A Chaotic Encryption System Using PCA Neural Networks

A Chaotic Encryption System Using PCA Neural Networks

A Chaotic Encryption System Using PCA Neural Networks Xiao Fei , Guisong Liu, Bochuan Zheng Computational Intelligence Laboratory, School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 610054, P. R. China Email: [email protected] Abstract—This paper introduces a chaotic encryption system (2) Randomness: If the selected parameters of a chaos is using a principal component analysis (PCA) neural network. The in the chaotic domain, the output sequence generated by the PCA neural network can produce the chaotic behaviors under chaotic system is chaotic and pseudo-random. certain conditions so that it serves as a pseudo-random number generator to generate random private keys. In this encryption (3) Certainty: Chaos is generated by a deterministic sys- system, the one-time pad encryption method is used, which is tem. When the system parameters and the initial values are regarded as the most secure encryption method. The proposed invariant, the output of the random sequence is invariant and system can encrypt any kind of data. The security and high repeatable. performance of encryption are illustrated via some simulations. (4) Extreme sensitivity to initial conditions: Even any minor Index Terms—Principal Component Analysis, Neural Net- works, Chaotic Encryption, Information Security. changes on the initial conditions will have a big impact. The relationships between the basic characteristics of chaos and the common cryptography have been discussed by Shan- I. INTRODUCTION non in his classic article [3], where he proposed two basic With modern technology and network developing, the dig- principles to guide the cryptography design: diffusion and ital information has been applied to many areas in the world. confusion. The chaotic mixing characteristic in an orbit is However, because of its easy replicability, information security corresponding to the traditional characteristics of diffusion in problem has become more and more important. Cryptography encryption system. Chaotic signals with random characteristic is one of basic methodologies for information security. In and the sensitivity to system parameters are related to the recent years, chaos theory consistently plays an active role conventional characteristics of confusion in encryption system. in cryptography [1][2]. The attractiveness of using chaos Therefore, these two characteristics of chaos are able to as the basis for developing cryptosystem is mainly due to guarantee the confusion and diffusion of a chaotic encryption its random behavior and sensitivity to initial conditions and system, like other traditional encryption methods. parameter setting that fulfill the classic Shannon requirements The rest of the paper is organized as follows. Section II of confusion and diffusion [3] [4]. In those classic encryption gives a detailed description of the chaos encryption algorithm. methods, one-time pad is thought as the most secure way in The designing method is also discussed. Security analysis and encryption. In this paper, an encryption algorithm is presented conclusions are given in Secions III and IV, respectively. by using chaos theory to implement one-time pad encryption. II. THE CHAOS ENCRYPTION ALGORITHM AND SYSTEM By this method, each data stream will correspond to a pseudo- DESIGN random private key. Therefore, even by brute-force attacks, it is impossible to decode in a short time. The chaos system introduced in this paper is implemented Chaos is a ruleless and random phenomenon and usually by PCA neural networks algorithm proposed by Xu[7]. In [5], occurs in a deterministic system. This phenomenon is very some chaotic characteristics of the algorithm is presented. The common in the natural life, such as paths of light, changes of PCA algorithm can be described by weather and fluctuations of economic. These disorderly and ω(k +1)=ω(k)+2η[ω(k) − ω3(k)], 0 <η<1, (1) unsystematic phenomena have some amazing features. The most famous case is the Butterfly Effect: while a butterfly fans its wings at Brazil, it may cause a tornado at Texas where η is a parameter and ω(k) is system output. If the of United States. The characteristic of Chaos, especially its initial value is set ω(0) = 0.5, the system output corresponding sensitivity to initial value, is shown in this phenomenon. to η is shown in Fig.1. In this figure, we can see that the system Chaos system has some geometric and statistical characteristics output will eventually converge to one point if parameter η is which deterministic system doesn’t have. Different from other between 0 and 0.5; the system will converge to two points complex phenomena, chaos has the following characteristics: when η is between 0.5 and about 0.65. Obviously, if we (1) Ergodicity: All trajectories of a chaos is confined to a increase η, the system will converge to more and more points, fixed region, a chaotic domain of attraction. The trajectories from 4, 8 to 2n. When η is between 0.8 and 1, the system no transit each output state in this domain. longer comes to convergence and has a chaotic state. 978–1–4244–1674–5/08/$25.00 c 2008 IEEE CIS 2008 Fig. 4. The chaotic encryption system Fig. 1. System chaos phenomena map when ω(0) = 0.5 B. The Encryption Algorithm The description of the encryption algorithm is stated as follows, (1) Ni is the byte array of the original data. The value of Ais: A = Ni(XOR)N2(XOR)N3...(XOR)Ni...(XOR)Nn (2) The value of n equals to the bytes number of the original data. (2) Then computing by Ni = Ni(XOR)A, i =1− n (3) (3) Setting parameter η, the initial ω, and the number of Fig. 2. The Lyapunov’s exponents λ with different η when ω(0) = 0.5 iteration ν to generate random private keys serial Si,i=1−n by the chaotic system. For a chaos dynamical system, the size of Lyapunov’s (4) Mapping the value of Si to a range between 0 and exponents λ with the system is related to the degree of chaos. 255, then encrypting data by calculating the bytes array of the By analyzing Lyapunov’s exponents λ,ifη is selected as encrypted data as follows different values, the system shows different degree of chaos. Fig. 2 shows the Lyapunov’s exponents λ with different η when Di = N1(XOR)Si,i=1− n ω(0) = 0.5. (5) In order to make the encrypted information more adapt- Based on the above analysis, the algorithm is designed as: able with actual requirements, we repeat step 3 to step 4 for ω(k +1)=ω(k)+2η[ω(k) − ω3(k)] M times. The repeat times M is decided by the encryption ω(0) = 0.5,η =0.95, requirements. A larger M will achieve a better chaotic of original information, but it is more time-consuming. where the Lyapunov’s exponent value λ =0.88006. A. The System Design C. The Decryption Algorithm Chaotic systems can generate an unpredictable random Decryption is an inverse operation to encryption. In de- orbit. It can be used as pseudo-random number generator to crypting, there exists four parameters to determine which are generate encryption keys and then make use of these keys needed and dispensable. encrypting. The flowchart of the Key Generator is shown in The decryption algorithm is stated by, N Fig.3. (1) i is the byte array of the encrypted data. We calculate One-time pad method is considered as the most secure the value of A by encryption method. In this method, each data stream in the A = N1(XOR)N2(XOR)N3...(XOR)Ni...(XOR)Nn (4) smallest units is corresponding to a random key. So that the encryption method is quite safe even under exhaustive attack. Thesizeofn is equate to the numbers of bytes of the encrypted It is because that the key space is large enough to decode in data. a short time. (2) Computing: The system has been deployed by C++ Language. The interface of the chaotic encryption system is shown in Fig.4. Ni = Ni(XOR)A, i =1− n (5) Fig. 3. The work flow of the encryption system (3) Input correct system parameter η, the initial ω ,the B. Key Space Analysis ν number of iteration and the repeat time M to make the For a good encrypt algorithm, the key space should be large chaotic system generate the same random private keys serial enough to against the brute-force attack. It should be very S ,i=1− n i as the encryption procedure. sensitive to the private key. If we use slightly different secret S (4) Mapping the value of i to a range between 0 and 255, keys in the proposed algorithm, it will give a quite different then decrypt data by calculating results. In order to test the sensitivity of the keys, we have Di = N1(XOR)Si,i=1− n done the following steps: (1) We still take the standard Lina map (Fig.5(a)) as the D Then, we get the decrypted data. i is the bytes array of the plain-image. encrypted data. (2) Encryption: We encrypt the original data (Fig.5(a)) with In the following experiments, we can see that even the the parameters in Table I. Then we get the encrypted data initial value or system parameters has very little changes, the (Fig.5(b)). decryption will fail. TABLE I III. SECURITY ANALYSIS THE PARAMETERS OF SYSTEM FOR ENCRYPTING A good encryption algorithm should not only have well con- items value description fusing information, but also have certain ability to against all n 2121 the number of iterations ω0 0.5 the initial value kinds of cryptanalytic attacks.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    5 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us