain based techniques operate at the bit level to change pixel value. This bit level change may be through shuffling the pixel bits or changing the quantization. Transform domain methods involve operation in DCT, DFT or DWT domain operation on coefficient. Popular techniques discussed in the literature are either value transformation based or the hybrid of above mentioned encryption methods. Owning to the fact, numbers of encryption schemes presented in the literature are not tested for all parameters of cryptographic security and performance. A decent encryption scheme must fulfill all the security requirements of an image cryptosystem. The performance should be comparable to other proposed schemes or must be acceptable with respect to a particular application. Moreover, image encryption schemes are characterized distinctly from text, as they have to take account of the redundancy in images. Several image encryption schemes take benefit of this redundancy and encrypt the image in such a way that decrypted image is not the exact replica of input plaintext image. This recovered image is perceptually similar to the plaintext image but it may have minor changes. So a benchmark for quality analysis of recovered image should be established which would be helpful in comparison. A.

Types of Cryptographic Attacks

Ciphertext-only Attack Ciphertext-only attack is a cryptanalysis method where the attacker has access to only a set of ciphertext. The attack is considered successful if the attacker is able to deduce the key or even the plaintext. Copyright Š 2016 MECS

19

Known Plaintext Attack Known plaintext attack is a cryptanalysis method in which the attacker has information of a set of plaintext and their corresponding ciphertext. These types of attacks are more successful in the deduction of secret keys. Chosen Plaintext Attack Chosen plaintext attack is a category of cryptanalysis in which the attacker has access to the encryption scheme as a black box. In this way, the attacker can get ciphertext of any random plaintext. The goal of such attacks is to deduce the relationship of plaintext to ciphertext by providing specific plaintext. Chosen Ciphertext Only Attack Chosen ciphertext only attacks are mostly used in public key cryptosystems. In this attack model, the attacker can choose a ciphertext and get its corresponding decrypted plaintext. Brute Force Attack Brute force attack is employed when the attacker is unable to get any advantage of other weaknesses. It is also referred to as, exhaustive key search, as it practically checks all the possible keys for decryption. This attack can be theoretically used against any ciphertext but the limitation arises with the computational time required to perform an exhaustive key search. Section-II presents the image encryption schemes containing three image encryption schemes which are to be analyzed for security and performance assessment. Section-III presents the metric of image quality describing four image quality measurement metrics along with tabulated and graphical representation of their testing on the selected image encryption schemes. Section-IV presents the metrics to evaluate the cryptographic security of the image encryption schemes. The section discusses the information entropy analysis, correlation coefficient analysis in the planner and 3-D view. The differential analysis presents the measurement of avalanche effect, mean squared error, number of pixels change rate, universal average change intensity, dispersion test analysis. The statistical analysis presents histogram analysis, maximum and irregular deviation measurements. Keyspace analysis presents exhaustive key search and key sensitivity test. Robustness tests include tamper detection, compression friendliness and noise tolerance. Section-VI draws the conclusion.

II. IMAGE ENCRYPTION SCHEMES Numerous image encryption techniques are presented in the literature with surprising characteristics. These techniques lack evaluation on common criteria. Three of these techniques are chosen for evaluation. First of these ciphers is AES based block cipher with demonstrated cryptographic security. Second, one is a compression and noise tolerant cipher and the third is a chaos-based image cipher with high randomness and unpredictability.

I.J. Computer Network and Information Security, 2016, 12, 18-29


20

A.

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

A. Normalized Correction

Advanced Encryption Standard

National Institute of Standards and Technology (NIST) selected Rijndael as Advanced Encryption Standard (AES) in 2011 [5]. The selection of AES was a tradeoff between performance, efficiency and overall security. It replaced the Data Encryption Standard (DES) and Triple-DES due to their weaker security against brute force attacks. It is a new generation symmetric block cipher with key sizes of 128, 192 and 256 bits. It is a linear transformation substitution cipher, which uses triple discrete invertible uniform transformations. It has a high degree of modular design, making it possible to counter any future attack mechanism or to introduce development. The algorithm has outperformed in 15 candidates for AES but has received criticism by some researchers due to its security. These criticisms are theoretically valid as the other algorithms provide better security but it does not mean that AES encrypted data is vulnerable to attack. Although it is not the most secure cipher but its security can be increased by adding more rounds.

It is a function of time lag to measure the similarity between two images. A numerical value of 1 indicates an identical image and deviation from unity indicates the difference between the two images. The formula for calculation of NC is provided below. (

)

∑∑

(

)

(

)

∑ (

(

))

(1)

B. Compression and Noise Tolerant Image Encryption Scheme

Fig.1. Normalized Correction Measurements between Original and Recovered Images by Cipher2.

Nisar et. al [6] proposed a compression and noise tolerant image encryption scheme. They have used orthogonal basis vectors to process the image to introduce confusion in the algorithm. The image is separated into 16×16 blocks and these blocks are permuted. These permuted blocks are DCT transformed and multiplied with orthogonal vectors generated from Singular Value Transformation (SVD) of a randomly generated matrix. The resultant cipher image has the horizontal correlation, which allows it for lossy compression.

Fig. 1 shows the result of normalized correction for Cipher2 [6] for seven test images used for analysis. Numerical values of NC for Cipher1 are presented in table 2. The results of NC for Cipher1 [5] and Cipher3 [7] were unity so their numerical results are provided in table 1.

C.

Chaos-based Image Encryption Scheme

Ruisong Ye [7] has used generalized Bernoulli shift maps to permute the image pixel position and change the grayscale values. Two chaotic orbits are generated for the permutation of image pixels and diffusion of image grayscale value. The first chaotic sequence is used to get an index sequence to permute the pixel positions. The second chaotic sequence is generated from generalized for of Bernoulli shift map by setting the initial values. Every cipher pixel of cipher image is obtained by taking XOR of plain image pixel with randomly generated pixel (through Bernoulli shift map) and the previous cipher image pixel multiplied by the mod of gray-level. The cipher has demonstrated high-security characteristics with a large key space.

B. Correlation Measure The correlation coefficient can be used to measure the similarity between two images. It measures the crosscorrelation between pixels of original and recovered. This test can provide numerical results to quantize the similarity measure and the graphical results will demonstrate the same correlation graphically. A diagonal line of points will indicate identical image, the spreading of points above and below this line will designate the amount of variance between two images. Fig. 2 shows the cross-correlation of plaintext image and recovered image by Cipher2 [6]. Numerical results of cross-correlation for the other images generated by Cipher2 [6] are provided in table 2 and table 1 provides the similar values for Cipher1 [5] and Cipher3 [7].

III. IMAGE QUALITY METRIC There are some image encryption schemes which doesn’t exactly reproduce the decrypted image and add slight distortion which is tolerable in some conditions if the visual quality of the image is not significantly degraded. Image quality metric describes the metrics which can be used to quantities the recovered image quality as compared to the original image. Copyright © 2016 MECS

Fig.2. Cross-Correlation between the Original Image (Archer) and Recovered Image with Cipher2.

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

21

C. Mean Squared Error

E. Structural Similarity Index

MSE provides the mean of the squares of the differences of the corresponding pixels of two images. It provides a numerical value of distortion in the recovered image. Below formula is used to calculate MSE between original and recovered image.

SSIM is intended to improve the similarity measure based on human visual perception on traditional methods such as PSNR and MSE. It differs from other techniques as it considers image quality degradation as observed variance in structural information. SSIM is based on the idea that the pixels have a strong relationship with its neighbors and this relationship has important information about the structure of objects. Moreover, SSIM is only applied to luminosity layer of the TrueColor image as all the structural information is contained in this layer. SSIM can be calculated from the following formula.

∑(

)

(2)

(

)

( (

)( )(

)

(4)

)

Where

Fig.3. The Mean Squared Error between Original Images and Recovered Images with Cipher2.

Fig. 3 provides the results of MSE between original and recovered image for Cipher2 [6]. Numerical values of it are provided in table 2. Table 1 provides the numerical values of MSE for Cipher1 [5] and Cipher3 [7]. D. Peak Signal-to-Noise Ratio (PSNR)

the average of & the average of . the variance of & the variance of the covariance of and ( ) ( ) two variables to stabilize the division with weak denominator the dynamic range of the pixel values and by default. Numerical results of SSIM are used for similarity evaluation, higher value indicates more similarity and a value of 1 is achieved in the case of identical images.

PSNR provides the peak of error between two images. It is an estimator for human visual perception of reconstruction quality. It is the most commonly used metric to check the recovered image quality. In some situation, PSNR may not produce actual results correlating with human visual perception [8]. We can calculate PSNR by the following formula. (

)

(3)

Fig. 4 shows the result of PSNR for original and recovered image by Cipher2 [6], the same is provided numerically in table 2. Table 1 provides the values of PSNR for Cipher1 [5] and Cipher3 [7].

Fig.5. SSIM for Original and Recovered Images with Cipher2.

Fig. 5 shows the results of SSIM for original and recovered image by Cipher2 [6]. Numerical values of the same are provided in table 2. Table 1 provides the numerical values of SSIM for Cipher1 [5] and Cipher3 [7]. Table 1. Numerical values of NC, CC, MSE, PSNR and SSIM for test images for Cipher1 and Cipher3. Image Archer Flower Glider Kodim15 Lena Mandrill Peppers

NC 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

CC 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

MSE 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

PSNR Inf Inf Inf Inf Inf Inf Inf

SSIM 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

Fig.4. PSNR for Original and Recovered Images with Cipher2.

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 18-29


22

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

Table 2. Numerical values of NC, CC, MSE, PSNR and SSIM for test images for cipher2. Image Archer Flower Glider Kodim15 Lena Mandrill Peppers

NC 1.0004 1.1164 1.0025 1.0022 .9416 1.0342 1.0639

CC 0.9978 0.9917 0.9978 0.9958 0.9963 0.9688 0.9976

MSE 22.4941 113.3242 16.3242 23.8359 243.2791 190.0615 91.4748

PSNR 34.6101 27.5983 36.0025 34.3585 24.2698 25.3419 28.5178

SSIM 0.9857 0.9715 0.9896 0.9809 0.9443 0.9246 0.9883

IV. BENCHMARK FOR CRYPTOGRAPHIC SECURITY EVALUATION

Fig.6. Histogram of Local Entropies for Cipher1: Archer Image

Visual examination of ciphertext image is the primary factor to quantify the encryption quality of an image encryption scheme. Nevertheless, visual examination is not enough to judge the quality of encryption. Thus, an evaluation benchmark is required to estimate the encryption quantitatively. An effective image encryption algorithm changes the pixel values in such a way to make it irregular. Thus, higher the change in pixel values, the more effective is the encryption. Following are the performance metric to evaluate the cryptographic security of encryption scheme. A. Information Entropy Analysis Information entropy is a mathematical parameter of information and coding theory, which reflects randomness and uncertainty of a source. It gives information about the source itself [9, 10]. It is an important concept for analyzing any cryptosystem as it measures its uncertainty and randomness. The entropy of a source can be calculated by following formula [11-15]: ( )

( )

( )

Table 3. Information Entropy Analysis of Three Encryption Schemes AES 7.9997 7.9998 7.9998 7.9998 7.9998 7.9998 7.9998

Copyright © 2016 MECS

FEA 7.1127 7.0432 7.0656 7.0415 6.9828 7.0925 7.1482

Moreover, the entropy of the source is not uniformly distributed so we have also calculated the local entropy. Local entropy is displayed graphically by calculating entropy for 16×16 blocks of cipher image and plotting their histogram.

(5)

Here, S is the source, P(Si) is the probability of occurrence of symbol Si, N is the number of bits to represent symbol Si. For an ideally random source with 2N symbols, the entropy is N. Therefore, for a grayscale ciphertext image, the ideal entropy should be 8. An actual information source is never actually random so its entropy value is smaller than the ideal one. However, in an actual cryptosystem, the entropy must be as closer to the ideal value as possible otherwise; it will threaten the security of the cryptosystem.

Archer Flower Glider Kodim15 Lena Mandrill Peppers

Fig.7. Histogram of Local Entropies for Cipher2: Archer Image

Chaos 7.9991 7.9991 7.9989 7.9991 7.9991 7.9991 7.9992

Fig.8. Histogram of Local Entropies for Cipher3: Archer Image

B. Correlation Coefficient Analysis Correlation determines the degree of similarity between two variables. It is used as an important metric to evaluate the quality of a cryptosystem [16, 17]. Natural images have a lot of correlation between their adjacent pixels as there are very few sharp edges [12]. The image cryptosystem is regarded as effective if it hides the original image content completely with the lowest correlation [11, 16, 18]. Correlation coefficient for an identical image is equal to one (-1 for negative image) and for a highly uncorrelated image is almost zero. Correlation of an images can be calculated in horizontally

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

adjacent pixels, vertically adjacent pixels and diagonally adjacent pixels. Mathematically, the correlation coefficient is calculated by below formulas [2, 11-13, 16, 17, 19, 20]. ( √

( )

(

)

( )

∑(

Fig. 9-11 provides the correlation plot of Archer image encrypted by the three ciphers. The plot of correlation between all the pixels in diagonal, horizontal and vertical directions are provided for comparison.

) ( )

(6)

( ))

( ))(

∑(

23

(7)

( ))

(8)

Here, C.C is correlation coefficient, x and y are the pixel values, Cov is the covariance between x and y, VAR(x) gives the value of variance at pixel value x, δx is standard deviation, N is the total number of pixels and E is expected value operator.

(a)

(b)

(a)

(c) Fig.10. Correlation Coefficient Analysis of Archer Image Encrypted Using Cipher2: (a) Correlation between Diagonally Adjacent Pixels (b) Correlation between Horizontally Adjacent Pixels (c) Correlation between Vertically Adjacent Pixels

(b)

(c) Fig.9. Correlation Coefficient Analysis of Archer Image Encrypted Using Cipher1: (a) Correlation between Diagonally Adjacent Pixels (b) Correlation between Horizontally Adjacent Pixels (c) Correlation between Vertically Adjacent Pixels

Copyright © 2016 MECS

(a)

I.J. Computer Network and Information Security, 2016, 12, 18-29


24

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

(c)

(d)

Fig.12. (a) Plain image (Archer) (b) Cipher image with Cipher1 (c) Cipher Image with Cipher2 (d) Cipher Image with Cipher3

C. Differential Analysis The differential analysis is based on the study of change in output pixels in response to a change in input pixels. This property of an image cryptosystem is referred as diffusion characteristics and was introduced by Shannon in his classical masterpiece in 1949 [9]. To withstand the differential cryptanalysis, a cryptosystem must ensure good diffusion characteristics. The output image should change entirely in an unpredictable manner for a change of single pixel of an input image. Following parameters are used to perform differential analysis of a cryptosystem to ensure good diffusion characteristics.

(b)

Avalanche Effect

(c) Fig.11. Correlation Coefficient Analysis of Archer Image Encrypted Using Cipher3: (a) Correlation between Diagonally Adjacent Pixels (b) Correlation between Horizontally Adjacent Pixels (c) Correlation between Vertically Adjacent Pixels

The gradient is another measure of image correlation. In a highly correlated image, the value of gradient will be very less and its 3D plot will be a plane surface. Fig. 12 provides the result of the gradient plot for plaintext Archer image and its corresponding cipher images by the three ciphers under test. Fig. 12 (a) provides the gradient map for plaintext cipher image which clearly shows homogeneous areas on the left. Fig. 12 (b) shows the same graph for cipher image of Cipher1, which indicates highly non-homogeneous distribution. Fig. 12 (c) on the other hand indicate homogeneity at some areas and nonhomogeneity at the other places. It accounts for the same correlation which is indicated in Fig. 10 (b). Fig. 12 (d) has the same plot as (b) but its color distribution is much wide and provides better no-homogeneity. Table 4. Numerical Results Of Correlation Coefficient Analysis For Archer And Kodim15 Images. Archer (diagonal) Archer (vertical) Archer (horizontal Kodim15 (diagonal) Kodim15 (vertical) Kodim15 (horizontal)

Cipher1 [3] 0.0018 0.0028 0.0094 0.0106 0.0086 0.0108

(a) Copyright Š 2016 MECS

Cipher2 [4] 0.0078 0.0035 0.9199 0.0473 0.0557 0.9143

(b)

Cipher3 [5] 0.0169 0.0097 0.0087 0.0055 0.0132 0.0007

Avalanche effect is used to measure the diffusion characteristic of an image cryptosystem, which is an important parameter that must be checked to verify the randomness and complexity of the system. The system is taken as a black box and one bit of the input plaintextimage is changed to observe the change in the output ciphertext-image. Small change in output image in response to 1-pixel changed input image will make it possible to construct a meaningful relationship between the two images. To avoid deduction of this relationship, the output image pixels of 1-pixel changed image must be more than 50% different from the original image. Let C1 is the cipher image with original plaintext image and C2 is the cipher image with a 1-pixel change in the plaintext image. Following are the test to measure the avalanche effect. Table 5. Avalanche Effect (MSE, NPCR, and UACI) Results for Cipher1 Archer Flower Glider Kodim15 Lena Mandrill Peppers

MSE 40.3882 40.3877 40.3994 40.3778 40.3755 40.3781 40.3837

NPCR 99.6216 99.6071 99.6147 99.6140 99.6040 99.6143 99.6403

UACI 33.5271 33.4798 33.5221 33.4387 33.4929 33.3709 33.4227

Table 6. Avalanche Effect (MSE, NPCR, and UACI) Results for Cipher2 Archer Flower Glider Kodim15 Lena Mandrill Peppers

MSE -16.0760 -15.8515 -16.7518 -15.5447 -17.7328 -16.7046 -15.8838

NPCR 99.9993 99.9985 99.9989 99.9985 99.9985 99.9969 99.9985

UACI 0.0129 0.1292 0.0459 0.0523 0.0408 0.0459 0.0502

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

Table 7. Avalanche Effect (MSE, NPCR, and UACI) Results For Cipher3 Archer Flower Glider Kodim15 Lena Mandrill Peppers

MSE 40.4000 40.3997 40.4338 40.4235 40.3978 40.4083 40.3758

NPCR 99.5972 99.6918 99.6735 99.6887 99.6338 9.6063 99.6078

UACI 33.3052 33.5325 33.7354 33.6324 33.6025 33.6323 33.4130

Mean Squared Error (MSE) Mean Squared Error is used to check the avalanche effect by calculating MSE between image C1 and C2 [19, 20]. indicates an evident difference between two images and their relationship is too complex to be predicted easily [23, 24]. Number of Pixel Change Rate (NPCR) The number of Pixel change rate is a test to measure the avalanche effect of an image cryptosystem. It measures the number of pixel difference between two cipher images C1 and C2. The theoretical critical value for this test is 99.6094% for 8-bit image [25]. ∑

(

)

25

[9], image ciphers can be attacked by statistical analysis. An image cipher should transform a meaningful and correlated image into a random looking image. Therefore, an image cipher should produce an encrypted image with uniform histogram distribution. Color image histogram is unlike the histogram of a grayscale image (intensity histogram). Usually, histogram for three RGB color channels is obtained separately and visually inspected for uniformity [12, 17]. Sometimes, the brightness is taken out by normalizing all the triplets and then plotting them sequentially. Aberration graphs are also used for image histogram analysis as they plot the intensity values in three dimensions [26]. The purpose of aberration graph can be served with a gradient map of Fig. 12. These techniques are simple but not effective for efficient histogram analysis of color images. A technique for drawing color histograms and color clouds originally developed for movie poster analysis by S.C. Gaddam [27] is presented here for histogram analysis of cipher images. Fig. 13-15 provides the color histogram for Archer image encrypted by the three ciphers. The histogram analysis of Cipher1 and Cipher3 fulfill the uniformity requirement whereas of Cipher2 is debatable.

(9)

Universal Average Change Intensity (UACI) Universal average change intensity measures the average intensity difference between the two images. Theoretical critical value for this test is 33.4635% [25]. *

|

(

)

(

)|

Fig.13. Color Histogram of Archer Image encrypted by Cipher1

+

(10)

Dispersion Test Analysis Dispersion test is performed to check the result of diffusion. A white image with a small black patch of and a black image with a small patch of white color are encrypted and the results of dispersion are checked in the output image. Table 8. Entropy Analysis of White Image (With 8x8 Black) and Black Image (With 8x8 White) Black Image White Image

Cipher1 7.9993 7.9994

Cipher2 0.0007 0.0012

Fig.14. Color Histogram of Archer Image encrypted by Cipher2

Cipher3 7.9968 7.9971

D. Statistical Analysis Histogram Analysis Image histogram shows the distribution information of pixel values and discloses statistical characteristics. It is regarded as an important statistical feature of an image and is taken as a metric for evaluation of the security of an image encryption scheme. In Shannon’s perspective Copyright © 2016 MECS

Fig.15. Color Histogram of Archer Image encrypted by Cipher3

I.J. Computer Network and Information Security, 2016, 12, 18-29


26

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

Maximum Deviation

E. Keyspace Analysis

A parameter to check the statistical security of encryption is the maximum deviation, which measures the deviation between pixel values of an original image and the encrypted image [16, 19]. Higher the value of maximum deviation more is the deviation in encrypted image from that of plaintext image. Below formula is used to calculate the value of maximum deviation.

Keyspace analysis is done to check robustness against brute force attacks. A good image encryption system should have large enough key space and high sensitivity to the secret key.

(11)

Here, di is the difference of histogram of the original image and that of cipher image at value i. d 0 and d255 are the difference values at index 0 and 255. Table 9. Results of Maximum Deviation for the Image Ciphers Archer Flower Glider Kodim15 Lena Mandrill Peppers

Cipher1 77357 122880 97164 110380 54894 44518 73422

Cipher2 67159 111430 77738 90674 33210 22014 44004

Encryption Scheme Cipher1 Cipher2 Cipher3

Key Size (bits) 128, 192, 256 128 312

Key Sensitivity Test

Maximum deviation alone is not enough to ensure statistical randomness of a ciphertext image. The encryption algorithm should randomly change the pixel values to become a statistically robust scheme [16, 19]. An algorithm, which makes a large change in some image pixel values and produces insignificant change in other, is not statistically secure. The procedure to calculate the value of irregular deviation is enlisted below. Take the histogram; say h, of absolute difference of plaintext image and ciphertext image. Calculate the mean value of h and name it Mh. Calculate the irregular deviation ID using the following formula: |

Key space size is the number of different keys, which can be used as secret key. Sufficiently large key space is necessary to prevent the execution of brute force attacks [13, 16]. Exhaustive key search is the number of operations required to check all the possible secret keys for decryption [16]. A cryptosystem with 256bits key will require 2256 number of operations to check all the keys. Table 12 provides the key space size for the three encryption schemes, which is large enough to be secure against brute-force attack. Table 11. Key size for the Encryption Schemes

Cipher3 48439 96096 55346 79296 22330 21718 23318

Irregular Deviation

∑|

Exhaustive Key Search

(12)

Key sensitivity is an extreme dependency on the exact key. This test measure, how much the cryptosystem is sensitive to small change in secret key [13]. A secure cryptosystem, even 1-bit change in the secret key would be enough to produce entirely different cipher image. Key sensitivity is checked in two different aspects: (a) completely different ciphertext images should be produced with 1-bit change in secret key, (b) 1-bit changed secret key should produce entirely random decryption image. The satisfactions of these two aspects of key sensitivity test are mandatory for the security of key space [13, 20]. F. Robustness Test Image robustness tests are used to check the dependence of decryption algorithm on the exact values of ciphertext image. The success of these tests indicates higher security but reduces the robustness to compression, noise, and small unintentional tampering.

A smaller value of ID indicates the histogram is close to uniformity and betters the statistical properties of encryption. Table 10. Results of Irregular Deviation for the Image Ciphers Archer Flower Glider Kodim15 Lena Mandrill Peppers

Cipher1 60404 64704 72376 59490 74780 81680 69736

Copyright © 2016 MECS

Cipher2 55292 53660 65086 56654 69920 74974 64428

Cipher3 47652 40128 53768 42396 59694 64776 56462

(a)

(b)

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

(c)

(d)

Fig.16. Image Decryption Results with 1bit changed Secret Key (a) Plaintext Archer Image (b) Decryption Results with Cipher1 (c) Decryption Results with Cipher2 (d) Decryption Results with Cipher3

Tamper Detection This test is performed to check the robustness against tempering in ciphertext image. It indicates high diffusion characteristics of an image cryptosystem. A small patch of 8Ă—8 is painted black in the cipher image and the decryption is performed to check the robustness against tampering [26]. The same can be done by changing the least significant bit of cipher image. Corresponding change between decrypted image and non-tempered decrypted image is checked for compliance [12].

(a)

27

same image. Huffman coding, run-length coding, arithmetic coding, LZW coding and Simplified MED are some types of lossless compression [29, 30]. The second type of compression is lossy compression, which reduces the amount of data that is not necessary for visual inspection. Under-sampling, reduced color maps, requantization and other such techniques are used as a mean of lossy compression. Lossy compression hence introduces small variations in the cipher image so an algorithm positive to cipher image sensitivity test or cut test will not be friendly to lossy compression. The compressed image produced with such encryption algorithm cannot be recovered with accuracy. If an algorithm produces good quality image after decryption of its compressed image and provides significant compression is said to be a compression friendly encryption algorithm. If the cipher image has high entropy, it would not result in compression as there is not enough correlated data to be compressed. In some cases, if a highly uncorrelated ciphertext image is compressed by JPEG it may result in an increase of image size [29, 31]. There are some encryption algorithms, which perform compression before or during the process [29].

(b) Fig.18. Compression Performance of Image Ciphers using Archer Image

(c)

(d)

Fig.17. Temper Detection Results with painted box (a) Original Archer Image (b) Recovered with Cipher1 (c) Recovered with Cipher2 (d) Recovered with Cipher3

Fig. 18 shows the reduction in cipher image size after JPEG compression. Highest compression ratio is achieved in the case of Cipher2 whereas Cipher3 provides minimum compression ratio. Fig. 19 provides the result of decryption after JPEG compression with QF of 90. Cipher1 fails to recover the image whereas Cipher2 has recovered the image with reasonable visual quality. Recovery with Cipher3 contain visible distortion but can be tolerated in special cases.

Compression Friendliness Image compression has vital importance in the field of cryptography. It reduces the transmission or storage bandwidth significantly making it a highly desirable property. Numerous image compression algorithms are in practices, which are developed, based on information theory [9]. There are two types of image compression methods; lossless and lossy. Lossless compression reduces the unnecessary redundancy in the image by reducing the required number of bits to represent the Copyright Š 2016 MECS

(a)

(b)

I.J. Computer Network and Information Security, 2016, 12, 18-29


28

A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

(c)

(d)

Fig.19. JPEG compressed image recovery results (QF=90%) (a) Archer Image (b) Cipher1 recovery (c) cipher2 recovery (d) cipher3 recovery

Noise Tolerance Image after encryption may go through a noisy channel and certain amount of noise can be introduced. Tolerance of the cryptosystem to such noise becomes a desirable property in some applications. It is true that such property will indicate some weakness in the encryption scheme but it can be dealt with that specific application. AWGN is added to the ciphertext image, it is decrypted with the exact decryption key, and its similarity with the uninterrupted recovered image is tested for compliance.

(a)

(b)

(c)

(d)

Fig.20. Noise Immunity Result with AWGN (mean 0, variance 0.01) (a) Archer Image (b) Cipher1 recovery (c) cipher2 recovery (d) cipher3 recovery

Cipher1 is highly insensitive to distortion as it is evident from Fig. 20 (a). Distortion in Cipher2 recovered image is slightly noticeable and in Cipher3 are a little bit more prominent.

V. CONCLUSIONS The provided benchmark has discussed a wealth of cryptographic evaluation and performance parameters. These parameters are implemented to evaluate a block Copyright Š 2016 MECS

cipher, a compression tolerant and a chaos-based encryption schemes. The result of evaluation has demonstrated that block based cipher (AES) is good for cryptographic security as far as the communication channel is free of any distortion and image is not needed to be compressed. Compression tolerant encryption scheme has little security and can be used only in noisy channels or to achieve greater compression at the cost of security. Chaos-based scheme is proven to be a cipher of choice as it has high cryptographic security and demonstrated performance along with some tolerance to tempering, compression or noise. The proposed cryptographic evaluation benchmark can be applied to any image encryption scheme to quantize its security and performance. REFERENCES [1] Stallings, W., Cryptography and network security, principles and practices, 2003. Practice Hall. [2] Li, C. and K.-T. Lo, Optimal quantitative cryptanalysis of permutation-only multimedia ciphers against plaintext attacks. Signal processing, 2011. 91(4): p. 949-954. [3] Omar M.Barukab, Asif Irshad Khan, Mahaboob Sharief Shaik , MV Ramana Murthy, Shahid Ali Khan,"Secure Communication using Symmetric and Asymmetric Cryptographic Techniques", IJIEEB, vol.4, no.2, pp.36-42, 2012. [4] Prabir Kr. Naskar,Atal Chaudhuri,"A Secure Symmetric Image Encryption Based on Bit-wise Operation", IJIGSP, vol.6, no.2, pp.30-38, 2014.DOI: 10.5815/ijigsp.2014.02.04 [5] Daemen, J. and V. Rijmen, AES Proposal: Rijndael, AES algorithm submission, September 3, 1999. URL http://www. nist. gov/CryptoToolKit, 1999. [6] Nisar Ahmed, Y.S., Hafiz Adnan Habib, Design and Analysis of a Compression Friendly Image Encryption Scheme. Computers & Electrical Engineering - Journal Elsevier, 2015. [7] Ye, R., An Image Encryption Scheme with Efficient Permutation and Diffusion Processes, in Advances in Computer Science and Education Applications2011, Springer. p. 32-39. [8] Winkler, S. and P. Mohandas, The evolution of video quality measurement: from PSNR to hybrid metrics. Broadcasting, IEEE Transactions on, 2008. 54(3): p. 660668. [9] Shannon, C.E., Communication theory of secrecy systems*. Bell system technical journal, 1949. 28(4): p. 656-715. [10] Gray, R.M., Entropy and information theory2011: Springer Science & Business Media. [11] Ragab, A.H.M., O.S.F. Alla, and A.Y. Noaman, Encryption Quality Analysis of the RCBC Block Cipher Compared with RC6 and RC5 Algorithms. IACR Cryptology ePrint Archive, 2014. 2014: p. 169. [12] Kanso, A. and M. Ghebleh, An efficient and robust image encryption scheme for medical applications. Communications in Nonlinear Science and Numerical Simulation, 2015. [13] Chen, J.-x., et al., A fast chaos-based image encryption scheme with a dynamic state variables selection mechanism. Communications in Nonlinear Science and Numerical Simulation, 2015. 20(3): p. 846-860. [14] Wang, X.-Y., S.-X. Gu, and Y.-Q. Zhang, Novel image encryption algorithm based on cycle shift and chaotic system. Optics and Lasers in Engineering, 2015. 68: p.

I.J. Computer Network and Information Security, 2016, 12, 18-29


A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes

126-134. [15] Yu, M.-y., Image Encryption Based on Improved Chaotic Sequences. Journal of Multimedia, 2013. 8(6): p. 802-808. [16] Elashry, I.F., et al., Homomorphic image encryption. Journal of Electronic Imaging, 2009. 18(3): p. 033002033002-14. [17] Kwok, H. and W.K. Tang, A fast image encryption system based on chaotic maps with finite precision representation. Chaos, Solitons & Fractals, 2007. 32(4): p. 1518-1529. [18] Kamali, S.H., et al. A new modified version of advanced encryption standard based algorithm for image encryption. in Electronics and Information Engineering (ICEIE), 2010 International Conference On. 2010. IEEE. [19] El Fishawy, N.F. and O.M.A. Zaid, Quality of Encryption Measurement of Bitmap Images with RC6, MRC6, and Rijndael Block Cipher Algorithms. IJ Network Security, 2007. 5(3): p. 241-251. [20] Ahmed, H.E.-d.H., H.M. Kalash, and O.S.F. Allah. Encryption efficiency analysis and security evaluation of RC6 block cipher for digital images. in Electrical Engineering, 2007. ICEE'07. International Conference on. 2007. IEEE. [21] Mohamed, A.B., G. Zaibi, and A. Kachouri. Implementation of rc5 and rc6 block ciphers on digital images. in Systems, Signals and Devices (SSD), 2011 8th International Multi-Conference on. 2011. IEEE. [22] Cheddad, A., et al. Securing information content using new encryption method and steganography. in Digital Information Management, 2008. ICDIM 2008. Third International Conference on. 2008. IEEE. [23] Liehuang, Z., et al., A novel image scrambling algorithm for digital watermarking based on chaotic sequences. International Journal of Computer Science and Network Security, 2006. 6(8B): p. 125-130. [24] Massoudi, A., et al., Overview on selective encryption of image and video: challenges and perspectives. EURASIP Journal on Information Security, 2008. 2008: p. 5. [25] Wu, Y., J.P. Noonan, and S. Agaian, NPCR and UACI randomness tests for image encryption. Cyber journals: multidisciplinary journals in science and technology, Journal of Selected Areas in Telecommunications (JSAT), 2011: p. 31-38. [26] Li, J. and H. Liu, Colour image encryption based on advanced encryption standard algorithm with twodimensional chaotic map. Information Security, IET, 2013. 7(4): p. 265-270. [27] Gaddam, S.C., Drawing Color Histograms and Color Clouds, in Color Histograms01-08-2010, Boston

29

University: Boston, MA 02215, United States. [28] Sivakumar, T. and R. Venkatesan, A Novel Image Encryption Approach using Matrix Reordering. WSEAS Transactions on Computers, 2013. 12(11). [29] Lian, S., Multimedia content encryption: techniques and applications2008: CRC press. [30] Mohamed M. Fouad, Richard M. Dansereau,"Lossless Image Compression Using A Simplified MED Algorithm with Integer Wavelet Transform", IJIGSP, vol.6, no.1, pp.18-23, 2014.DOI: 10.5815/ijigsp.2014.01.03 [31] Shah, J. and V. Saxena, Performance Study on Image Encryption Schemes. arXiv preprint arXiv:1112.0836, 2011.

Authors’ Profiles Nisar Ahmed is a PhD scholar at Department of Computer Science and Engineering, University of Engineering and Technology Lahore. He has done MS Computer Engineering from the same institute. His areas of interest includes Multimedia Security, Computer Vision and Machine Leanring.

Shahzad obtained his Ph.D. degree in Informatics from University of Edinburgh, UK in 2012. He is working as associate professor at Department of Computer Science & Engineering, University of Engineering & Technology, Lahore.

Gulshan Saleem has done her MS Software Engineering from College of Electrical and Mechanical Engineering, National University of Science and Technology, Rawalpindi and her BS Software Engineeing from Fatima Jinnah Women University, Rawalpindi. Her areas of interest includes Machine Learning, Information Retrieval and Digital Image Processing.

How to cite this paper: Nisar Ahmed, Hafiz Muhammad Shahzad Asif, Gulshan Saleem,"A Benchmark for Performance Evaluation and Security Assessment of Image Encryption Schemes", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.18-29, 2016.DOI: 10.5815/ijcnis.2016.12.03

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 18-29


I. J. Computer Network and Information Security, 2016, 12, 30-35 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.04

Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection Jaspreet Kaur M.Tech student CEC Landran /Information technology, Mohali, Punjab, India E-mail: jaspreetsohi0@gmail.com Dr. Parminder Singh Associate Professor CEC Landran /Information technology, Mohali, Punjab, India E-mail: singh.parminder06@gmail.com Abstract—The energy has undergone a major concern in wireless sensor networks. One direction of the LEACH protocol would be towards a future in which four levels will continue to dominate. The other direction would be towards to improve these levels further and increasingly smaller amount of energy consumption. The support provided by the LEACH protocol has been crucial in which upgrade these levels and build smart distributed network. The previous scenario is not motivated by the scientific altruism. They are large number of clients in the proposed scenario and therefore have a good reason to encourage a scalable alternative for communication. LEACH protocol to take a leaf out of the proposed scenario and it has a good energy saver, less energy consumption. The proposed scenario needs to be better understanding the catalytic role played by the previous three levels LEACH protocol. It should also see whether there is scope for deploying more nodes through collaboratively proposed protocol with three level protocols.

hundred sensor nodes were imported in the first time and 97% were followed the base station node and the remaining were peer to per fashion. Every node on the network was look upon the sensor request or reply packet. Now as the request is complete the services go up with multiple devices, it is the time for other devices to ride on the path for message transmission. The base station has look upon the routing table and on the basis of routing table it decides to restart the process for data. In the earlier number of nodes failed to send the data again. The node deployment is usually done randomly by scattering nodes in the sensor field. In some applications, actuators (anodes) that control various devices can also be positioned within the sensor network. A collector node (cnode), which is often more capable than the other nodes in the field, is also located either inside or close to the sensor field.

Index Term—Throughput, LEACH Protocol, Multilevel, Interference.

I.

INTRODUCTION

Understanding and solving the challenges of sensor networks will require a unique framework. Such a framework needs to have four parameters: energy, Throughput, Delay and Interference. By their nature, systems comprise parameters that are distinct and yet inter-related. Sensor Nodes are crucial to fulfill the objectives of IEEE 802.15 standard. Sensor Network Protocol need to work on energy related issues. Base Station node provides a comprehensive report of the node communication and its issues that create problem in future. Base Station can create time-bound application so every node give confirmation on limited time. Sensor nodes have discovered a hidden node on the network and remove the entire path of the network. This information has been conveyed to each node on the network. Over Copyright Š 2016 MECS

Fig.1.Wireless Sensor Networks [9]

Cnodes, usually called sinks or base stations, are

I.J. Computer Network and Information Security, 2016, 12, 30-35


Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection

responsible for collecting the sensed data from snodes and then serving the collected data to users. They are also responsible for starting task disseminations in many applications. The sensed data by snodes is conveyed through the sensor network by multiple hops in an ad hoc manner, and gathered in cnodes that can be perceived as the interface between sensor networks and users. Multiple sensor networks can be integrated into a larger network through the Internet or direct links between either cnodes or gateways. A. Need of Multi-Level Protocol The sensor nodes that have completed their communication be released from the occupied channel immediately. Noise level of the network has increased in every time and reduction policy applied by the given method. This method is improvement in the case of best channel selection. There are multiple interfaces to route the data in every corresponding direction that the particular node needs. Every interface attaches to each sensor node on the network and thus the performance graph increases. Noted that we maintained the performance factor in every level but still we are too far from performance of the network. Till now we are struggling.

II.

RELATED WORK

Ravi Kishore Kodaliet. al. [1], Pointed out multi-level LEACH protocol and every round has checked the energy level of each node. TDMA technique has deployed on the multi-level LEACH protocol and increased the performance of the network. On an average result have been analyzing values Joules in energy consumption, they added. The other protocol analyzed by the authors [2], where the results has come down to previous approach results in wireless sensor networks. Divide and rule based technique applied on the density based network; while the throughput of the protocol value in previous work value comprising better in current work. Heinzelman, et al. [3] Low energy adaptive clustering hierarchy (LEACH) is a cluster based protocol in which the role of a member node in the cluster is to sense the surrounding environment and transmit the sense data to the BS. The data collection area is randomly divided into several clusters. Based on time division multiple accesses (TDMA), the sensor nodes transmit data to the cluster heads, which aggregate and transmit data to Base station. Kranthi K. Mamidisetty. al. [6], discussed dissemination of data or flooding. In this concept each and every node in the network sends the message to every node if the node does not want to that data from the source node. To avoid this spreading data all over the nodes in network, the author uses the multistage queuing network strategy. Amit Patwardhan. al. [10], discussed path planning for wireless sensor network The author used 5*5 grid nodes means 25 nodes communicating with each other and performed four functions sleep, idle, listening and Copyright Š 2016 MECS

31

transmitting, with amount of energy consumption being in ascending order. The nodes used in the scenario maintain a constant path for all the traversals of the base node is equally bad since even this will result in the formation of energy holes in the network. In [13] author addressed the issue related to throughput maximization and delay minimization and suggested the linear programming based solution. Author in [14] conducted a comprehensive study on the evaluation of LEACH based clustering protocols, where evaluation multilevel routing protocols on the basis of scalability and traffic constrain

III.

WORKING WITH MULTILEVEL LEACH

Multi-Level LEACH protocol has been looked upon something that is synonymous with regulating sensor size and reproductive energy. The planning of the network is such that the vision of reaching out to an additional new sensor which was enters by the approval of base station node. The goal of this paper is to improvement of throughput and lesser delay during communicative node. It enforced the nodes to make a decision according to routing table and this decision relating to their recommunication, new communication or any link failures. Previous energy related issues are not optimal solution or contraception, especially nodes, must have say in decision to re-communication. A.

Comparing of LEACH and Its Variants

In this paper Table 1.1, shows the comparison of Three Levels with LEACH protocol and thus we analyzed that we need to modify these levels for improvement in the area of sensor network. Table 1. Comparing Multi-Level LEACH Protocols Parameter

LEACH

Level-1

Level-2

Level-3

Algorithm Implemented

Every Node

Centralized only

Centralized only

Scheduling

TDMA

Base Station No TDMA

TDMA

TDMA

Energy Conserved

Less

Less

Less

More in Level 1 and Level-2

Coverage Area

Less

More than LEACH

More than LEVEL-1

More than Level-2

Parameter

LEACH

Level-1

Level-2

Level-3

IV.

PROPOSED ALGORITHM

The gap between proposed and previous approach is huge and there is trust deficit even more among the proposed approach. Proposed model pick up higher energy nodes from the network pool. If the centralized main node is ensured of a good linkage with communicating sensor nodes could benefit to route the data to other hundreds of nodes. The source node sent a message to provide additional amenities at node 15 near node 5 where a noted node dedicated to node 16 is

I.J. Computer Network and Information Security, 2016, 12, 30-35


32

Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection

located. This model is closely associated with MultiLevel LEACH and has a separate channel dedicated to each node. As the network is located in an area 1000 meter X 1000 meter, there is a facilitating to retransmission of packet, alarm collision in the channel to accommodating more sensor node. The main node has developed a channel of source node and receiver node. There has been spurt in the number of sensor nodes visiting in the cluster network which provides a separate channel dedicated to receiver node. The proposed work has completed the preliminary work including making random channel and to build channel to carry data or messages for the sensor networks. The algorithm after some modification would be applied to the network model. The main sensor node would be providing routing table and other necessary facilities to the entire sensor nodes on the network model. Referring ail [7] said various other. With the central sensor node rejecting the other node who did not give the MAC address to make amendments in the proposed algorithm. In the reply of MAC address given by the requesting node and is to be accepted by the central main node. If the node struggle for sending the data then central main node has assured to provide another shortest route for data transfer. Representatives of the sensor nodes in the network model have extent to support the central main sensor node. Assume that if there are n antennas of the communicating nodes in the network and they want for communication, thus, we allocate the channels to each node on the network. We are calculating the power of the sensor nodes individually from equation 1.

path channel selection but we adopt wideband channel to resolve the issues faces in narrowband selection technique. Suppose there is channel pool C and ÂĽC is the random channel selecting from the channel pool. The channel is selecting in such a way that no two channels with have a same frequency band. Then, the corresponding equation configure on the network simulation model.

In the channel selection approach the delay has been calculated by subtracting d2, d1and observed time is shown in t.

V.

SIMULATION AND RESULTS

During experiment conducted on a test bed of 100 sensor nodes, packet loss for each link was measured every minute for two hours under various topology setting. Although the majority of links were good, 10% of the nodes exhibited a packet rate loss greater than 2%. The test bed consisted of running the modified fixed probabilistic version of LEACH over a 100m x 1000m area. Each node has a transmission range of 250m, and the distance between two successive nodes was 180m and 200m. The constant bit rate (CBR) data traffic of 4packets/sec connecting node 0 to node 4 was used.

For any path loss in the channel the equation 2 calculates the energy loss during link disconnection.

Fig.3. Multi-Level Experimental Model Fig.2. Multi-Path Wideband Channels

The multi-path communication has not possible in sensor networks due to narrowband channel selection technique. Thus, the energy consumption tends to increase such a way that affect the performance of the Multi-Level Network. Below figure 1.2, shows the multiCopyright Š 2016 MECS

In the Experimental model presented in Figure 2, nodes 3, 4, 9, 38, and 43 sense the presence of a nearby nodes. Nodes 3–4 each send a message to node 38 with their observed sensor data. Node 3 forwards the received messages, along with its own set of sensor readings, to the next node along the path to the base station.

I.J. Computer Network and Information Security, 2016, 12, 30-35


Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection

Thus, node 3 sends a total of 5 messages, which are all subsequently relayed from node to node, until they reach the base station. In total, 29 messages are transmitted throughout the network. A reduction in communication and energy costs are possible if collected sensor data is aggregated prior to relaying.

VI.

ANALYSIS AND RESULTS

The following results have been calculating that was studied in figure 3 and now we are explain it results A.

Throughput

Throughput is typically defined as the rate at which messages are serviced by a communication system. It is usually measured either in messages per second or bits per second. In wireless environments it represents the fraction of the channel capacity used for data transmission. Throughput increases as the load on the communication system increases initially. After the load reaches a certain threshold, the throughput ceases to increase, and in some cases, it may start to decrease. An important objective of a MAC protocol is to maximize the channel throughput while minimizing message delay.

33

with on-going transmissions in the adjacent frequencies. The range within the intended receiver may be subject to interference from other transmission sources, thereby causing the rate of transmission errors to be higher than desired. For the Physical layer configuration, the interference of individual sensor nodes varied from 0.3 to 0.9, with an overall average interference of 0.3 Watt for the entire sensor network. The nodes with the best reliability were those placed closest to the sink node. Nodes located farthest from the sink node and along the edges of the sensor network exhibited the most packet loss. The hop count for messages to reach the destination varied from a minimum of 1 hop to maximum of 8 hops.

Fig.5. Channel Interference of Multilevel LEACH Protocol

VII.

CONCLUSION

It should be analyzed from the results that we have analyzed from the section V1. Table 2. Comparison of Interference

Fig.4. Throughput of the Multi-Level LEACH Protocol

From above figure (figure 4) clearly shows that the normalized throughput of Base Station increases to a maximum of 20 to 30 packets/seconds respectively when the sent is increased from 10 to 70% increases, and dropped to approximately 10 to 40 seconds respectively when the forwarding packet is increased. This throughput is examined in more than hundreds of nodes. B.

Channel interference

802.11 does not have support aided by transmit power for transmit power control (TPC), sub channelization or dynamic channel selection and support for adaptive (DCS). A number of PHY considerations were taken into account for the target environment. At higher frequencies, line of sight (LOS) is a must. In adjacent channel interference, the signals in nearby frequencies have components outside their allocated frequency ranges, and these components may interfere Copyright Š 2016 MECS

Parameter Previous Work Proposed Work

Minimum

Average

Maximum

0.6

0.7

0.9

0.5

0.6

0.7

As per the information received from the network simulator; the interference from the previous work is more than that of proposed work. The value is slightly down in the case of proposed when more than 50 nodes complete their communication and base station maintained the MAC Table. Table 3. Comparison of Throughput Parameter Packet Sent (Previous) Packet Receive (Previous) Packet Sent (Proposed) Packet Receive (Proposed)

Minimum

Average

Maximum

7

7

8

6

7

8

15

20

25

15

20

25

I.J. Computer Network and Information Security, 2016, 12, 30-35


34

Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection

We have been studying the throughput of the network and we concluded that the proposed scenario gives higher value than previous work. This throughput figure out the comparison of packet received divided with packet sent. From the comparison table if we add some modification in the multi-level LEACH protocol as discussed in section4; the performance of the network remains high.

[12]

ACKNOWLEDGEMENT

[15]

We wish to thank our parents for their good upbringing which help us to do good works in every field of life. We would like to thanks all our professors in our respective departments to help and guide us in the ways we needed to get success in this field of research. Without their stimulating guidance, encouragement and suggestions, this research work would not have been possible. We also thank all faculty members of Information technology Department for constructive suggestions to improve the quality of this research work and providing such a great academic environment. Finally, thanks to all our friends for their support and encouragement.

[13] [14]

[16] [17]

[18]

[19]

[20]

REFERENCES [1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10] [11]

Jianli ZHAO, Lirong YANG, “LEACH-A: An Adaptive Method for Improving LEACH Protocol", Sensors & Transducers, Vol. 162, Issue 1, January 2014, pp. 136140. Jiman Hong, Joongjin Kook, Sangjun Lee, Dongseop Kwon, Sangho Yi, “T-LEACH: The method of threshold-based cluster head replacement for wireless sensor networks", Inf Syst Front, Springer Science, 2009, pp.513-521. N.Israr, I.Awan, “Converge based inter cluster communication for load balancing in heterogeneous wireless sensor networks”, Telecommun Syst, 2008, pp.121-132. S.Koteswararao, Dr. M. Sailaja, P. Ramesh, E. Nageswararao, V. Rajesh, “Sensor Networks Simulation in NS2.26”, IJECT, 2011, pp.251-254. Ravi Kishore Kodali, Naveen Kumar Aravapalli, “Multi-level LEACH Protocol model using NS-3”, IEEE, 2014, pp.375-380. Gilbert Chen, Joel Branch, Michael J. Pflug, Lijuan Zhu, Boleslaw K. Szymanski, “SENSE: A Wireless Sensor Network Simulator”, Advances in Pervasive Computing and Networking, Springer, 2004, pp.249-267. Amit Patwardhan, “Energy based path planning for wireless sensor networks”, International Journal on Emerging Technologies, 2010, pp.16-18 P. Chevillat, J. Jelitto, A. Noll Barreto, H.L. Truong, “A Dynamic Link Adaptation Algorithm for IEEE 802.11a Wireless LANs”, IEEE, 2003,pp.1141-1145. A. Ahmad, K. Latif, N. Javaid, Z. A. Khan, U. Qasim, “Density Controlled Divide-And-Rule Scheme For Energy Efficient Routing In Wireless Sensor Networks”, 2013. The Network Simulator ns-2. http://www.isi.edu/nsnam/ns/ Kiran Maraiya Kamal Kant Nitin Gupta, “Application based Study on Wireless Sensor Network”, International Journal of Computer Applications, volume 21– No.8, 2011.

Copyright © 2016 MECS

[21]

Amir Akhavan Kharazian, Kamal Jamshidi, Mohammad Reza Khayyambashi, “Adaptive Clustering in Wireless Sensor Network: Considering Nodes with Lowest Energy”, IJASUC, 2012, pp.1-12 Amiya Nayak and Ivan Stojmenovic, “Wireless Sensor and Actuator Networks”, Wiley, 2010. Jaspreet Kaur, Parminder Singh “Measuring the data routing efficacy in LEACH Protocol”, IEEE INDIACom-2016. Parminder Singh “Comparative study between unicast and Multicast Routing Protocols in different data rates using vanet”, IEEE February 2014, pp. 278-284. Sheffy Jasuja, Parminder Singh “Accountability of WMNs using BEB Algorithm”, Volume 5, 2015 Manish Kumar jha, Atul kumar pamdey, Dipankar pal, “Energy-efficient multilayer Mac protocol for wireless sensor network” n.209 216. Bachar Rachid, Haffaf Hafid “Distributing Monitioringfor wireless Sensor Networks: a Multi-agent approach”, September 2014 IJCNIS 13-23. Kushlendra kumar pandey, Neetesh Purohit, Ajay Agarwal “Efficient Clustering Technique for cooperative wireless sensor network”, IJCNIS 2014 404. Sachin Gajjar, Nilav Choksi, Mohanchur Sarkar, Kankar Dasgupta “LEFT: A Latency and Energy EfficientFlexible TDMA Protocol for Wireless Sensor Networks”, IJCNIS januray 2015 vol 7 N2, pp 1-14. T.S. Rappaport. “Wireless communications: principles and practice”, Prentice Hall, 2nd edition, 2002.

Authors’ Profiles Jaspreet Kaur She was born in 27 August in 1991 in India. She did her B.Tech in Information Technology from Doaba Group of College, Kharar India in 2013 and now she currently perusing M.Tech from Chandigarh Engineering College Landran, India. She has published paper in National/International Journal Conferences. Her area of interest is networking.

Dr. Parminder Singh is a young dynamic personality with a proven record of a good academician and researcher having an outstanding academic record. He has been working as an Associate Professor in Information Technology Department and has more than Ten years of rich experience as an academician and researcher. He has published over 70 Journal and conference papers in the areas of Networking, Wireless Networks, sensor computing and Network security. He has also contributed various articles published in Elsevier Book and Springer Berlin Heidelberg. He has won best-paper awards including the IEEE "Best Paper Award" in the Year 2012 and 2014. He has served/serving in International conferences as a general pc co-chair, and steering committee member, and presented Expert Lectures in the areas Wireless Networks and sensor computing. He has received faculty excellence and research awards in the year 2011, 2013 and 2015 from his Institution for excellence in research, teaching and service.

I.J. Computer Network and Information Security, 2016, 12, 30-35


Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection

35

How to cite this paper: Jaspreet Kaur, Parminder Singh,"Improving Energy of Modified Multi-Level LEACH Protocol by Triggering Random Channel Selection", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.30-35, 2016.DOI: 10.5815/ijcnis.2016.12.04

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 30-35


I. J. Computer Network and Information Security, 2016, 12, 36-43 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.05

An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks Reza MotahariNasab Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran E-mail: rezamotahari68@eng.ui.ac.ir

Ali Bohlooli and Neda Moghim Assistant Professor, Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran E-mail: {bohlooli, n. moghim}@eng.ui.ac.ir

Abstract—Underwater Wireless Sensor Networks (UWSNs) consist of certain number of sensors and vehicles interacting with each other to collect data. In recent years, the use of Autonomous Underwater Vehicle (AUV) has improved the data delivery ratio and maximized the energy efficiency in UWSNs. Clustering is one of the effective techniques in energy management which increases the lifetime of these networks. One of the most important parameters in creating optimized clusters is the choice of appropriate cluster head (CH), which not only increases the lifetime of the network and the received data in the sink, but also reduces energy consumption. Clustering of networks was primary done via distributed methods in previous researches. It spends too much energy and also involves too many nodes in the clustering process and fades their main functionality, which is gathering data in sensor networks. It also causes more damping of the network. However, in the proposed protocol, instead of having them distributed by the network and the nodes, the stages of clustering and selecting the appropriate CH is the task of the AUV (Autonomous Underwater Vehicle). Since all the necessary measures to cluster in the network will be carried out by the AUV by this method, many control overheads in the process of clustering the network will be removed and energy consumption caused by nodes reduces significantly. With this method, the network scalability will also be manageable and under control. For simulating and implementing our method we mainly used the OPNET software. The results show that energy consumption of nodes in the proposed algorithm has been significantly improved compared to previous results. Index‌Terms—Energy Optimization, Data Gathering, Underwater Sensor Networks, Autonomous Underwater Vehicle (AUV), Clustering. I. INTRODUCTION There has been a recent trend in deployment of underwater sensor networks (UWSNs) in various Copyright © 2016 MECS

applications related to environmental and surveillance data acquisition in oceanic fields. Instead of radio and optical signals, most of the underwater communication systems use acoustic signals. However, the use of acoustic signals imposes many design challenges on communication protocol for UWSNs due to the high bit error rate, narrow bandwidth and long propagation delay. Naturally, sensor nodes are forced to communicate with each other over a short distance (a possible way). Various methods have been proposed for data gathering in these networks, and specific parameters of them have been improved which leads to the design and development of the autonomous underwater vehicle (AUV). AUV will facilitate network deployment, maximize network coverage, and enable high-speed networking. In general, data gathering method can be divided into two categories based on the path a mobile sink travels: path-constrained and path-controlled. By using the path-constrained scheme, the path of the mobile data sink is predetermined, whereas it is dynamically optimized in the path-controlled mode to conserve energy and improve data delivery rate. In this paper, we design a communication protocol for application of an AUV gathering data from an underwater sensor network. Recent advances in the onboard processing capabilities of AUVs, permits us to perform increasingly sophisticated data collection tasks currently[1]. In particular, these advances make the in situ adaptation of AUV data-collection survey routes possible. This capability has the potential to ensure that the data collected at sea by an AUV meets the standards necessary for various subsequent objectives. The remainder of this paper is organized as follows: Section II describes and examines the related works. Section III propose the network model. Section IV proposes our protocol for data gathering and energy consumption model used in underwater sensor networks. Section V validates our protocol through simulated deployments. Finally section VI presents our conclusions.

I.J. Computer Network and Information Security, 2016, 12, 36-43


An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

II. RELATED WORK Domingo and Prior [2], [3], [4] proposed a clustering approach where data is forwarded to a destination node in a single hop manner. Although this data forwarding technique is effective way to minimize the energy consumption in case of a large set of nodes, the issue of uniform consumption of energy still remains unsolved for cluster-heads. This arises the need for using a mobile node to collect data from neighbors in an UWSN. In addition, the issue of switching cluster heads is also very important that the leader should be considered. It is shown that an AUV acting as a mobile sink can effectively reduce the transmission range of sensors[5], which leads to energy saving of transmission. In this way, an AUV travels a specified path and stops at location called "tour-point" for data gathering. Same as the previous method [6], AUV is used to collect data UWSN networks. In such approaches, the probabilistic nature of the neighborhood, where successful communication probability is low, may result in retransmissions over the acoustic link. Beside AUV operational cost, these retransmissions add additional cost in term of excessive resource consumption. Authors have analyzed the usage of a special fixed node to gather data from static acoustic sensor network[7, 8]. In this approach, fixed nodes collect data from neighbor nodes and forwards it to AUV during data gathering tour. However, the authors have not discussed the issue of non-uniform energy consumption during AUV data gathering and the rapid depletion of energy at the fixed node. This could be one of the important drawbacks of this approach. Clustering is another important issue that should be considered in underwater networks. One of the most famous hierarchical routing protocols based on clustering, is the LEACH protocol. In this method, each cluster member send their data to cluster head. The cluster head aggregate this data and send it to the BS, thus the cost of communication will reduce dramatically. Many improvements have been made in LEACH protocol so far. LEACH-C method is an example of these improvements. In LEACH-C, the forming of clusters is done using a centralized algorithm by base station in starting of each period. Base Station uses the received information from nodes for finding the predetermined number of cluster heads and network configuration within the clusters. Some algorithms that based on not sending the correlated data are considered. The TINA algorithm is one of them. In this algorithm the sensor node compares the value of sampled data with previous data, if it found them having different values, it will send them; otherwise it goes to sleep mode. The proposed improvement to this algorithm could be that sensor node decides to send data after comparing the value of new sample with last reported data. Maximum Amount Shortest Path (MASP) scheme is proposed to increase the network throughput with better energy efficiency by optimizing the assignment of sensor Copyright Š 2016 MECS

37

nodes[9]. Genetic algorithm (GA) is utilized jointly with the integer linear programming to derive the mobile sink’s path. Authors have presented a data collection scheme that is optimized by considering multi-hop routing[10]. The authors [11] extended the concept of the Traveling Salesperson Problem (TSP) and proposed an algorithm for AUV path planning so as to maximize the amount of collected information and minimize the travel time.

III. NETWORK MODEL The proposed network architecture is made up from sensor nodes that are fixed on the ocean floor. Sensor nodes are organized into clusters and a cluster has a cluster head node. In order to choose a cluster head node, a special algorithm is applied, wherein various criteria are considered, including the remaining energy and the distance to the path of the AUV. Fig.1. shows this architecture.

Fig.1. Static Two-Dimensional Underwater Acoustic Networks [12]

In this method, the cluster head nodes collect data from the sensor nodes and delivers it to the mobile sink (AUV). Clustering with the help of AUV is done via this method, where after placing all the sensor nodes on the ocean bed and fixing them, the AUV enters the area of the sensor network and after each seconds it makes a signal with an acoustic range. This acoustic signal expands in the underwater space in the form of threedimensional shape and active nodes that are in the acoustic range of the signal. In other words, the nodes that this released signal covers by AUV will receive this signal. The content of the Hello packet which is sent by the AUV includes the id, which is the same number of the determined clusters by the AUV. Each node that receives this packet saves the number of clusters and thus becomes a member of this cluster. After every seconds, the AUV moves forward and the value of one id unit increases. Finally the identified clusters with different id will be formed. Fig.2. shows how a cluster with the above stated method is created.

I.J. Computer Network and Information Security, 2016, 12, 36-43


38

An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

Fig.2. Clustering Network using AUV

The time span, after which the AUV sends the Hello packet, is obtained from Equation 1: (1) Where, 2R is the maximum communication range of two distanced nodes and v is the speed of the AUV. Fig.3. shows clustering flowchart from the perspective of AUV.

Using this method for clustering the nodes has several advantages; for example it improves the energy consumption of the system. The other advantage of this method is the ability to adjust and control the size of the cluster and the number of nodes in the cluster. By changing the power of the sent signal from the AUV side, the range of the released signal is changed and therefore the number of nodes in the cluster can also change to lees and more. Improving the scalability is also one of the other important advantages of using this method in clustering the network. Also, due to the low volume of the package sent from AUV, which only includes one id, the use of this method therefore will have no overhead in transportation and in calculations. For this reason, the amount of memory used by the nodes will also decrease.

IV. THE PROPOSED PROTOCOL FOR DATA GATHERING After the formation of clusters in question, data gathering from the network will be conducted in three phases: 1) First Phase: The phase of selecting the cluster head nodes 2) Second phase: Data gathering phase, where the data is sent to the cluster head node by nodes. 3) Third phase: Transition phase, where the collected data by the cluster head node is sent to the AUV. A. The Method of Selecting the Cluster Head Node Energy parameters and the geographical locations are among the most important parameters that affect the determining of the cluster head node. In the proposed method of this paper, a node is chosen as a cluster head node. Beside having the initial conditions (needed energy), it also has the shortest distance to moving path of the AUV. Therefore, for each node, the Wi weight is considered for underwater sensor network. Equation 2 describes how to calculate the Wi weight. (2) In this equation, Ci is the remaining energy level of the node and Di shows the distance of node from the AUV path. B. The Steps of the Proposed Algorithm In order to select the cluster head node, a timer is assigned for each node. The stages of selecting the cluster head node by the help of the node weight (Wi) and the included timer is as follows:

Fig.3. Clustering flowchart from the Perspective of AUV

Copyright Š 2016 MECS

First step: In order to use the timer in the nodes, we firstly need to synchronize the nodes with each other in such a way that with releasing a strong signal, the AUV synchronizes the nodes in the network. When the nodes receive the setup signal from the AUV, they proceed to I.J. Computer Network and Information Security, 2016, 12, 36-43


An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

set their timer, the value of which is (ti) and is set corresponding to the node weight (Wi). It should be noted that the amount of ti should have an inverse relation with the node weight (Wi). Second step: The first node that its timer reached zero will be known as the cluster head node and will send a special data packet called CH-ADV to all nodes. This packet includes the node number, the cluster number that the AUV has sent to it in the development phase of the cluster and the amount of CH-Count, which indicates the number of times that this node is selected as a cluster head node. The nodes that receive this packet specify whether or not the number of the cluster in the message is equal to the number of the cluster they had saved themselves. If there is a match, the node in question accepts the data packet sent from the cluster head node and stops its timer then saves the number of that node as the number of the cluster head node. Otherwise, it will destroy the received packet and continue to decrease its timer until it receives a new packet from an appropriate cluster head node or its timer hits zero. Third step: After all nodes of a cluster receive the CH-ADV message of the cluster head node, they send their weight (Wi) and their other data in a Data message to the cluster head node. The cluster head node stores these values in its memory so they can be delivered to the AUV later. Fourth step: When a cluster head node receives the Data message from a sensor node, it waits for a certain time (about 70 ms), then proceeds to send the END message to the entire cluster. This message will indicate the end of the data gathering phase and the starting of the transition phase. In order to save energy as in the period one, each node goes to sleep mode upon receiving the END message. Fifth step: As the transition phase begins, the cluster head node will wait to be located in the communication range of the AUV. Upon arrival of the AUV to the considered area and after it sends the message of Data – Req, the cluster head node starts communicating with AUV and sends the stored data to it. Other nodes also start to calculate their own weight by getting the message (Wi) and then adjust their amount based on Wi . Sixth step: At this stage, besides receiving the sent data from the cluster head node, the AUV also compares the node weight with the threshold weight (Wth). If the cluster head node weight (Wch) is smaller than the threshold weight, the Setup message is sent to all nodes by AUV (back to the first stage) and then the process of the selecting of the head node will be resumed. If the cluster head node weight is not less than the threshold weight, the AUV does not send a message to the nodes and so they are allowed to continue to work, which is sending the collected data to the cluster head node. C. Model of Energy Consumption Used In Underwater Sensor Networks Sensor networks lifetime depends on the energy consumption of the nodes in that network. In order to estimate the energy consumed by each type of sensors in Copyright Š 2016 MECS

39

the underwater networks, the acoustic signal propagation model, proposed by Yorick [13], is used. Equation 3 [14] shows the ratio of the signal to noise in the underwater propagated signal in the receiver: (3) In this equation, SL is the noise of the source level, TL represents weakness, NL represents the source noise level and DI represents the direction of propagation. All the values in the formula are based on dB. Factors that are influential in obtaining the NL in underwater sensor networks include: disturbance (perturbation) (Nt), shipping (transport) (Ns), noise from rotating impeller (Nw) and thermal noises (Nth) based on the influencing parameters on the noise, the overall noise of NL is calculated by Equation 4: (4) The DI value for the network is considered to be zero, because it is assumed that the direction of propagation and the receiving of the underwater acoustic signals are from all the directions. The weakening (TL) related to the parameters of absorption and spread is obtained by Equation 5[15]: (5) Where in equation 5 the first part is related to the parameter of spread and the second part is related to the parameter of absorption. K factor indicates the type of signal expansion. Therefore, k = 1 represents cylinder release, k = 2 represents spherical release, and k = 1. 5 shows it is in practical and useful status. also represents expansion distance of the signal in meters. Thorpe Equation [14,15] also calculates the absorption rate: ( ) (6) And

(

) (7)

We assume the amount of noise in receiver is15dB. To calculate the amount of SL in Formula 3, the transferred signal intensity (It) should be calculated. â „

(8)

Therefore, the energy transfer based on every bit can be calculated as follows:

I.J. Computer Network and Information Security, 2016, 12, 36-43


40

An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

(9) Where h is the depth of network in the ocean in meters. With the help of the above equations, one can calculate the consumed energy for each transmission of the END and CH-ADV messages. Equation 10 is used to calculate the consumed energy of the sensor nodes:

(10) And the consumed energy by the cluster head node is also calculated by Equation 11:

(11)

the functions included in these process models there can be the production of desired packets, functions of receiving the incoming packets, the transfer function of the normal node to the cluster head node and vice versa, the function of introducing the cluster head node, etc. It should be noted that the conditions of transition from one state to another, demand the establishing of the necessary conditions that these conditions should be defined in the simulator. Table 1. The Initial Conditions And acoustic Link Parameters used In the Simulation Parameters Channel Bandwidth Central Frequency Data Packet Length Control Packet Length Data Rate Power AUV Speed Number Of Clusters Number Of Nodes In Each Cluster Total Number Of Nodes

Values 4KHz 25 KHz 1024bits 32bits 2500bps 50w 2m/s 8 6 48

In equations 9 and 11, nch refers to sensor nodes, ch refers to head node, T refers to transmission, R means receiving and f represents transmission frequency. Equations 3 to 11 are used for estimating the network lifetime based on the consumed energy of different nodes as well as the cluster’s various operations such as the formation of cluster, data gathering, and data transmission to the AUV [16].

V. EVALUATING THE PERFORMANCE OF THE PROPOSED PROTOCOL The results presented in this section are the outcomes of simulation in OPNET simulator, which are implemented based on the parameters in Table 1. The results for the different seeds are repeated several times, and the results in each test are recorded and the final result is obtained via averaging the results of the carried out tests. In our simulation model, we consider packet error rate of 10-3 for binary shift keying (BPSK) modulation. In addition, we assume the nominal speed of sound as 1500 m/s in acoustic channel to simulate propagation delay. The detail of the acoustic link parameters used in our simulation model is given in Table 1. It should be noted that each of the nodes in the network and the mobile node (AUV) is equipped with acoustic modes WHOI (Micro-Modem) for making acoustic voice communication [17], [18]. For this reason, the modem features should be considered in the simulation. Fig.4. shows the process model related to the AUV and Fig.5. shows the process model of the sensor node in OPNET simulator. As shown in Fig.4. and Fig.5., these process models are composed of multiple functions, each of which are responsible for specific tasks and as soon as the necessary conditions are established, they are performed. Among Copyright Š 2016 MECS

Fig.4. Process Model of the AUV

Fig.5. Process Model of the Sensor Node

The battery power of the underwater acoustic wireless sensor network node is limited. So, when the energy runs out, the node will fail, resulting in the network partitioning or even crashing the entire network. Therefore, how to reduce energy consumption and prolong network life cycle are key issues the study of underwater acoustic wireless sensor network. Fig.6. shows the maximum energy consumption by nodes and Fig.7. shows the minimum energy consumption based on

I.J. Computer Network and Information Security, 2016, 12, 36-43


An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

simulation time. This time is roughly equal to 3600 Seconds.

41

similarity function can save energy by reducing the number of transmissions from cluster heads to the AUV. Fig.8. compares the ratio of the signal to the noise in the AUV and the node. As it is clear, by the increase in simulation time, the ratio of signal to the noise is decreased up to approximately 40dB.

Fig.6. Maximum Energy Consumption by the Node

Fig.8. Signal to the Noise Ratio (SNR)

In Fig.9., the results are compared with the BGAF algorithm [18], which is a clustering algorithm in the underwater networks. Comparison of the charts shows that the proposed algorithm has a better function than the BGAF algorithm. Experiments as well as show that the proposed algorithm is very reasonable for the mechanisms to balancing network load and extending the network life cycle.

Fig.7. Minimum Energy Consumption by the Node

The maximum power consumption (Fig.6.) is related to the cluster head node (CH) and the minimum energy consumption (Fig.7.) is related to the typical sensor node. As shown in the Figures, since it has more message exchanges compared to other nodes, the CH consumes more energy than other nodes. Nevertheless, according to Fig.6. and Fig.7., as long as each node has no message exchanges with other nodes or the AUV, its energy consumption is almost zero and after establishing the necessary and desired data exchange (between nodes and CH or between CH and AUV) the energy consumptions would have an increasing process. In this proposed scheme, the energy is saved at each phase of the clustering scheme. For example, only AUV is allowed to work during the initial phase. Also, data aggregation with Copyright Š 2016 MECS

Fig.9. Comparing the Energy Consumption of the Proposed Protocol with BGAF Algorithm

VI. CONCLUSION Various techniques are known for collecting data from acoustic underwater networks, each of them has their own advantages and disadvantages. In this paper we proposed a method for decreasing energy consumption in

I.J. Computer Network and Information Security, 2016, 12, 36-43


42

An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

sensor networks; given that the energy consumption is a subject undergoing intense study with broad and current interests. We continue our work through selecting the most appropriate method, which is the clustering of nodes and determining a cluster head node for it. Beside decreasing the consumed energy of the nodes, this method will also create flexibility in the network and will increases its lifetime. The OPNET simulator is used for implementing this algorithm in our simulation environment. The results of this simulation indicate that the proposed algorithm was quite successful in its task. We foresee this method could be used in collecting data from the sensor nodes by the AUV robots in future. REFERENCES Z. Latifi, K. Jamshidi and A. Bohlooli, ―Increasing the Efficiency of IDS Systems by Hardware Implementation of Packet Capturing‖, in I. J. Computer Network and Information Security (IJCNIS), Vol. 10, pp. 30-36, 2013. [2] M. C. Domingo, R. Prior, ―Energy analysis of routing protocol for underwater wireless sensor networks,‖ Elesvier. Computer Communications, 31, 2008. [3] M. C. Domingo, R. Prior, ―A distributed clustering scheme for underwater wireless sensor network, ‖ In Proc. IEEE PPIMRC, Athens, September, 2007. [4] H. Harb, A. Makhoul, and R. Couturier, ‖ An Enhanced K-means and ANOVA-based Clustering Approach for Similarity Aggregation in Underwater Wireless Sensor Networks‖, in IEEE Sensors Journal, MARCH 2015. [5] G. A. Hollinger, S. Choudhary, P. Qarabaqi, C. Murphy, U. Mithra, G. S. Sukhatme, M. Stojanovic, H. Singh, and F. Hover, ―Underwater Data Collection Using Robotic Sensor Networks, ‖In IEEE. Journal on Selected Area in Communications, Vol. 30, No. 5, June, 2011. [6] F. Favaro, P. Casari, F. Guerra and M. Zrozi, ―Data upload from a static underwater network to an AUV: Polling or random access?, ‖ In Proc. MTS / IEEE Oceans Yeosu, Korea, May, 2012. [7] F. Favaro, L. Brolo, G. Toso, P. Casari and M. Zrozi, ― A Study on Remote Data Retrieval Strategies in Underwater Acoustic Network, ‖ In Proc. MTS IEEE Oceans San Diego, US, September 2013. [8] N. Ilyasa, T. A. Alghamdib, M. N. Farooqa, B. Mehbooba, A. H. Sadiqa, U. Qasimc, Z. A. Khand, N. Javaid, ‖ AEDG: AUV-aided Efficient Data Gathering Routing Protocol for Underwater Wireless Sensor Networks‖, in Elesvier Procedia Computer Science, 2015. [9] S. Gao, H. Zhang, and S. Das, ―Efficient data collection in wireless sensor networks with path-constrained mobile sinks,‖ IEEE Trans. Mobile Comput. , vol. 10, no. 4, pp. 592–608, Apr. 2011. [10] J. L. and J. P. Hubaux, ―Joint mobility and routing for lifetime elongation in wireless sensor networks,‖ in Proc. IEEE INFOCOM, vol. 3, Mar. 2005. [11] G. A. Hollinger, U. Mitra, G. S. Sukhatme, ‖ Autonomous data collection from underwater sensor networks using acoustic communication‖, In Intelligent Robots and Systems (IROS), IEEE/RSJ International Conference on, pp. 3564-3570, 2012. [12] D. Jabba And M. Labrador, ―A Data Link Layer In Support of Swarming of Autonomous Underwater Vehicles,‖ In Proceedings Of The MTS/IEEE Oceans Conference, Bremen, Germany, May 2009. [1]

Copyright © 2016 MECS

[13] M. Stojanovic. ―Acoustic (Underwater) Communications‖. In John G. Proakis, editor, Encyclopedia of Telecommunications. John Wiley and Sons, 2003. [14] C. Pelekanakis, M. Stojanovic, and L. Freitag. ―High rate acoustic link for underwater video transmission‖. InProc. of MTS/IEEE OCEANS 2003, vol.2, pp. 1091–1097, San Diego, CA, USA, September. [15] T. Melodia, H. Kulhandjian, L. C. Kuo, E. Demirors, ‖ Advances In Underwater Acoustic Networking‖, in Wiley Online Library, May 2013. [16] Z. Heidarian, N. Movahedinia, N. Moghim, P. Mahdinia, ―Intrusion Detection Based on Normal Traffic Specifications‖, in I. J. Computer Network and Information Security (IJCNIS), Vol. 9, pp. 32-38, 2015. [17] D. Richard Blidberg, ―The Development of Autonomous Underwater Vehicles (AUV); A Brief Summary‖, Autonomous Undersea Systems Institute, Lee New Hampshire, USA, 2010. [18] S. Sendra, J. Lloret, J. M. Jimenez, and L. Parra, ―Underwater Acoustic Modems‖, in IEEE Sensors Journal, Vol. 16, No. 11, June 1, 2016. [19] G. Liu, W. Wen, ‖A improved GAF clustering algorithm for three-dimensional underwater acoustic networks‖, in IEEE International Symposium on Computer, Communication, Control and Automation, 2010.

Authors’ Profiles Reza MotahariNasab, received the BS degrees in Computer engineering from the Payam-e-Noor’s Isfahan University, Iran in 2012. He is now completing the MSC degree at the University of Isfahan, Iran. His research interests include wireless communication and wireless sensor networks.

Ali Bohlooli, received the B.S. and M.S. degrees in Computer engineering (with honors) from the department of Electrical & Computer Engineering, Isfahan University of Technology, Iran in 2001 and 2003, respectively. He received his Ph.D. degree from the University of Isfahan, Iran in 2011. Now he is an assistant professor at faculty of Computer Engineering, University of Isfahan, Iran. His research interests include wireless networks and network modeling.

NedaMoghim, received the B.S. and M.S. degrees both from Isfahan University of Technology, Iran, Isfahan in 1999 and 2002 respectively and the Ph.D. from Amirkabir University of Technology, Iran, Tehran in 2009. She is the author of several technical papers in telecommunications journals and conferences. Currently she is an assistant professor with the Department of Information Technology Engineering, University of Isfahan, Iran. Her research interests are in the area of admission control and bandwidth

I.J. Computer Network and Information Security, 2016, 12, 36-43


An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks

43

management/ traffic engineering for QoS-enabled IP networks, next generation networks, and wireless mobile/fixed networks.

How to cite this paper: Reza MotahariNasab, Ali Bohlooli, Neda Moghim,"An Energy-Aware Data-Gathering Protocol Based on Clustering using AUV in Underwater Sensor Networks", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.36-43, 2016.DOI: 10.5815/ijcnis.2016.12.05

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 36-43


I. J. Computer Network and Information Security, 2016, 12, 44-50 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.06

A Novel Technique to Prevent PUE Attack in Cognitive Radio Network Poonam 1a, Ekta gupta 2a, C.K. Nagpal a a

YMCA University of Science and Technology, India 1 2 E-mail: poonamgarg1984@gmail.com , ektagupta43@gmail.com

Abstract—Need of wireless communication is increasing to work from distance. That is why new applications are made everyday which increases demand of spectrum but due to limitation of spectrum and inefficient utilization of spectrum. A new paradigm is constituted which is called Cognitive Radio Network (CRN). It get more attention in recent times due to most promising solution for the efficient utilization of spectrum. Spectrum sensing in CRN makes it prone to many attacks on each layer. One of these attacks is PUE attack where a malicious user pretends to be a primary user and not let others to use primary user's channel in its absence. It may cause Denial of Service attack in the network. There are many techniques available in the literature for detection and prevention of PUE attack but still there are some limitations in these approaches. Current research provides detection results based on the energy level of all users in the network. In this paper we provide a novel approach to prevent PUE attacker based on signal activity patterns. Simulation is done in MATLAB-2013 and results show that proposed method gives excellent performance. Index Terms—Cognitive radio network, security, primary user emulation attack, trust-based mitigation, reputation-based mitigation, primary user emulation attack prevention.

So to counterbalance all these problems a solution was proposed which is called Cognitive Radio Network (CRN) [31] [32]. In this network when the licensed users are not using their channels other unlicensed users come and use their channels till the licensed users do not come again on their channels. These licensed users are called primary users and unlicensed users are called secondary users. To use the primary user's channel secondary users need to sense their channels continuously to find holes in the network [18] [23]. This process of using the primary user's channels is divided into many layers in the network. On each layer there is presence of secondary users which are continuously sensing the channels. So to increase their time of presence in the network or to stop primary users to use their channels some secondary users may behave maliciously. One of these malicious activities is to capture the identity of any primary user to use its channel and to stop other secondary users to use it. It leads to a very harmful attack in the network called Primary User Emulation Attack as shown in Fig 1. This is done on the physical layer in the network. In this paper we will give a brief about many detection and prevention techniques for this attack those are proposed till now. We propose a method to prevent this attack that is based on Signal Activity Pattern and give the simulation results using MATLAB 2013(b) which shows our proposed method gives better performance than previous.

I. INTRODUCTION Wireless communication is done by radio spectrum. For this the spectrum is divided into number of channels and these channels are distributed among many groups of users which may be either any government organization, any business firm or may be a group of users. These users have to pay to the service providers so that they can become the licensed users for that channel. Then this channel will be dedicated only to that group of user who owns it. Now everybody in that group is using that channel as per their need. But the problem here is that these users are not using these channels all the time even if they are doing their work and paying for the spectrum [16] [20]. Another problem is the ever increasing demand for the wireless applications for which there is need of extra spectrum. As the radio spectrum is a natural resource no one can increase it for extra usage and almost all the spectrum is already allocated to licensed users leaving some part of the spectrum for unlicensed users. Copyright Š 2016 MECS

Fig.1. PUE Attack

II. RELATED WORK There is much vulnerability in the network at each layer. But attacks during the sensing of spectrum most

I.J. Computer Network and Information Security, 2016, 12, 44-50


A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

affect the network. There are many techniques for detection and prevention of PUE Attack in the literature [30]. A technique is proposed to detect malicious user by taking the difference of secondary user from primary transmitter and by calculating ratio. This is done by using GPS [10]. Another approach based on threshold value which is calculated by taking the ratio of maximum and minimum Eigen value [2]. Energy detection is also used by Fenton's approximation method to calculate mean and variance of distribution [3]. An attacker is found on the basis of signal characteristics and position of primary user is located by GPS [4]. Another technique is based on the probability of missing the primary user and probability of successful PUEA [5]. Now a new technique is proposed based on the trust value of secondary user [6]. In the next approach signal strength is calculated to find the attacker [30]. Another technique is

45

based on increasing the detection probability of PU which decreases attacker's probability [8]. Attacker can also be found by calculating the belief value and compared with the threshold value [9]. Another method to detect attacker is to focus on the authentication of primary user [15]. Now a new method is proposed by making interference signals and then compared with the position of primary user to locate the attacker [11]. After this a new method is proposed where a reference signal is generated then it is encrypted with AES and it is regenerated at the receiver's end [12]. After this a new proposal came which combines both energy detection and location verification to provide better results [13]. After this in a method sensing results are combined with routing information [14]. In the current research energy level of users are checked by cyclostationary feature [1]. A comparative study of all mentioned techniques is in table 1.

Table 1. Brief Summary of PUE Attack Solutions Technique

Year

Advantage

Physical layer approach

2005

Easy to implement antennas

Network layer approach

2005

Network bandwidth is not wasted due to ignorance of some SUs in path

Limitation Require prior knowledge about signal characteristics Require prior knowledge about location of attacker

DRT&DDT

2006

GPS system is used

Deprivation in result

MME

2007

Doesn't require prior knowledge

Calculating Eigen values is quite complex

Mac layer approach

2008

QoS of network improved

Fenton's method

2008

Easy to implement

Loc based verification

2008

Works effectively in hostile environment

NPCHT& WSPRT

2009

Flexible in maintaining successful PUEA high

More time complexity

2009

Works effectively for one malicious user

Doesn’t check for multiple malicious users

Robust spectrum decision protocol

2010

Probability of successful PUEA detection is quite good

Individual detection is applied for all users

Cooperative Sensing

2011

Maximize detection probability of primary user

Focus is on detection of primary user

Cross layer approach

2011

Works effectively after combining the results from both layers

Conveying the information between layers may be sensitive

Belief Propagation

2012

Probability of accuracy is more

Node may pass wrong belief value

2012

Authentication of every primary user reduces presence of malicious user

Require knowledge of cryptography

PNC

2013

Use additive nature of electromagnetic waves

Quite complex to find the location

AES

2013

Work effectively even under very low SNR values

May not be that much effective after a range

SPARS

2014

Doesn't need any prior knowledge about primary users

Works when Pus having same SAP

2015

A fruitful technique to increase probability of false alarm

Somewhat complex

2015

Robust technique

Results are based on energy detection technique

approximation

Collaborative sensing

Primary Authentication

Database Approach

spectrum

Spectrum

User

Assisted

Intense explore algorithm

Copyright Š 2016 MECS

Do not have focus on preventing malicious activity Mean & variance is calculated for all users at SU Might not useful when transmitter power is low

I.J. Computer Network and Information Security, 2016, 12, 44-50


46

A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

III. PROBLEM DEFINITION

B. Procedure

Current research in prevention of Primary User Emulation Attack is based on either energy detection or location based technique. Here two sets of secondary users are made, users from one group sense the behavior of their neighboring users in other group by using energy detection technique. If the energy level of any user is found to be greater than the threshold value of signal energy then it is considered to be suspected user. Further if another user in first group sense the same behavior about that user then fusion center believe it to be a malicious user and alerts all other secondary users about it. Energy detection technique provides good results but due to environment conditions energy level of signal may change under the noise uncertainty. In that condition it may not work accurately. In location based techniques position of users are found by using GPS, which is not convenient because GPS may not work everywhere and all the time.

IV. PROPOSED WORK Security in CRN is an important topic of research and attacks in CRN is an important component. Here a solution is proposed to provide better results based on Signal Activity Pattern (SAP) of a transmitter that works on the ON and/or OFF periods of the transmitter. An ON period refers to the duration of a busy period that the transmitter is transmitting and SUs must be denied from communications. An OFF period refers to the time limit of an idle period between two adjacent ON periods. A. Working environment Figure 2 shows environment of working where sensing result table contains SAPs of secondary users obtained by their neighboring secondary users. Table fusion center contains SAPs of all secondary users and primary users. Fusion center derives the result by comparing these SAPs.

To set up the environment two types of nodes are taken one is fixed and the other is movable. Nodes which are kept fixed called primary user nodes and the nodes which are movable called secondary user nodes. Secondary user nodes keep moving in the network to locate the holes in the network and whenever finds a vacate channel of primary user use it. SAP contains information of activation and deactivation time of that node in the network along with the energy levels of the signals produced by the node at any particular time. So one secondary user node senses the behavior of its neighboring secondary user node and makes its SAP, store it in a table. Similarly all the secondary nodes make SAP of their neighboring nodes and store the SAPs in a table named 'Sensing Results'. This process continues for 5 seconds after starting. Now these SAPs are given to the fusion center. As the same node may be in neighborhood of more than one node so, all these neighboring nodes sense the behavior of that particular node and make its SAP. Now when there is more than one entry in the sensing table for the same node then the SAP of that node is selected on the basis of maximum energy level. Similarly SAPs of all the secondary nodes are selected and given to the Fusion Center. It is considered that all the primary users are active at the same time in the network i.e. their ON/OFF period is same in the network and their energy level is also same due to equal priority of licensed users in the network. So SAP of all primary users will be same and it is saved in the Fusion Center already. Now all the SAPs are with Fusion Center. It compares the SAPs of all secondary user nodes with the SAP of primary user nodes. Then the secondary user node which is active in the off period of primary user node and its energy level is also equal to or greater than the primary user node is considered as attacker node. After detection of attacker node it is blocked by the Fusion Center and then it broadcasts a message in the network telling all the nodes about the attacker node. Now detection latency is obtained by calculating the number of malicious users that are detected by our proposed method within the given time. Throughput of the network is also calculated on the basis of results obtained by the simulation. Then these results are compared with the previous technique and it seems to give better results. Step by step working is shown as under:E (p) – Signal energy level of primary user E(s) - Signal energy level of secondary user T (p) - Time of activation of Primary user in network T (s) - Time of activation of Secondary user in network

Fig.2. Working Environment

Copyright Š 2016 MECS

Step 1: Set up CRN environment. Step 2: Initially Signal Activity Pattern (SAP) of all SUs is same and these patterns are stored at Fusion Centre (FC).

I.J. Computer Network and Information Security, 2016, 12, 44-50


A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

47

Step 3: Get SAPs of each SU from their neighboring SUs. Step 4: Give all SAPs to FC after each interval of time. Step 5: If ((E (p)>= E(s ) )&& (T (s)!=T(p)) then i. ii. iii.

SU is malicious. FC will block it. FC broadcast this message about attacker in the network.

Step 6: Else SU is not malicious.

V. IMPLEMENTATION RESULTS

Fig.4. Nodes Positions after Start

A simple network terminology is used for simulation. Simulations are conducted with six primary users and 7 secondary users that are located randomly surrounding the primary users. PUs are fixed whether SUs are moving with random speeds in random directions. Here we consider the minimum distance between primary and secondary user 200 meter and maximum distance is considered 400 meter. Parameters used for implementation are given in table 2.

Step 1: Secondary nodes start moving and sense their neighboring secondary nodes for 4 seconds and make their SPA.

Table 2. Parameters used in Implementation Parameter Total number of nodes, N Number of PU nodes Number of SU nodes Number of channels

Parameter value 13 6 7 1

Maximum velocity with which SU nodes can move

10kmph

Total area

1000m*1000m

Area within which nodes are considered adjacent to each other

200m Fig.5. Sensing Results

The implementation results of the proposed method are as follows:

Fig 4 shows corresponding sensing results in sensing table 5 after sensing for 4 seconds. Step 2: SUs give sensing results to FC and nodes having energy > PUs' energy is shown in FC database. Fig 6 shows the corresponding results in FC's database

Fig.3. Nodes Positions before Start

Fig. 3 shows the positions of Pus and SUs in the environment at a time before starting the implementation. Fig. 4 shows the positions of Pus and SUs in the environment after starting the implementation. Copyright Š 2016 MECS

Fig.6. FC Database

I.J. Computer Network and Information Security, 2016, 12, 44-50


48

A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

Step 3: Plot SPAs of all SUs and PUs

increases because every secondary user that is active in the network continuously sense the activity of its neighboring secondary users and give its SPA to FC at an interval of time. Fig 10 shows no. of malicious users versus throughput of network in our proposed method and its comparison with the existing technique. In our approach throughput increases as the number of malicious users increases because probability of detection of attackers increases as the number of malicious users increases. Due to this the channels are utilized efficiently thereby increasing the throughput of network.

Fig 7 shows the corresponding results in graphs

Fig.7. SPAs of SUs and PUs

Step 4: FC finds attacker nodes, block them and broadcasts message in the network. Fig 8 shows the corresponding results in nodes blocked table

Fig.10. Comparison of Throughput

VI. CONCLUSION AND FUTURE WORK Cognitive radio was introduced to utilize the unused spectrum efficiently to improve the spectrum utilization and hence to reduce spectrum scarcity. Spectrum sensing is one of the important aspects of cognitive radio network. In this paper, we address the problem of preventing PUE attacks in mobile CRNs. Its various detection and prevention methods are also presented. Mitigation of PUE attack is considered in this paper through our novel approach, Signal Activity Patterns. The simulation results prove the method is robust.

Fig.8. Blocking Nodes and Broadcasting Message by FC

Table 3. Comparison between Existing and Proposed Technique

Fig.9. Comparison of Detection Accuracy

Fig 9 shows no. of malicious users versus detection accuracy in our proposed method and its comparison with the existing technique. In our approach detection accuracy increases as the number of malicious users Copyright Š 2016 MECS

Parameters Techniques

Detection accuracy

Throughput

Existing

Up to 65%

Up to 60%

Proposed Model

Up to 70

Up to 65%

In our future work, we will improve the accuracy of detecting malicious user by taking variable activation time with energy and acquiring SUs trust value. . In our future work, we will improve the accuracy of detecting malicious user by taking variable activation time with energy and acquiring SUs trust value. Since all secondary user nodes are not trustworthy so, to find attacker node among these secondary user nodes is difficult task.

I.J. Computer Network and Information Security, 2016, 12, 44-50


A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

REFERENCES [1]

[2]

[3]

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

C. Sumathi, & R. Vidhyapriya, “Intense explore algorithm-A Proactive Way to Eliminate PUE attacks in cognitive radio networks”. International Journal of Applied Engineering Research, 2015, 10 (2), 3827–3842. Y. Zeng, & Y.C. Liang,, “Maximum-Minimum Eigenvalue Detection for Cognitive Radio”. IEEE International Symposium on Personal, Indoor, and Mobile Radio Communications, 2007, 1-5. S. Anand, Z. Jin, & K. P. Subbalakshmi, “Analytical model for primary user emulation attacks in cognitive radio networks”. IEEE Symposium of New Frontiers in Dynamic Spectrum Access Networks (DySPAN’2008), 16. R. Chen, J.-M. Park & J. H. Reed, “Defense against Primary User Emulation Attacks in Cognitive Radio Networks”. IEEE Journal on Selected Area in Communications, 2008, 26(1), 25-37. Z. Jin, S. Anand, & K. P. Subbalakshmi, “Mitigating primary user emulation attacks in dynamic spectrum access networks using hypothesis testing”. ACM Mobile Computing and Communications Review (MC2R): Special Issue on Cognitive Radio Networks, 2009, 13(2), 74-85. W. Wang, “Attack-Proof Collaborative Spectrum Sensing in Cognitive Radio Networks”, 130–134. S. Bhattacharjee, S. Sengupta, & M. Chatterjee, “Vulnerabilities in cognitive radio networks” A survey. Elsevier-Computer Communications, 2013, 36(13), 1387– 1398. Chen, H. Cheng, & Y. Yao, “Cooperative Spectrum Sensing in Cognitive Radio Networks in the Presence of the Primary User Emulation Attack”, IEEE Transactions on Wireless Communications, 2011,10(7), 2135–2141. Z. Yuan, D. Niyato, H. Li, J.B. Song & Z. Han,“ Defeating Primary User Emulation Attacks Using Belief Propagation in Cognitive Radio Networks”. Selected Areas in Communications, IEEE Journal, 2012, 30 (10), 1850-1860. R. Chen & J.-M. Park, “Ensuring Trustworthy Spectrum Sensing in Cognitive Radio Networks” Networking Technologies for Software Defined Radio Networks. SDR ’06.1st IEEE Workshop on, 2006, 110–119. X. Xie, & W. Wang, “ Detecting primary user emulation attacks in cognitive radio networks via physical layer network coding”, Proceeding of Computer Science, 2013, 21, 430–435. Alahmadi, M. Abdelhakim, J. Ren, & T. Li, “Mitigating primary user emulation attacks in cognitive radio networks using advanced encryption standard”. Global Communications Conference (GLOBECOM), IEEE, 2013, 3229–3234. R. Yu, Y. Zhang, Y. Liu, S. Gjessin, & M. Guizani, “Securing cognitive radio networks against primary user emulation attacks”. IEEE Networks, 2015, 29(4), 68–74. Sorrells, P. Potier, L. Qian & X. Li, “Anomalous spectrum usage attack detection in Cognitive Radio Wireless Networks”. IEEE International Conference on Technologies for Homeland Security (HST), 2011, 384389. Meena Thanu, “Detection of Primary User Emulation Attacks in Cognitive Radio Networks”. International Conference on Collaboration Technologies and Systems (CTS), 2012, 605-608 Clancy, C., Hecker, J., Stuntebeck, E., and O'Shea, T.,

Copyright © 2016 MECS

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

49

“Applications of machine learning to cognitive radio networks”, Wireless Communications, IEEE, 14(4):47-52, (2007). S. Haykin, “Cognitive Radio : Brain-Empowered wireless communications”, IEEE Journal on Selected Areas in Communications , 2005, 23(2), 201–220. Das, “Primary User Emulation Attack in Cognitive Radio Networks ” A Survey. International Journal of Computer Networks and Wireless Communications, 2013, 3(3), 312–318. Y. C. Liang, K. C. Chen, G. Y. Li, &, P. Mahonen, “Cognitive radio networking and communications: An overview. IEEE Transactions on Vehicular Technology”, 2011, 60(7), 3386–3407. Z. Jin, S. Anand& K. P. Subbalakshmi, “Impact of Primary User Emulation Attacks on Dynamic Spectrum Access Networks”. IEEE Transaction Communications, 2012, 60(9), 2635–2643. Abhilasha Singh, Anita Sharma,” A Survey of various Defense Techniques to detect Primary User Emulation Attacks”, International Journal of Current Engineering and Technology, Vol.4, No. 2, April 2014. Deepa Das, Susmita Das, “Primary User Emulation Attack in Cognitive Radio Networks: A Survey”, IRACST, Vol.3, No3, June 2013. Carl R. Stevenson, Gerald Chouinard, “IEEE 802.22: The First Cognitive Radio Wireless Regional Area Network Standard”, IEEE Communications Magazine, January 2009. Zeng, Y., Liang, Y.-C., Hoang, A. T., and Zhang, R., “A review on spectrum sensing for cognitive radio: challenges and solutions”, EURASIP J. Adv. Signal Process, 2010. V. Kukreja, S. Gupta, B. Bhushan, & P. Mittal, “Enhancement of Spectrum Efficacy using Cognitive Radio Networks”. International Journal of Future Generation Communication and Networking, 2015 8(2), 265–272. C. Kiruthika, A.C. Sumathi, “A Study on Primary User Emulation Attack in Cognitive Radio Networks”, International Journal of Computer Science Engineering and Technology (IJCSET), 2014, 4(10), 260–262. Adelantado, & C. Verikoukis, “Detection of malicioususers in cognitive radio ad hoc networks”, A non-parametric statistical approach. Ad Hoc Networks, 2013, 11 (8), 2367–2380. Jin, Z. Anand, S., "Impact of Primary User Emulation Attacks on Dynamic Spectrum Access Networks, "Communications, IEEE Transactions, vol.60, no.9, pp.2635, 2643, September 2012. Z. Jin, S. Anand, & K.P. Subbalakshmi, “Robust Spectrum Decision Protocol against Primary User Emulation Attacks in Dynamic Spectrum Access Networks”. Global Telecommunications Conference (GLOBECOM) IEEE, 2010, 1-5. E. Gupta, Poonam & C. K. Nagpal, “Survey on PUE Attack Detection and Prevention Techniques”, International Journal of Emerging Technologies in Engineering Research (IJETER-2016), 4(4), 90–95. Vivek Kukreja, Shailender Gupta, Bharat Bhushan and Chander Kumar, “Towards Performance Evaluation of Cognitive Radio Networks in Realistic Environment”, published in IJCNIS-2014, pp. 61-77, DOI: 10.5815. Poonam Mittal, Mehak Jain and C. K. Nagpal, “A Throughput and spectrum aware fuzzy logic based routing protocol for CRN”, published in IJCNIS-2016, pp. 58-64, DOI: 10.5815.

I.J. Computer Network and Information Security, 2016, 12, 44-50


50

A Novel Technique to Prevent PUE Attack in Cognitive Radio Network

Authors’ Profiles Poonam received her B.Tech. and M.Tech. from YMCA University of Science and Technology, Faridabad, India. She is currently working as an Assistant Professor in Computer Engineering Department in the same university. She is currently pursuing Ph.D. Her interests include networking and algorithm design.

C.K. Nagpal, Professor, head of department and Ph.D. supervisor in Computer Engineering Department in YMCA University of Science and Technology, Faridabad, India. His interests include networking and fuzzy expert system. He has published more than 35 papers in various national and international publications.

Ekta Gupta, received her B.Tech. from MVN University and pursuing M.Tech. from YMCA University of Science and Technology, Faridabad, India. Her area of interest is Security in MANET.

How to cite this paper: Poonam, Ekta gupta, C.K. Nagpal,"A Novel Technique to Prevent PUE Attack in Cognitive Radio Network", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.4450, 2016.DOI: 10.5815/ijcnis.2016.12.06

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 44-50


I. J. Computer Network and Information Security, 2016, 12, 51-58 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.07

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator Er. Richa Thakur Department of Computer Science, Himachal Pradesh University, Shimla, India E-mail: richathakur.uiit@gmail.com

Prof. K.L. Bansal Department of Computer Science, Himachal Pradesh University, Shimla, India E-mail: kishorilalbansal@yahoo.com

Abstract—Delay/disruption Tolerant Networks (DTNs) provide connectivity in those networks which lack continuous connectivity or considerable delays like that of terrestrial mobile networks, military ad-hoc networks, sensor or planned network in space. They lack in an endto-end path between Source and Destination resulting in long variable propagation delays. The Internet Protocols do not operate properly in these networks, thus raising a variety of new challenging problems in this area. The DTN effectively improves the network communications where the connectivity in the network is intermittent or is prone to disruptions. Routing in DTNs is challenging because of long and frequent time durations of nonconnectivity. There are several routing approaches that have been proposed with strategies ranging from flooding to forwarding approaches. In this paper these protocols are analyzed based on the quantitative data gathered by simulating each protocol in ONE simulator environment. The performance is discussed and compared for different routing protocols and results are discussed for different performance metrics. Index Terms—Delay Tolerant Networks (DTNs), ONE simulator tool, Routing in DTN.

I. INTRODUCTION Today wired and wireless networks have enabled a wide range of devices to be interconnected all over the world. Mobile devices like smart phones are gaining increasing importance both in private and professional sector. Around 4.61 billion mobile phone users exist today, in 2016, which are estimated to become 5.07 billion in 2019 [1]. 3.4 billion Internet users [2], which make around 40% of the world population, shows that today existing networks are very successful networks. But in a strong contrast to omnipresent Internet access in the developed world, there are still 4.0 billion nonInternet users in the world at present [2]. Further, developing countries have very low internet rates where only 22% internet rates are there in India [3]. According to a new United Nations report on 21st September 2015broadband Internet is failing to reach billions of people Copyright © 2016 MECS

living in the developing world, including 90% of those living in the poorest nations [4]. One of the reasons of the non-availability of Internet access to the people is our over-burdened existing technology. The traffic volume generated by today’s mobile-connected devices shows exponential growth with a 26-fold increase between 2010 and 2015, reaching over 6 Exabyte per month in 2015 [5]. Moreover, the current network technology still cannot reach everywhere, and also for some applications their infrastructure cost is very high. The two main reasons for these limitations are: the infrastructure-based design of the existing technologies and the fundamental assumptions on which these technologies rely. The first and most important of these assumptions is an end-to-end connectivity from the source to the destination, possibly via multiple intermediate nodes which can be easily violated due to power savings (as in case of sensor networks), mobility or unreliable networks connections in which the nodes are exposed to long delays or may be disconnections. So, here comes the Delay-tolerant networking (DTN) which is an attempt to provide the infrastructure-less networking, beyond the reach of existing networks. . Due to their infrastructure-less nature, they can be deployed where infrastructure access is not available or possible, or to offload the congested infrastructure networks. The main feature of DTNs is disruption or delay which is mainly because of limited wireless radio range of widely scattered mobile nodes, limited energy resources, interference and attacks etc. Hence routing in these networks become a challenging task and is an active research area. Unlike the conventional routing strategies, DTNs lack permanent network connectivity which makes its routing more challenging. The lack of instantaneous path connections result in high latency of data delivery, overall low data rates, long queuing delays and limited longevity of individual nodes. This paper aims to analyze the performance dynamics of various existing protocols on the basis of essential resources and performance metrics of a network i.e. buffer size, message creation-rate, delivery probability, average latency and overhead ratio. The organization of the paper is as follow: Section I gives introduction about the DTN. Section II briefly describes the related work by

I.J. Computer Network and Information Security, 2016, 12, 51-58


52

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

various authors in the same field. Section III is focused on routing approaches and at last Section IV is the detailed description about the simulation results and analysis of the work done.

Finally Ari Keranen et al. [23] have given the detailed description about the ONE simulator which is a Javabased simulator for the evaluation of the DTNs. It offers a wide variety of tools to create complex mobility scenarios that come closer to reality than many other mobility models.

II. RELATED WORKS AND CONTRIBUTIONS The very first work of collecting and classifying the existing routing protocols of delay tolerant networks was done by E.P.C. Jones et al. [6] which classified the protocols on two key properties: Replication and Knowledge. The authors then identified the need of a hybrid technique that exploits both of the properties to work on real scenarios. Among various routing protocols Epidemic protocol [7] is a popular one using complete flooding approach for message transmission in the delay tolerant network. Ram Ramanathan et al. [8] have extended the work to get Prioritized Epidemic which uses expiry time information and topology awareness to decide which bundles to delete or hold back when faced with a resource crunch. Many papers have studied ways to make Epidemic Routing consume fewer resources [9, 10]. Flooding families create a lot of redundancy by generating message replicas and there is a need to remove these replicas from DTN [11]. On the other hand are forwarding protocols MaxProp [12] and Probabilistic Routing Protocol using History of Encounters and Transitivity (PRoPHET) which make use of mobility patterns for routing decisions. The DTN design makes a different set of choices in the architectural design of the protocols [13]: messages versus packets, a form of hopby-hop reliability and security versus end-to-end, name based routing versus address based routing, and a routing abstraction of partially-connected rather than fullyconnected network graph making the TCP/IP protocols useless in this scenario. Jian Shen et al. [14] have also surveyed the existing/proposed routing protocols. The protocols are mainly classified for Flooding and Forwarding families and both families are compared and analyzed in terms of various performance metrics. A further classification is given by Salman Ali et al. [15]. The authors classified the protocols on the basis of replication, knowledge and coding based and gave a comprehensive comparison of the DTNs routing strategies. Further a simulation based performance comparison of protocols is done in [16]. New protocols keep on emerging in DTNs. Lately Ahmad El Shoghri et al. [17] have proposed Augur that uses spatiotemporal information of the networking nodes and route the message in the network using this information. The authors then compared the performance of the existing protocols with the Augur on ONE simulator tool. Similarly CASPaR [18] is a congestion avoidance shortest path protocol for DTNs. Applications of DTNs are significantly recognized in various scenarios today [19]. The Disaster Response system in [20], HimSwan [21] - a healthcare system, Military [22] and Mobile Adhoc NETworks (MANETs) are some of the application areas which need DTNs. Copyright © 2016 MECS

III. ROUTING IN DTNS The important issue of routing in DTNs is a challenging task. Delay Tolerant Networks have to deal with disconnections, waiting time might range from seconds to days, buffer space of intermediate nodes must meet the demand of the network and finally energy consideration of individual nodes is an important task [24]. Delay Tolerant Networks forward messages opportunistically and cooperatively on occurrence of contacts between physical devices when mobile devices come into mutual communication range. They employ a store-carry-forward routing strategy where messages are stored for longer duration, carried through mobile devices and forwarded if the destination device or a better suited device is encountered. A. Store-Carry-Forward: Store-carry-forward is a message passing approach that a node follows after receiving a message. The “Store” phase adds the message to the node’s buffer that allows the data to wait for a suitable time to forward the message. In the “Carry” stage the message is propagated to other regions of the network physically by the movement of the node carrying the data. Finally, “Forward” is the stage when the node decides to send the message to another node due to the availability of other better candidates or to the message’s final destination [25].

Fig.1. Message Delivery in a DTN

Fig.1 shows an exemplary DTN scenario where a message is carried through and then delivered to the destination device: At time t1 device 5 sends a message destined for device 7. It forwards the message to device 3 as it is more capable of carrying the message to device 7.

I.J. Computer Network and Information Security, 2016, 12, 51-58


Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

At the network time t2 device 3 and destination device 7 comes in the communication range of each other and then the message is delivered to destination device by device 3. Note that there was no communication path between device 5 and device 7. The path only exists in parts over time. B. Flooding-based Routing strategies: Strategies in the flooding family transmit multiple copies of each message to a set of nodes called relays. These relays store the messages until they got connected with the destination, at which point the message is delivered. Usually, these strategies are studied in the context of mobile ad-hoc networks, where random mobility has a good chance of bringing the source into contact with the destination. Message replication is then used to increase the probability that the message gets delivered. The basic protocols in this family do not need any information about the network; however more advanced schemes use some knowledge to improve performance [6]. Direct-Contact/Direct-Delivery (DD): This is a simple routing approach in which the source delivers the message only to the final recipient. Clearly it is a one hop routing scheme. Because of its simple characteristics it does not consume many resources. First Contact (FC): In the first contact routing only a single copy of the message is used. The source node forwards the message to the first available contact, which in turn forwards it to the next first available contact and so on. Epidemic Routing (Epi): Epidemic routing is the extreme end of the flooding family approach. It works as follows: when a message is sent to a destination, it is first saved in a local buffer and tagged with a unique ID. When two nodes come in the contact range of each other they exchange the list of all the message IDs they have in their buffer and exchange those messages they don’t have in their local buffer. Hence in the end of this exchange both the nodes will have same messages. This process continues until all the nodes have all messages in their buffer. In epidemic routing, it is ensured that messages can be delivered with a high probability. However, the network resources are consumed heavily [26]. Spray and Wait: Spray and Wait is a controlled flooding scheme. It consists of two phases: Spray phase and Wait phase. Given an initial number of allowed replicated copies per message N, the spray phase sprays the message to the first encountered devices, keeping N/2 replicas and forwarding rest of it. The process continues recursively until N=1. If the destination is not found in the spraying phase, each of the N nodes carrying a message copy performs direct transmission to the destination node [27].

53

family require some knowledge about the network. They typically send a single message along the best path; hence they do not use replication. Probabilistic Routing Protocol using History of Encounters and Transitivity (PRoPHET): PRoPHET uses the statistics of previous encounters made by a node with other neighbors. In this routing strategy each node locally gathers meeting probabilities with other nodes when in mutual communication range. Replication of a message between two nodes in transmission range is performed if the node currently not storing the message has a higher meeting probability for the message’s destination than the node currently storing the message. In PRoPHET to forward data from one node to another it uses a probabilistic metric called delivery predictability, P (a, b) Ε [0, 1], at every node a for each known destination b. This metric indicates the likelihood of a node to deliver a message to that destination [28]. Delivery predictability is stored in internal delivery vector and it gets updated whenever nodes meet each other. This delivery predictability metric is used by each node and is recalculated at each opportunistic encounter according to these rules: (1) When a node a encounters another node b, the predictability for b is increased. This is shown in Equation (1). đ?‘ƒ(đ?‘Ž,đ?‘?) = đ?‘ƒ(đ?‘Ž,đ?‘?)đ?‘œđ?‘™đ?‘‘ + (1−đ?‘ƒ(đ?‘Ž,đ?‘?)đ?‘œđ?‘™đ?‘‘) Ă— đ?‘ƒđ?‘–đ?‘›đ?‘–đ?‘Ą

(1)

where đ?‘ƒđ?‘–đ?‘›đ?‘–đ?‘Ą is an initialization constant. (2) The delivery predictability of nodes must age because if two nodes do not encounter/meet each other in a while, then they are less likely to forward messages to each other. Equation (2) shows this ageing equation. đ?‘ƒ(đ?‘Ž,đ?‘?) = đ?‘ƒ(đ?‘Ž,đ?‘?)đ?‘œđ?‘™đ?‘‘ Ă— Ć”đ?‘˜

(2)

where Ć”đ?‘˜ is an aging constant. (3) The delivery predictability follows the transitive property i.e. if a node a frequently meets node b and node b frequently encounters node d, then node d probably is a good node to forward message intended for node a. The effect of transitivity on delivery predictability is shown in Equation (3). đ?‘ƒ(đ?‘Ž,đ?‘‘) = đ?‘ƒ(đ?‘Ž,đ?‘‘)đ?‘œđ?‘™đ?‘‘ + (1−đ?‘ƒ(đ?‘Ž,đ?‘‘)đ?‘œđ?‘™đ?‘‘) Ă— đ?‘ƒ(a, b) Ă— đ?‘ƒ(đ?‘?,đ?‘‘) Ă— β (3) where β is the scaling constant that decides how much large impact the transitivity should have on the delivery predictability.

C). History/ Prediction-based Routing Strategies: The routing strategies in this family use network topology information to select the best possible path in the network and the message is then forwarded from one node to another along this path. The strategies in this Copyright Š 2016 MECS

IV. PERFORMANCE EVALUATION AND SIMULATION To analyze the performance of various routing protocols we have used the Opportunistic Network

I.J. Computer Network and Information Security, 2016, 12, 51-58


54

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

Environment (ONE) simulator tool to create a real time scenario. We analyzed our simulation results mainly on the basis of three performance metrics viz. Delivery Probability, Average Latency and Overhead Ratio. Also an important resource in DTNs which is the Buffer Size is explored for the considered protocols. The main aim of the evaluation is to analyze the various aspects of the DTNs in order to gain the better insight and understanding of the limitations of the existing protocols in terms of various performance metrics. A. Simulation Tool: Simulation plays an important role in analyzing the behavior of DTN routing and application protocols. We used the ONE to extensively evaluate the different performance dynamics of the various routing strategies. ONE is a powerful simulator used to implement realistic DTN scenarios. It is a Java-based tool offering a broad set of DTN protocol simulation capabilities in a single framework [23]. The ONE simulator can be run on Linux, Windows or any other platform supporting Java. It is an agent-based discrete event simulation engine that is designed for evaluating the performance of DTN routing protocols. At each simulation stage, the engine updates a number of modules that implement the main simulation functions. Unlike other DTN simulators which focus only on routing simulation, the ONE combines mobility models, inter-node contacts, DTN routing protocols, message handling and visualization in one package and provides a rich set of reporting and analyzing modules. A detailed description of the simulator is available in [23] and the ONE simulator project page [29], where the source code of the simulator is also available. Source codes are written in Java programming language.

nodes. On the right hand side is the list of nodes taking part in the simulation with each node having a unique id (e.g. p0, p1, p2‌). Messages created in a simulation run also have unique ids. Two green circles in the snapshot are the radio communication range of the two nodes. All settings related to the simulation and for the nodes are done in default_settings file provided with the simulator. In order to analyze the performance metrics of different DTN routing protocols we visualized the simulation in real-time and analyzed the various reports generated by ONE simulator using the settings shown in Table 1. Table 1. Settings used for ONE Simulator Parameters Simulation time Bus Movement Model Number of Nodes Message size Message ttl (time-to-live) Interface transmission speed Interface transmission range

Values 7200s = 2 hours Map Based Movement 126 500KB – 1MB 120min 250kbps 100m

B. Performance Metrics used: The DTN protocols are evaluated and analyzed using the following performance metrics: Delivery Probability: It is the ratio of the number of delivered messages to the total number of messages created by the source node. đ?‘’đ?‘™đ?‘– đ?‘’ đ?‘ƒ đ?‘œđ?‘?đ?‘Žđ?‘?đ?‘–đ?‘™đ?‘–đ?‘Ą đ?‘šđ?‘?đ?‘’ đ?‘œđ?‘“ đ?‘šđ?‘’ đ?‘Žđ?‘”đ?‘’ đ?‘‘đ?‘’đ?‘™đ?‘– đ?‘’ đ?‘’đ?‘‘ đ?‘?đ?‘?đ?‘’ đ?‘“ đ?‘™đ?‘™ đ?‘œđ?‘Ąđ?‘Žđ?‘™ đ?‘› đ?‘šđ?‘?đ?‘’ đ?‘œđ?‘“ đ?‘šđ?‘’ đ?‘Žđ?‘”đ?‘’ (4) Average Latency: It is the average time taken by the messages from their creation to their first delivery at the destination node. đ?‘’ đ?‘Žđ?‘”đ?‘’ đ?‘Žđ?‘Ąđ?‘’đ?‘›đ?‘? đ?‘’đ?‘™đ?‘– đ?‘’ đ?‘Ąđ?‘–đ?‘šđ?‘’ − đ?‘’đ?‘Žđ?‘Ąđ?‘–đ?‘œđ?‘› đ?‘Ąđ?‘–đ?‘šđ?‘’ ( ) đ?‘œđ?‘Ąđ?‘Žđ?‘™ đ?‘‘đ?‘’đ?‘™đ?‘– đ?‘’ đ?‘’đ?‘‘ đ?‘šđ?‘’ đ?‘Žđ?‘”đ?‘’ (5) Overhead Ratio: Overhead ratio is another important metric which shows how efficient a protocol is in terms of correct relay decisions. (

Fig.2. ONE Simulator at Work

Fig. 2 shows a snapshot of the ONE simulator at work. The graphical user interface appears once the simulator starts running. The main window in the snapshot shows the map of Helsinki city which is the by-default map provided in the simulator. Bottom left window shows the event log controls which can be controlled by various checkboxes provided in there. Beside this window is the event log window showing various connections between Copyright Š 2016 MECS

đ?‘’ đ?‘•đ?‘’đ?‘Žđ?‘‘ đ?‘Žđ?‘Ąđ?‘–đ?‘œ đ?‘šđ?‘?đ?‘’ đ?‘œđ?‘“ đ?‘Ą đ?‘Žđ?‘› đ?‘šđ?‘–đ?‘Ąđ?‘Ąđ?‘’đ?‘‘ đ?‘šđ?‘’ đ?‘Žđ?‘”đ?‘’ đ?‘•đ?‘–đ?‘?đ?‘• đ?‘Ž đ?‘’ đ?‘›đ?‘œđ?‘Ą đ?‘‘đ?‘’đ?‘™đ?‘– đ?‘’ đ?‘’đ?‘‘ ( đ?‘œđ?‘Ąđ?‘Žđ?‘™ đ?‘› đ?‘šđ?‘?đ?‘’ đ?‘œđ?‘“ đ?‘Ą đ?‘Žđ?‘› đ?‘šđ?‘–đ?‘Ąđ?‘Ąđ?‘’đ?‘‘ đ?‘šđ?‘’ đ?‘Žđ?‘”đ?‘’ (6)

C. Simulation Results: Impact of Buffer Size: Buffer storage is a valuable resource the participating

I.J. Computer Network and Information Security, 2016, 12, 51-58


Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

nodes have in a network. We start our study by exploring this resource for various participating protocols and then choosing an optimum buffer size on which all protocols can run optimally. We run our experiment for following buffer sizes: 2MB, 4MB, 8MB, 12MB and 15MB. Message creation rate is set to 1 every 30 seconds. Our aim is to see the effect of varying the buffer size on delivery probability of different routing protocols. Fig. 3 shows the simulation results for various protocols on different buffer sizes. The performance metric delivery probability is analyzed in this simulation. The simulation results show that in Direct Delivery (DD) protocol buffer sizes greater than 2MB do not results in high delivery probability. The reason for this behavior is that increasing the buffer size will surely increase the number of messages stored in the local buffer but it does not guarantee the availability of destination node for receiving the messages. Same results are obtained for First Contact (FC) routing protocol. However Epidemic (Epi), Spray and Wait (S&W) and PRoPHET deliver more messages with increase in buffer size. This is because Epidemic and Spray and Wait protocols use flooding approach for message transmission and more buffer size means less number of messages dropped. Note that there is no increase in delivery probability of all protocols except Epidemic when increasing the buffer size from 12 MB to 15 MB. Epidemic routing performs well compared to other routing protocols but at the cost of extensive resource usage.

55

at a scale from 0 to 1. An interesting thing to observe is that all these protocols are far from a sure delivery probability (i.e. a probability of 1 in current scenario) which shows all DTN protocols are still in their infancy and a lot improvement in routing approaches is needed to enhance the probability of delivering messages. A further exploration of buffer size is shown in Fig. 5. The figure shows the number of messages dropped using different buffer sizes (message statistics are given in Table 3). On using a small size buffer of 2MB the number of messages dropped are high as there is less space to store the created messages. Epidemic routing has the highest message dropping rate which is because of extensive flooding used by it. The approach here for dropping message is drop-oldest-message. When increasing the buffer size, the number of messages dropped is decreased and this number goes down to 0 for 15MB buffer size except Epidemic routing. Using the result of this simulation we chose our optimal buffer size for all protocols as 12MB to further analyze the different performance metrics of a DTN protocol.

Fig.5. Number of Messages Dropped for Different Buffer Sizes

Table 2 shows the message statistics for different protocols at different buffer sizes which are used for analyzing different protocols. Table 2. Message Statistics for different Buffer Sizes Message creation Rate = 1 message/30seconds Messages

Fig.3. Delivery Probability vs. Buffer Size

Created Started

Dropped

Delivered

Fig.4. Delivery Probability in a Range from 0 to 1

Fig. 4 shows the delivery probability of the protocols Copyright Š 2016 MECS

Protocols All DD FC Epi S&W PRoP HET DD FC Epi S&W PRoP HET DD FC Epi S&W PRoP HET

2MB

4MB

8MB

12MB

15MB

242 29 1739 2937 1797

242 33 2111 3491 2224

242 33 2143 3572 2272

242 33 2143 3542 2277

242 33 2143 3523 2277

1806

2141

2177

2135

2130

67 135 1196 621

6 27 1180 472

0 0 585 95

0 0 183 10

0 0 52 0

631

545

197

23

0

13 9 19 12

14 14 19 25

14 14 30 30

14 14 31 31

14 14 32 32

17

22

28

27

27

I.J. Computer Network and Information Security, 2016, 12, 51-58


56

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

Overhead Ratio Overhead ratio is another important metric which shows how efficient a protocol is in terms of correct relay decisions. We run our simulation for the chosen buffer size of 12MB and analyzed the overhead created by each protocol.

taking any routing decisions and use the simple direct delivery option since it would not load nodes further. While on low data rates (1 message per 240s) Epidemic routing performs well. Table 3. Average Latency in Seconds for Different Message Creation Rate Message Rate (in sec) 1/30 1/60 1/120 1/240

Fig.6. Overhead created by considered protocols at 12MB Buffer size

Fig. 6 shows the overhead created by considered protocols at 12MB buffer size. Obviously Direct Delivery routing has zero overhead as no messages are transmitted by it to the intermediate nodes. From rest of the considered protocols PRoPHET which is a forwarding family protocol performs the best on optimal buffer size. This is mainly because of correct relay decision taken by it at the time of transmitting a message to the intermediate node using the history data of the nodes in the network. Epidemic routing has a high overhead ratio while Spray and Wait protocol has overhead almost equal to PROPHET. This shows that a controlled flooding as done by S&W significantly reduces the overhead in a high traffic scenario.

DD

FC

Epi

S&W

PRoP HET

1827.2 2022.0 1145.6 3700

2197.2 2697.6 1211.5 4136.6

2610.8 3314.3 3140.7 3473.6

2555.9 2381.7 2454.3 2417.2

2464.5 2869.9 3854.7 3411.8

Simulation Results: On the basis of the detailed analysis and quantitative data gathered by us the performance of the different protocols can be summarized in the Table 4. On a scale from 1 to 5 where 1 is least, 2 is low, 3 is moderate, 4 is high and 5 is highest the different protocols are given values. Table 4. Simulation Results Buffer Size = 12MB Metric

DD

FC

Epi

S&W

PRoPHE T

Overhead Ratio

0

High

Highest

Low

Least

Least

Low

Highest

High

Moderate

High

Highest

Low

Least

Moderate

Low

Low

High

High est

Moderate

At high traffic At low traffic Delivery Probability

Avera ge Laten cy

Average Latency: Finally the Average Latency at different loads is analyzed. Simulation is carried out on buffer size of 12MB and message creation rate is varied. Message creation rates are varied from 1 message created per 30 seconds to 1 message created per 240 seconds.

The overhead ratio of PRoPHET is the lowest while of Epidemic protocol is the highest. Average Latency at high traffic (i.e. 1 message/30s) for Epidemic is the highest while for Direct Delivery protocol it is the lowest. At low traffic (1 message/240s) Spray and Wait performs considerably good while First Contact routing performs the worst. Delivery probability of Spray and Wait is the highest. With these results it can clearly be seen that no one protocol is best in every metric. There is always a tradeoff between one or more performance metrics.

V. CONCLUSION AND FUTURE WORK

Fig.7. Average Latency vs. Message Creation Rate

Fig. 7 shows the average latency of the protocols with detailed statistics given in Table 3. Epidemic routing has the highest average latency rate while Direct Delivery protocol has the lowest on high traffic (1 message per 30s). This shows that at higher traffic load one must stop Copyright Š 2016 MECS

Routing in Disruption Tolerant Networks is a challenging task. Different routing protocols have tried to minimize the delays, which are unavoidable, in DTNs but there is still a trade-off between high resource consumption and high delivery ratio of a protocol. In this paper we have analyzed the five DTN protocols on quantitative data gathered from simulating the protocols environment on ONE simulator. The results show that simple flooding protocol like Epidemic has high delivery probability but the message latency is high. On the other

I.J. Computer Network and Information Security, 2016, 12, 51-58


Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

hand forwarding protocols performs good but at the cost of resource consumption. Also all these protocols are far away from the guaranteed delivery probability of 1. Future work in the DTN could be to consider the tradeoff between the different performance metrics of various protocols and exploit a hybrid technique which makes use of flooding as well as forwarding to gain the best performance for a specific application. For this purpose the deep analysis of the different aspects of these protocols from different angles are required. REFERENCES [1] [2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

Mobile phone users worldwide 2013-2019, http://www.statista.com, (retrieved July 20, 2016). Internet users in the world, http://www.internetlivestats.com, (retrieved July 20, 2016). World Internet Users-Internet World Stats, http://www.internetworldstats.com, (retrieved July 21, 2016). Billions of people in developing world still without Internet access, UN News Centre, http://www.un.org, (retrieved July 21, 2016). Cisco Systems Inc. Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2010-2015. http://newsroom.cisco.com/dlls/ekits/Cisco_VNI_Global_ Mobile_Data_Traffic_Forecast_ 2010_2015.pdf February 2011. E. P. C. Jones and P. A. S. Ward, “Routing Strategies for Delay-Tolerant Networks”, Submitted to Computer Communication, 2008. A. Vahdat and D. Becker, “Epidemic Routing for Partially Connected Ad Hoc Networks” Tech. rep. CS200006, Duke University, Apr 2000. Ram Ramanathan et al., “Prioritized Epidemic Routing for Opportunistic Networks”, Proceedings of the 1st international MobiSys workshop, p 62-66, NY USA 2007. Teerapong Choksatid, Worrawat Narongkhachavana and Sumet Prabhavat, “An Efficient Spreading Epidemic Routing for Delay-Tolerant Network”, 13th IEEE Annual Consumer Communications & Networking Conference (CCNC), 2016. Bhed Bahadur Bista and Danda B. Rawat, “A Robust Energy- Efficient Epidemic Routing Protocol for Delay Tolerant Networks”, IEEE International Conference on Data Science and Data Intensive Systems, 2015. H. Bindra and A. L. Sangal, “Need of Removing Delivered Message Replica from Delay Tolerant Network- A Problem Definition”, MECS publisher, International Journal of Computer Network and Information Security, IJCNIS Vol. 4, No.12, November 2012. John Burgess et al., “MaxProp: Routing for Vehiclebased Disruption Tolerant Networks”, Proceedings of IEEE Infocom, 2006. Kevin Fall, "A delay-tolerant network architecture for challenged internets," 2003 Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications, pp.27-34, 2003. Jian Shen, Sangman Moh, and Ilyong Chung, “Routing Protocols in Delay Tolerant Networks: A Comparative Survey”, The 23rd International Technical Conference on Circuits/Systems, Computers and Communications (ITCCSCC 2008).

Copyright © 2016 MECS

57

[15] Salman Ali, Junaid Qadir, Adeel Baig, “Routing Protocols in Delay Tolerant Networks – A Survey”, 6th International Conference on Emerging Technologies (ICET), 2010. [16] Tamber Ambdelkader, Kshirasagar Naik, Amiya Nayak, Nishith Goel and Vineet Srivastava, “A Performance Comparison of Delay-Tolerant Network Routing Protocols”, IEEE Network, March/April 2016. [17] Ahmad El Shoghri, Branislav Kusy, Raja Jurdak, Neil Bergmann, “Augur: A Delay Aware Forwarding Protocol for Delay-Tolerant Networks”, IEEE 11th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), 2015. [18] Michael F. Stewart, Rajgopal Kannan, Amit Dvir and Bhaskar Krishnamachari, “CASPaR: Congestion Avoidance Shortest Path Routing for Delay Tolerant Networks”, International Conference on Computing, Networking and Communications, Wireless Ad Hoc and Sensor Networks, 2016. [19] Wei Sun et al., “On Delay-Tolerant Networking and its Applications”, International Conference on Computer Science and Information Technology (ICCSIT 2011). [20] H. Chenji et al., “A Wireless Sensor, Ad-hoc and delay tolerant network system for disaster response”, Technical Report LENSS-09-02, Sept 2011. [21] Vikram Jeet Singh et al., “Research on Application of Perceived QoS Guarantee through Infrastructure Specific Traffic Parameter Optimization”, I.J. Computer Network and Information Security, Feb 2014. [22] Ziyi Lu and Jianhua Fan, “Delay/Disruption Tolerant Network and its Application in Military Communications”, International Conference on Computer Design And Appliations ICCDA 2010. [23] Ari Kerane et al., “The ONE Simulator for DTN protocol evaluation”, 2nd International Conference on Simulation Tools and Techniques, 2009. [24] E. P. C. Jones, L. Li, P. A. S. Ward, “Practical Routing in Delay-tolerant Networks”, Proc. of ACM SIGCOMM workshop on Delay-tolerant networking, pp. 237-243, Sep. 2006. [25] S. Yamamura, A. Nagata, and M. Tsuru, "Store-carryforward based networking infrastructure: vision and potential," 2011 Third International Conference on Intelligent Networking and Collaborative Systems (INCoS), pp.594-599, 2011. [26] Jian Shen, Sangman Moh , and Ilyong Chung, “Routing Protocols in Delay Tolerant Networks: A Comparative Survey”, The 23rd International Technical Conference on Circuits/Systems, Computers and Communications ,ITCCSCC 2008. [27] Thrasyvoulos Spyropoulos, Konstantinos Psounis, Cauligi S. Raghavendra,” Spray and Wait: An Efficient Routing Scheme for Intermittently Connected Mobile Networks”, Proceedings of the ACM SIGCOMM Workshop on Delay-Tolerant Networking (WDTN’05), August 2005. [28] Anders Lindgren et al. “Probabilistic Routing in Intermittently Connected Networks”, ACM SIGMOBILE Mobile Computing and Communications Review, Volume 7 Issue 3, July 2003. [29] Project page of the ONE simulator, https://www.netlab.tkk.fi/tutkimus/dtn/theone.

I.J. Computer Network and Information Security, 2016, 12, 51-58


58

Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator

Authors’ Profiles Er. Richa Thakur has completed her Bachelor of Technology (B.Tech.) in Information Technology from University Institute of Information Technology, Himachal Pradesh University, India, in 2013 and is currently pursuing her M.Tech in Computer Science from Department of Computer Science, Himachal Pradesh University, Shimla, India.

Dr. Kishori Lal Bansal is a Professor in the Department of Computer Science, Himachal Pradesh University with a teaching experience of more than 20 years. He has completed his Ph.D. in Computer Science and Applications in 2005. He is an active research scholar with more than 50 research papers published in reputed journals.

How to cite this paper: Richa Thakur, K.L. Bansal,"Delay Tolerant Networks: An Analysis of Routing Protocols with ONE Simulator", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.5158, 2016.DOI: 10.5815/ijcnis.2016.12.07

Copyright Š 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 51-58


I. J. Computer Network and Information Security, 2016, 12, 59-66 Published Online December 2016 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2016.12.08

Energy Efficient Clustering Protocol for Sensor Network Prachi The NorthCap University, Gurgaon, 122017, India E-mail: prachiah1985@gmail.com

Shikha Sharma The NorthCap University, Gurgaon, 122017, India E-mail: shikhasharma.mr@gmail.com

Abstract—Energy efficiency is a very crucial issue for battery operated Wireless Sensor Networks (WSNs). Routing plays a major in energy dissipation and it is shown in the literature that Cluster based approach is the most energy effective in any network in comparison to direct or multi hop based approach. Therefore, optimized Clustering became a key point to achieve energy efficiency in Wireless Sensor networks. In this paper, we have designed and implemented a novel protocol in MATLAB in which Cluster Heads are chosen on the basis of energy threshold, minimum average distance from surrounding nodes and farthest distance among Cluster Heads to provide optimal coverage. This paper also compare results of randomly selected CHs and farthest CHs and results demonstrates that farthest chosen CHs provide much better results than randomly selected CHs. To further evaluate performance of our protocol, results of our protocol are compared with LEACH and proposed protocol dominates LEACH in terms of minimizing transmission distance, energy dissipation and hence increasing network lifetime. Apart from this, proposed protocol is based on Poisson distribution because simulation results clearly states that Poisson distribution is very well suited for WSN in comparison to Uniform and Random distribution. Index Terms—Clustering, Sensor Network, Energy Efficient, Routing, network lifetime.

I. INTRODUCTION Wireless Sensor Network (WSN) typically comprises of large number of sensor nodes [1]. These sensor nodes are spatially distributed in random or uniform manner to form a network often operated with dynamic topology. These networks are widely used for applications that require constant monitoring of any environmental phenomena or real-time reporting of any special event and transmit the sensed data to Base Station (BS). The flexibility and cheapness of using WSNs to sense and communicate make them favorable for a wide spectrum of real-time applications. Deployment of sensors is usually done in a random Copyright © 2016 MECS

fashion as they exhibit the capacity of autonomously organizing themselves into a wireless sensor network. In order to increase the utility of sensors, the deployment can also be done in pre-defined location to ensure optimal coverage and connectivity. After deployment, sensors form a dynamic topology of the network thus offering ease of deployment even in inhospitable regions. Due to these advantages of sensors, it is almost inevitable that they are being used heavily in today‟s world. Sensors offer many advantages in terms of their small size, inexpensive nature and ease of usage. However, these advantages come with many other drawbacks such as limited battery lifetime, restricted memory and processing power. Energy consumption is one of the major problem in battery operated WSN. However, majority of energy consumption occurs (around 98%) due to transmission of messages over long distances [2]. Transmission of messages in WSN can be performed either through direct, multi-hop or cluster based routing. Direct communication consumes very high energy when distance between sensor nodes and the Base Station (BS) is large. Multihop based communication increase number of transmissions and receptions between sensor nodes and the BS and hence consumes high energy. Moreover, nodes near the BS forward packet on the behalf of other nodes lying far from the BS quickly deplete their energy and this phenomenon results in cascading effect and further decreases network lifetime. It has been proved in Literature by a number of researchers that cluster based communication effectively utilize network bandwidth, reduce overall transmission distance of nodes and helps in reducing energy consumption and enhancing network lifetime. Clustering partitions all the nodes into groups called clusters in which one of the node is selected as Cluster-Head (CH). Cluster Head performs the tasks of data aggregation after collecting the data from member nodes to minimize the redundancy which helps in conserving energy. Clustering can be classified in two types: Static and Dynamic [3]. In Static Clustering, Clusters that are formed initially doesn‟t change throughout the network lifetime. This type of scenario avoids re-establishment of clusters. However, Static Clustering results in selection of un-optimal nodes as CH. Dynamic Clustering suits best to

I.J. Computer Network and Information Security, 2016, 12, 59-66


60

Energy Efficient Clustering Protocol for Sensor Network

the requirement of homogenous networks where task of Cluster Heads is shifted from one node to another node after certain amount of time. However, a clustering protocol must takes into consideration some important factors before designing the clustering protocol. These parameters include: placement of nodes within a network, selection of Cluster Head and total transmission distance involved in transmission from the nodes to the BS. Placement of nodes within a network is very crucial because denser deployment of nodes increases the redundancy of information in the network and if nodes are sparse in the network then it may left some portion of network uncovered and also increase the transmission distance of messages in the network. Also, data loss in transmissions may affect the information received at the base station due to lack of redundancy in data. Cluster-Head (CH) Selection can be done in a number of ways. Broadly it is the frequency with which a protocol runs CH selection code and the node‟s favorability to be chosen as a CH in the network. Frequency of CH selection can be controlled by putting a threshold on parameters such as the node‟s residual energy. Another factor i.e. node‟s suitability to be chosen as CH can be determined on the basis of many facets like the number of nodes that are present in its neighborhood and the sum of distances to these nodes and so on. Because a CH carries the data on behalf of its member nodes, the distance to these member nodes count as an important parameter in CH selection. Additionally if the node is placed more central in its cluster, it automatically results in lesser intra cluster distance ultimately leading to lesser drainage of batteries associated with nodes. Likewise other factors have their own similar reasons. Total transmission distance involved in the transmissions. It is not only the intra-cluster distance that matters; energy dissipation is also heavy when there is large CH to BS station distances. Often BS is located on the boundaries of network or even farther which causes heavy drainage in energies of CH nodes which lays the foundation for reselection of CHs in the protocols when criteria lies in the threshold of CH‟s residual energy as explained previously. As a result, in this paper we propose dynamic clustering protocol for homogenous WSN by taking into consideration all the above stated parameters. Our protocol reduces the transmission distance significantly and helps in increasing network lifetime. Further, this paper explores three types of network distribution: Poisson distribution, Uniform Distribution and Random Distribution in order to evaluate which suits best for coverage of Wireless sensor Networks. Rest of this paper is organized as follows: Section II contains related work in this area. A novel protocol is proposed in section III. Section IV demonstrates the simulation results and comparisons of proposed protocol with state of art protocol. Conclusion is presented in section V.

Copyright © 2016 MECS

II. RELATED WORK In recent times, a number of clustering protocols have been proposed by various researchers for wireless sensor network with their own advantages and drawbacks. In this section, we discuss a handful of them that particularly focus on decreasing energy dissipation and increasing overall lifetime of the network. Later on, this section describes the requirement of developing a new protocol when lots of clustering protocols for WSN still exist in the literature. Low Energy Adaptive Protocol (LEACH) [4] was the first dynamic clustering protocol proposed for Wireless Sensor networks. Authors used the conventional way of clustering i.e. randomly select some of the nodes as CH. In order to evenly distribute load and balance energy dissipation among nodes, a node decides whether to act as a CH or not on the basis of number of CHs required and number of times that node has been CH so far. However, location of nodes is not taken into account while selecting CHs so there was high possibility that neighboring nodes may elect themselves as CH. Moreover, residual energy of nodes was not taken into consideration. To tackle the problem of randomized CHs and residual energy, LEACH-C [5] was presented where CHs are elected at optimal locations to ensure coverage of entire network. Authors implemented centralized LEACH-C where BS decides which node should act as a CH on the basis of information sent by all nodes regarding their location and energy levels to the BS. However, this increases lots of data transmission in the direction of the BS and congest transmission lines. Weighted Clustering Algorithm (WCA) [6] further tried to break the limitation of LEACH-C by using a weight based approach of clustering. Weights of each node are evaluated on the basis of several parameters such as number of neighbors, average distance to neighbors, mobility factor and cumulative time during which the node has acted as a CH. WCA designates a node as CH on the basis of weights that are assigned to each node, the node having minimum weight is designated as a CH. WCA biggest advantage is flexibility of adjusting the multiplicative factors of these parameters depending upon the requirements of a particular application. Moreover, reselection of CH can be deferred until the energy of the CH is declined to a minimum level but weight assigning process and dynamic clustering added overhead in the algorithm and consume lots of energy. To reduce the overhead of dynamic clustering, EEPSC (Energy Efficient Protocol with Static Clustering) divided the whole network in form of static clusters [7]. In the first round, a temporary CH is chosen randomly, but in subsequent rounds, current CH chooses minimum energy node as temporary CH for next round. The temporary CH collects the data regarding energy levels from all the member nodes. On the basis of this information, it selects the node with maximum energy as CH. An Enhanced version of EEPSC (EEEPSC) came into view after founding that EEPSC often selected nodes located near

I.J. Computer Network and Information Security, 2016, 12, 59-66


Energy Efficient Clustering Protocol for Sensor Network

boundary as CHs. This causes an increase in communication overhead of nodes located on opposite boundary [8]. Enhanced EEPSC considers spatial distribution of nodes and their residual energy to select CHs. Both Energy Efficient Protocol with Static Clustering (EEPSC) and Enhanced EEPSC (EEEPSC) selected temporary CH that adds overhead to both the protocols and hence Front Leading Energy Efficient Cluster Heads (FLEECH) [9] came into picture. FLEECH also divided the network into unequal clusters, starting from BS at the centre and reducing the size of clusters in outward fashion. Like LEACH-C, all nodes forward their residual energy and distance (from the BS) to the BS so that the BS can choose appropriate nodes as CH. This increase lots of traffic in direction of the BS as in LEACH-C. Although, the size of message is shorter than LEACH-C message, still these packets contributed to increase the traffic. Improved energy aware distributed clustering protocol (EADC) was proposed in 2016 to extend the network lifetime in different scenarios. It is clear from the above discussion that some of the clustering protocols suffer from randomized clustering, other suffer from the drawback of huge traffic that ultimately leads to heavy energy dissipation. Taking into consideration all the above points, we implement a new protocol that effectively handles the above stated points and propose an energy efficient clustering solution for WSNs.

III. PROPOSED APPROACH In this section, we propose an energy-efficient clustering approach to maximize lifetime of Wireless Sensor Networks. Network coverage plays a very crucial role in energy dissipation of the network because deployment of nodes affects quality and coverage of sensing thus making it a crucial factor from the perspective of Quality of Service (QoS). In node denser areas, data redundancy will increase. Although, certain amount of redundancy is helpful in reducing the effects of Data loss during transmission but it also results in more energy consumption during aggregation and compression of the data. On the other hand, intra-cluster average distance will increase for sparser nodes and this will ultimately result in increase of energy consumption. Moreover, with sparser nodes, effects of data loss may increase. Node deployment can be done either by deploying sensor nodes at pre-defined locations in the network or by using probability distributions (random, uniform and Poisson) based deployments. Former case is not possible when deployed in remote areas. Therefore, we need to focus on second approach instead. Random distribution of nodes facilitates ease of deployment and scalability. Random distribution is used in unstructured network applications such as battlefield surveillance, etc. In uniform distribution, nodes are uniform distributed Copyright © 2016 MECS

61

over a region. The probability distribution function of uniform distribution is given as [10]:

f x  

1 for A  x  B B A

(1)

where, A is location parameter and B-A is scale parameter. For standard Uniform Distribution, A = 0 and B=1. Poisson distribution is used when number of events occur within a given time slot. In Poisson distribution, both uniformity and randomness are taken into consideration while deployment of nodes [11]. It has been stated in the literature that Poisson distribution offers optimal coverage of network [12-13]. The Probability distribution function when a sensor node exists at point (x,y) is given as[14]:

f  x, y  

N A

(2)

Here, N is the total number of sensor nodes deployed and A is the deployment field. In order to compare these distributions, network parameters are set as follows: Table 1. Simulation Parameters Parameters Deployment Region Transmission Range Number of Nodes Initial Energy Message Size Locate of Base Station

Values 100*100 50 40 to 100 in steps of 20 0.5J 2000 bits (50,50)

All sensors nodes are homogeneous in nature and deployed in same region to analyze performance of various distributions. Additionally, all nodes possess same transmission range and this sensing range is symmetrical and circular in nature. We have simulated the scenarios in MATLAB.

Fig.1. Node Deployment using Random Distribution

I.J. Computer Network and Information Security, 2016, 12, 59-66


62

Energy Efficient Clustering Protocol for Sensor Network

Fig.2. Node Deployment Using Uniform Distribution

Fig.3. Node Deployment Using Poisson Distribution

Simulation results clearly demonstrate the fact that Poisson distribution provides better coverage in comparison of Random and Uniform distribution. As a result, for the implementation we are assuming a network that exhibits the following properties:    

Sensors nodes of the network are homogenous and immobile in nature. Position of Base station is fixed and it is not energy constraint. Nodes sense the environment at fixed rate. Network nodes are deployed using Poisson distribution.

Overall functioning of proposed protocol is shown in the flowchart depicted in figure 4. Copyright © 2016 MECS

Fig.4. Overall Functioning of Proposed Protocol

Proposed protocol is divided into 2 phases named as: A. Cluster Head Selection phase B. Cluster Formation Phase

I.J. Computer Network and Information Security, 2016, 12, 59-66


Energy Efficient Clustering Protocol for Sensor Network

Details of all the phases of proposed protocol are as given below: A.

NCF

63

1 D avg

Cluster Head Selection Phase

In Cluster Head Selection Phase, first step is to determine optimal number of Cluster Heads. This is a very crucial parameter because optimal number of CHs helps us to achieve minimum energy dissipation, load balancing as well as minimum average distance while maintaining network coverage and connectivity. In order to make the balance, Dahnil et. al. in [15] deduced the below mentioned formula:

koptimal 

N  fs R 2 2  mp d

(6)

Where Davg is the average distance of nodes from their CH. According to the above mentioned equation, minimum average distance suits best to our requirement. This has been illustrated in figure 5 & 6.

(3)

Here k is the optimal number of clusters per round, N is the total number of nodes, R is the transmission range, d is the distance of CHs from the BS, εfs and εmp are the energy dependent factors. The possible candidates for CH (say M) are determined on the basis of following parameters: 1.

Node Concentration of a sensor node if elected as CH Current residual energy of network nodes

2.

Fig.5. High Node Centrality

Node Concentration is the number of nodes that are present in transmission range of a node.

N  n nd  Rt 

(4)

N is the Node Concentration of the candidate CH. Distance of neighbor nodes from candidate CH (nd) should be less than equal to node‟s transmission range. Ideal node concentration is given as:

N min imum  5.1774 log n

(5)

According to above equation, if we maintain 5 cluster per 100 nodes, then atleast 20 member nodes are needed in a cluster. In our network, if any node has 20% of network nodes within its transmission range and its minimum energy is above certain threshold then that node became possible candidate for CH. After determining possible candidates for CH, final selection of CH is done on the basis of 2 parameters: 1. 2.

Node Centrality Distance among CHs.

Node Centrality Factor (NCF) evaluates the candidate CH on the basis of how central the node is in its cluster. To evaluate this, average distance of CH to its possible member nodes is found. Relation between Node Centrality and distance is given as: Copyright © 2016 MECS

Fig.6. Low Node Centrality

Apart from this, while finalizing, the distance between possible candidates of CHs is also taken into account. As a result, possible candidates who have farthest distance among themselves and high centrality i.e. minimum average distance from surrounding nodes are chosen as final CHs in our scheme. Subsequent CH selection can be done on the basis of remaining energy. Energy threshold is set as the minimum energy required to carry out the responsibilities of every round or after the selection of a node as cluster head, it will remain at its designation till 50% of its initial energy has been reached.

I.J. Computer Network and Information Security, 2016, 12, 59-66


64

B.

Energy Efficient Clustering Protocol for Sensor Network

Cluster Formation Phase

After finalization of CHs, members of CHs are determined on the basis of 1. 2.

Ideal degree of CH Distance of node from CH

Nodes will determine their possible CH by calculating the distance of each node to each of CHs and assign the node to that CH with which its distance is lowest while maintaining the ideal degree. If a CH has reached its highest ideal degree limit then we look for second closest CH for any particular node. This method is followed for each of network node until all the nodes comes under the territory of one or another CH. Our proposed method differs with LEACH in following ways: Fig.7. Deployment with Poisson, Uniform and Random Distribution

 CH selection is not randomized. Farthest nodes with ideal degree are chosen as CH in case of proposed protocol. This avoid LEACH drawback of choosing nearby nodes as CH.  Apart from this, nodes with minimum average distance are chosen as CH in case of proposed protocol.  Optimal numbers of CHs are chosen using the formula stated in literature.  CH ideal degree is also taken into consideration by proposed protocol for evenly load distribution.

Figure 7 demonstrates the average distance of nodes to their respective CHs in case of different distributions. It is clear from the figure that average distance of nodes from their CH is lesser in case of uniform initially. However, as the number of nodes increases in the network Poisson distribution perform much better than Uniform and Random distribution. Since, Wireless Sensor Networks comprise of large number of sensors due to fragile nature of sensors so Poisson distribution remains the best possible distribution for them.  Distance of nodes from CH

IV. RESULTS & DISCUSSION To evaluate the performance of proposed protocol, we simulated proposed protocol in a network of 100 nodes deployed with Poisson distribution between (x=0,y=0) and (x=100,y=100). BS is placed at (x=50,y=50). Initial energy of nodes is assumed to be 0.5J. Size of message is 2000 bits long. For our simulation, radio based energy dissipation model is used. Free space energy dissipation, εfs = 10 pJ/bit/m2 and amplifier energy, εmp = 0.0013 pJ/bit/m4 for a message of l bits at a distance of d. A.

It is clear from the figure 8 that distance of nodes from their CH is lesser in proposed protocol because it chooses CHs on the basis of minimum average distance from surrounding nodes. Apart from this, placement of CHs is done in a more distributed manner by selecting farthest nodes as CH. Therefore, farthest nodes as CH and minimum average distance from surrounding nodes helps to minimize the distance of nodes from their CH

Analysis of Simulation Results

A set of simulations were conducted to validate our theoretical claims of proposed protocol on the basis of impact of different distribution on the distance, distance of nodes from the CH and by comparing the proposed protocol with LEACH.  Impact of Different Distribution on the distance

Fig.8. Distance of Nodes from CH in Proposed Protocol and Protocol with Random CH

Copyright © 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 59-66


Energy Efficient Clustering Protocol for Sensor Network

Equation 7 also clearly proves the point that distance is a very critical parameter in sensor networks because energy dissipation is directly proportional to square of distance upto the certain distance threshold and as distance reaches above threshold energy dissipation increases in power of 4 of distance. Therefore, even small decrease in transmission distance of messages reduces energy dissipation drastically and increase network lifetime substantially.

one of the major pitfalls with LEACH. This type of deployment also increases the transmission distance of nodes lying far from the elected CHs if there is no CH in their surroundings. Proposed protocol addresses this issue in a very effective manner by selecting CHs with farthest distance so there is high possibility that nodes all over the network can find a CH in their territory.

Et l , d   lEelec  l *  fs * d 2 if d  d o

V. CONCLUSION

Et l , d   lEelec  l *  mp * d if d  d o 4

where,

do 

 fs  mp

and

Er  lEelec

65

(7)

In above equation Et and Er are the energies dissipated in transmitting and receiving a message of l bit over a transmission distance of d. Eelec = 50nJ/bit is the electronics energy, εfs = 10pJ/bit/m2 is the free space energy and εmp is 0.0013pJ/bit/m4 is the energy used by amplifier. Distance threshold is denoted by do.  Comparison of Proposed protocol with LEACH

In this paper, a novel clustering protocol is presented based on inter-cluster and intra-cluster distance. The proposed protocol works upon the solution space of those nodes that have ideal degree of member nodes and 50% or more current energy. Further, final CHs from this solution space is chosen on the basis of based on farthest distance between any two candidate CHs and average distance of surrounding nodes from CH. Selection of such nodes provides better spatial division of clusters. This division supports optimal coverage of nodes with Poisson distribution. Different distribution (Uniform, Random and Poisson) are evaluated and simulation results support Poisson distribution over other two distributions. Proposed protocol is also compared with random CH and LEACH protocol and results show that transmission distance in proposed protocol is much lesser than the other two. Reduction in transmission distance will ultimately help to minimize energy consumption and maximize network lifetime. Further improvement can be done by exploring the solution space created in the methodology. Some heuristically search on location based selection of nodes can be done to bring advancement in clustering technology. REFERENCES [1]

[2]

[3]

[4]

Fig.9. Distance of Nodes from CH in Proposed Protocol and LEACH

[5]

It is clear from the figure 9 that distance is lesser in proposed protocol because CHs in our protocol are more evenly distributed over the network and they are elected on the basis of minimum average distance of surrounding nodes from their CH. Possibility of electing neighboring nodes as CH was Copyright © 2016 MECS

[6]

F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, “Wireless Sensor Networks: A Survey”, Computer Networks: The international Journal of Computer and Telecommunications Networking, Vol. 38, No. 4, pp. 393422, March 2002. G. J. Pottie, W. J. Kaiser, “Wireless Integrated Network Sensors”, Communications of the ACM , Vol. 43, no. 5, pp. 51-58, may 2000, DOI : 10.1145/332833.332838. A. Abbasi, M. Younis, “A survey on Clusteing Algorithmsfor Wireless Sensor Networks”, Computer Communications, Vol. 30, No. 14-15, pp. 2826-2841, Oct 2007. Heinzelman W. B., Chandrakasan A., Balakrishnan H., ‟Energy Efficient Communication Protocol for Wireless Microsensor Networks‟, 33rd IEEE International Conference on System Sciences, ISBN: 0-7695-0493-0, Jan-2000. Heinzelman W. B., Chandrakasan A., Balakrishnan H., „An Application-Specific Protocol Architecture for Wireless Microsensor Networks‟, IEEE Transactions on Wireless Communications, Vol. 1, Issue 4, ISBN: 15361276, pp. 660-670, Oct-2002. Chatterjee M., Das S. K., Turgut D., „WCA: A Weighted Clustering Algorithm for Mobile Ad hoc Networks‟, in Cluster Computing, pp. 193-204, 2002.

I.J. Computer Network and Information Security, 2016, 12, 59-66


66

[7]

[8]

[9]

[10]

[11]

[12]

[13]

Energy Efficient Clustering Protocol for Sensor Network

Amir Sepasi Zahmati, B. Abolhassani, A. A. B. Shirazi, A. S. Bakhtiari, „Energgy Efficient Protocol with Static Clustering‟, World Academy of Science, Engineering and Technology, International Journal of Computer, Electrical, Automation, Control and Information Engineering, Vol. 1, No. 4, 2007. Chaurasiya S. K., Pal T., Bit S. P., “An Enhanced EnergyEfficient Protocol with Static Clustering for WSN”, International Conference on Information Networking (ICOIN), ISBN: 978-1-61284-661-3, pp. 58-63, Jan 2011. Nayak B. K., Mishra M., Rai S. C., Pradhaan S. K., „A Novel Cluster Head Selection Method for Energy Efficient Wireless Sensor Network‟, IEEE International Conference of Information Technology (ICIT), ISBN: 978-1-4799-8083-3, pp. 53-57, Dec-2014. Carle, J. and J. Myoupo, “Topological Properties & Optimal Routing Algorithms for Three Dimensional Hexagonal Networks”, In Proceedings of International Conference on High Performance Computing in the Asia Pacific Region, Beijing, China, pp. 116-121, 2000. Gayatri Devi, Rajeeb Sankar Bal, Sasmita Manjari Nayak, “Node Deployment and Coverage in Wireless Sensor”, International Journal of Innovative Research in Advanced Engineering, Issue 1, Volume 2, January 2015. Wang Y., Li F., Fang F., “Poisson Vs Guassian Distribution for Object Tracking in Wireless Sensor Networks”, 2nd International Workshop on Intelligent Systems and Applications, ISBN: 978-1-4244-5872-1, pp. 1-4, May 2010. Kan Yu, Zhi Li, Qiang Li, Jiguo Yu, A Poisson Distribution Based Topology Control Algorithm for Wireless Sensor Networks Under SINR Model, Wireless Algorithms, Systems, and Applications, pp 706-714, August 2015.

[14] Leon-Garcia, “Probability and Random Processes for Electrical Engineering”, 2nd Edition, Addison-Wesley, July 1993. [15] Dahnil et. al., „Connectivity Aware and Minimum Energy Dissipation Protocol in Wireless Sensor Networks‟, International Journal of Distributed Sensor Networks, Article Id: 153089, pp. 1-8, 2013.

Authors’ Profiles Dr. Prachi is working as Associate Professor in The NorthCap University. She has more than seven years of teaching and research experience. She has completed her Ph.D. in Computer Science from Banasthali University of Rajasthan, India. Her current research interests include wireless sensor network, security in underwater sensor networks and Cyber Security. Prachi received the B.Tech. degree from M.D. University, Rohtak in 2007 and the M.Tech. degree in Computer Science from the Banasthali University at Rajasthan in 2009. She has published 26 papers in referred journals and reputed IEEE and Springer conferences. Ms. Shikha Sharma is pursuing her M.Tech from The NorthCap University, Gurgaon. Her research area is energy efficient clustering in Wireless Sensor Networks. She has published 3 papers in International IEEE Conferences.

How to cite this paper: Prachi, Shikha Sharma,"Energy Efficient Clustering Protocol for Sensor Network", International Journal of Computer Network and Information Security(IJCNIS), Vol.8, No.12, pp.59-66, 2016.DOI: 10.5815/ijcnis.2016.12.08

Copyright © 2016 MECS

I.J. Computer Network and Information Security, 2016, 12, 59-66


Instructions for Authors Manuscript Submission We invite original, previously unpublished, research papers, review, survey and tutorial papers, application papers, plus case studies, short research notes and letters, on both applied and theoretical aspects. Manuscripts should be written in English. All the papers except survey should ideally not exceed 18,000 words (15 pages) in length. Whenever applicable, submissions must include the following elements: title, authors, affiliations, contacts, abstract, index terms, introduction, main text, conclusions, appendixes, acknowledgement, references, and biographies. Papers should be formatted into A4-size (8.27″×11.69″) pages, with main text of 10-point Times New Roman, in single-spaced two-column format. Figures and tables must be sized as they are to appear in print. Figures should be placed exactly where they are to appear within the text. There is no strict requirement on the format of the manuscripts. However, authors are strongly recommended to follow the format of the final version. Papers should be submitted to the MECS Publisher, Unit B 13/F PRAT COMM’L BLDG, 17-19 PRAT AVENUE, TSIMSHATSUI KLN, Hong Kong (Email: ijcnis@mecs-press.org, Paper Submission System: www.mecs-press.org/ijcnis/submission.html), with a cowering email clearly staring the name, address and affiliation of the corresponding author. Paper submissions are accepted only in PDF. Other formats are not acceptable. Each paper will be provided with a unique paper ID for further reference. Authors may suggest 2-4 reviewers when submitting their works, by providing us with the reviewers’ title, full name and contact information. The editor will decide whether the recommendations will be used or not.

Conference Version Submissions previously published in conference proceedings are eligible for consideration provided that the author informs the Editors at the time of submission and that the submission has undergone substantial revision. In the new submission, authors are required to cite the previous publication and very clearly indicate how the new submission offers substantively novel or different contributions beyond those of the previously published work. The appropriate way to indicate that your paper has been revised substantially is for the new paper to have a new title. Author should supply a copy of the previous version to the Editor, and provide a brief description of the differences between the submitted manuscript and the previous version. If the authors provide a previously published conference submission, Editors will cheek the submission to determine whether there has been sufficient new material added to warrant publication in the Journal. The MECS Publisher’s guidelines are that the submission should contain a significant amount of new material, that is, material that has not been published elsewhere. New results are not required; however, the submission should contain expansions of key ideas, examples, and so on, of the conference submission. The paper submitting to the journal should differ from the previously published material by at least 50 percent.

Review Process Submissions are accepted for review with the same work has been neither submitted to, nor published in, another publication. Concurrent submission to other publications will result in immediate rejection of the submission. All manuscripts will be subject to a well established, fair, unbiased peer review and refereeing procedure, and are considered on the basis of their significance, novelty and usefulness to the Journals readership. The reviewing structure will always ensure the anonymity of the referees. The review output will be one of the following decisions: Accept, Accept with minor revision, Accept with major revision, Reject with a possibility of resubmitting, or Reject. The review process may take approximately three months to be completed. Should authors be requested by the editor to revise the text, the revised version should be submitted within three months for a major revision or one month for a minor revision. Authors who need more time are kindly requested to contact the Editor. The Editor reserves the right to reject a paper if it does not meet the aims and scope of the journal, it is not technically sound, it is not revised satisfactorily, or if it is inadequate in presentation.

Revised and Final Version Submission Revised version should follow the same requirements as for the final version to format the paper, plus a short summary about the modifications authors have made and author’s comments. Authors are requested to the MECS Publisher Journal Style for preparing the final camera-ready version. A template in PDF and an MS word template can be downloaded from the web site. Authors are requested to strictly follow the guidelines specified in the templates. Only PDF format is acceptable .The PDF document should be sent as an open file, i.e. without any date protection. Authors should submit their paper electronically through email to the Journal’s submission address. Please always refer to paper ID in the submissions and any further enquiries. Please do not use the Adobe Acrobat PDFWriter to generate the PDF file. Use the Adobe Acrobat Distiller instead, which is contained in the same package as the Acrobat PDFWriter. Make sure that you have used Type 1 or True Type Fonts(cheek with the Acrobat Reader or Acrobat Writer by clicking on File>Document Properties>Fonts to see the list of fonts and their type used in the PDF document).

Copyright Submission of your paper to this journal implies that the paper is not under submission for publication elsewhere. Material which has been previously copyrighted, published, or accepted for publication will not be considered for publication in this journal. Submission of a manuscript is interpreted as a statement of certification that no part of the manuscript is under review by any other formal publication. Submitted papers are assumed to contain no proprietary material unprotected by patent or patent application; responsibility for technical content and for protection of proprietary material rests solely with the author(s) and their organizations and is not the responsibility of the MECS Publisher or its editorial staff. The main author is responsible for ensuring that the article has been seen and approved by all the other authors. It is the responsibility of the author to obtain all necessary copyright release permissions for the use of any copyrighted materials in the manuscript prior to the submission. More information about permission request can be found at the web site. Authors are asked to sign a warranty and copyright agreement upon acceptance of their manuscript, before the manuscript can be published. The Copyright Transfer Agreement can be downloaded from the web site. Publication Charges and Re-print No page charges for publications in this journal. Reprints of the paper can be ordered with a price of 150 USD. Electronic: free available on www.mecs-press.org.To subscribe, please contact the Journal Subscriptions Department, E-mail: ijcnis@mecs-press.org. More information is available on the web site at http://www.mecs-press.org/ijcnis.


IJCNIS-V8-N12  

International Journal of Computer Network and Information Security (IJCNIS) ISSN Print: 2074-9090, ISSN Online: 2074-9104 Volume 8, Number 1...

Read more
Read more
Similar to
Popular now
Just for you