What email address or phone number would you like to use to sign in to Docs.com?
If you already have an account that you use with Office or other Microsoft services, enter it here.
Or sign in with:
Signing in allows you to download and like content, and it provides the authors analytical data about your interactions with their content.
Embed code for: Ch-1
Select a size
The last decade has seen explosive growth in unclassified researches in all aspect of cryptology, and cryptanalysis has been one of the most active areas. Many cryptosystems, which had been thought to be secure, have been broken, and a large set of useful mathematical tools in cryptanalysis has been developed. Security of cryptographic systems is directly related to the difficulty associated with inverting encryption transformation of the system. The protection afforded by the encryption procedure can be evaluated by the uncertainty facing an opponent in determining the permissible key [SEB 89]. The cryptanalysis problem can be described as an identification problem [MAH 00], and the goal of the cryptography is to build cryptographic system that is hard to identify.
In recent years, a growing field of interest in “adaptive systems” has resulted in a variety of adaptive automatons whose characteristics resemble the target system, and in limited ways resemble certain characteristics of living organism and biological adaptive processes.
System identification concerns with inferring models from observation and studying of systems behavior and properties. System identification deals with the problem of building mathematical models of dynamical systems based on observed data from the system. There are two approaches for system identification, depending on the available information, which describe the behavior of the system [LJU 89]. The first approach is; State-Space approach (internal description), which describes the internal state of the system, and used whenever the system dynamical equation is available. The second approach is; Black-Box approach (input-output description) which is used when no information is available about the system except its input and output.
Artificial Neural Networks (ANNs) are statistical simplified models of the real world systems and central nervous system. They are networks of highly interconnected neural computing elements that have ability to respond to input stimuli and their ability to learn adaptively from dynamic environments to establish a generalized solution through approximation of the underlying mapping between input and output. Neural networks have several well-known shortcoming; perhaps the most significant one is the black-box problem (i.e., they capture hidden relations between inputs and outputs without explicitly identifying the nature of the mapping between inputs and outputs or giving reason for the outputs. Neural networks can be regarded as Black-Box that transforms input vector of m-dimensional space to an output vector in n-dimensional space, which make them ideal tools for black-box system identification [PAT 96].
Identification of a system consists of finding a model relationship and determining the system orders and approximation of the unknown function by neural network model called Neuro-Identifier (NID), using a set of input and output data, [MAH 00]. Neuro-Identifiers (NID) are basically a Multi-Layer Feed-Forward artificial neural networks (MLFF) with an input layer (buffer layer), a single or multiple hidden layer with biases, and a linear/or nonlinear output layer. Therefore ANN have been used to construct a neuro-identifier in black-box model to cryptanalysis cipher systems [MAH 00], and this model present successful results; but, the time of training is, at best, very large at most, and the accuracy is towered down when the cipher algorithm is become more complex.
During the past decade, fuzzy systems (FS) have supplemented conventional technologies in many scientific applications and engineering systems. The word fuzzy would put a negative impression on anything it describes. It indicates thing that are ambiguous, imprecise, or vague [SANT 93]. Fuzzy Logic (FL) is a method of easily representing analog processes on a digital computer to solve problems that deal with ambiguous data using a mathematical logic to represent a crisp logic system.
Fuzzy Neuro (FN) means the employment of fuzzy system (membership function and fuzzy set) in the generation of the input/output set to the heuristic learning strategies of ANNs. For instance an ANN can be fuzzified in such a way that it learns the mapping between input-output fuzzy sets. Furthermore, fuzzy logic can be used to determine the learning step of ANNs according to the state of convergence. By incorporating fuzzy principles into ANNs, more user flexibility is attained and the resultant network or system becomes more robust [CHI 96].
1.2 LITERATURE SURVEY:
For many years the concept fuzzy-neuro used at most for classification mode, but in this research we present the fuzzy-neuro in identification mode. And precisely, as Black-Box cryptanalysis model and we try in this research to give the ANNs and fuzzy logic (FL) a meaning in the context of fuzzy identification system.
The theses or the researches that concerned with this thesis are very little. The only available research is “ Black-Box Attack Using Neuro-Identifier” which is a Ph.D. thesis submitted to the Department of computer Science and Information Systems to the University of Technology by Mahmood Khalel Al-Ubaidy,2000. The other researches that discussed one or more concept such as:
Fuzzy Rule Extraction : a Ph.D. thesis submitted to the Department of computer Science and Information Systems to the University of Technology by Ban Nadeem Al-Kallak,2001. This thesis presented the concept of fuzzy-neuro classification system.
Hybridizing of Intelligent Systems: a Msc. thesis submitted to the Department of computer Science to the University of Baghdad by Hayder Shihab Ahmed, 1998. Main subject of fuzzy logic and ANNs and these connected is submitted with the genetic algorithm.
1.3 TOOLS USED IN THE RESEARCH:
The Fuzzy-Neuro-Identifier (FNID) used in this research is constructed, trained, and simulated in the MATLAB V188.8.131.521 (by Math Work Inc. 2000), and conducted on the following personal computer configuration:
PII 350 MHz & FULL CASH RAM
128 MB RAM
10.1 GB HD
The following programs, that represent the main work, have been written in Matlab, The total number of programs is (18) program:
Datcsr, Dataffin and Datvgnr: prepare crisp data to the system.
Fuzdatcsr, Fuzdataffin and Fuzdatvgnr : prepare fuzzy data to the system.
Fuzatcsr, Fuzataffin and Fuzatvgnr : represent identification training programs.
Fuzemcsr, Fuzemaffin and Fuzemvgnr : represent emulation training programs.
Simfuzatcsr, Simfuzataffin, Simfuzatvgnr, Simfuzemcsr, Simfuzemaffin and Simfuzemvgnr : represent simulation programs.
1.4 AIM OF THE RESEARCH:
The objectives of this research are to present a combination of Fuzzy Logic (FL) and Neuro-Identifier (NID) to produce Fuzzy-Neuro-Identifier (FNID). And use this combination as a tool to attack cipher systems by presenting the cryptanalysis problem of unknown cipher system as an identification problem depending on the fuzzification of observable input and output data of the system, and use the resulting system to achieve two goals; the first goal is to determine the encryption key, while the second goal is to construct an equivalent model for the unknown cipher system. The main aim of the proposed fuzzy-neuro system used in this research is to enhance the performance of traditional Black-Box using neuro-identifier (NID) by increasing the accuracy and decreasing time and number of epochs.
1.5 THESIS OVERVIEW:
This thesis consist of six chapters, the first chapter presented the introduction of all thesis, and other chapters contents:
Chapter two (Fuzzy Neural Networks):
This chapter consists of three parts; part one gives an overview of ANNs, models of artificial neuron, arithmetic operation, activation function, and the learning process. Part two gives an overview of FL, its features, some of its classification, membership functions, fuzzy sets, and linguistic variable. Part three present the fusion way of ANNs and FL, integrating system, and the models of FN systems.
Chapter three (System Identification):
This chapter present fundamental concepts and theoretical foundation of basic mathematical models for system identification. State-Space models presented briefly; while Black-Box model is discussed with more detail, as it is the chosen approach to stick with throughout the research.
Adaptive theory is also presented to complete the discussion of Black-Box identification using Neuro-Identifier (NID). This Neuro-Identifier (NID) is discussed too. Finally a proposed Fuzzy-Neuro-Identifier (FNID) is presented.
Chapter four (Cipher Systems Cryptoanalysis):
This chapter deals with three parts, the first part presents a main concept of cryptographic system, the general type of cipher systems and explain its classification. The second part presents a concept of cryptanalysis systems and its types for each cipher system. Finally, the third part presents the cryptanalysis of cipher systems using Neuro-Identifier (NID) (Black-Box attack).
Chapter five (Black-Box Attack Using Fuzzy-Neuro-Identifier (FNID)):
In this chapter the experimental work have been presented with two modes which describes the objectives of the attack. The first mode is the cryptanalysis and the second is the emulation mode. Two types of data, which are the crisp data (input and output data) prepared to the network and the fuzzy data used in the process, have been presented and discussed for each mode.
After that, the description of the training and testing of each mode were stated taking in consideration the training and testing data for each mode.
Finally, the final results of the network were stated and discussed. Then, the simulation of the Fuzzy-Neuro-Identifier (FNID) of each mode was made using the save weight (W) and biases (B). Furthermore, the proposed network structures were drawn with the experimental output figures of all cases.
Chapter six (Conclusion and Suggestion of Future Work):
The general conclusions obtained from this work were presented in this chapter with some recommendations for future work.
Chapter one Introduction
re crisp data to the system.
This chapter present fundamental concepts and theoretical found