DE102004006570B4 - One-time key generation method on a fractal basis for block encryption algorithms - Google Patents

Abstract

Once key generation method for generating unique base key sequences for block encryption algorithms based on the interaction of three main features for manipulation of digitally coded data, characterized in that after the preliminary calculation of a sufficiently large byte field by means of fractal Calculation methods and storage in a byte field, via a Multitude of interchangeable selection functions (selection functions) Sequence of bytes from an indexed byte field to a sequence of base keys, being an independent, per data block incremented pseudo-random number generator with a sufficiently large one Period is used to change the current random number value Selection function, as well as derive the index starting values with those the next Basic key off derived from the fractal-computed byte field.

Description

Field of the invention

The Invention belongs into the field of encryption and decryption procedures, in the extended context in the area of the one-time-key-method.

background to the invention

encryption method are used to protect information from access by unauthorized third parties used. Here is the secure transmission of information to attach the same importance as the storage of information on mass media in archive systems of digital information technology. Aim of encryption procedures is it the plaintext information and especially the used ones key to help secretly.

In Block encryption method plain text information is in small pieces of information in so-called data blocks split and encrypted or decrypted in blocks. Symmetric encryption methods assume that the used keys at transmitter and receiver is known. This key must be at least once between transmitter and receiver be exchanged and will usually changed at irregular intervals.

The Plaintext data blocks be with a key encrypted in a fixed number of rounds and thus in a ciphertext transformed. The ciphertext should be after the encryption no statistical dependence have plain text. [3]

This requirement can be met approximately with modern block encryption methods. For this purpose, for example, the known block encryption algorithm AES-Rijndael [1, 2] uses simple data manipulations, such as linking a data block with the key, moving characters within the block structure, exchanging characters within the block structure and replacing individual characters with so-called "S The manipulations are usually performed repeatedly over several rounds with a round key derived from the fixed base key. 1 )

modern Attack methods allow the used in many ways base key spy. Become only symmetric in encryption used, and the used base key is not replaced regularly, so an attacker can capture all ciphertext data previously recorded decipher and evaluate the plain text information.

In Extension to use one-time key method in conjunction with block encryption algorithms ideally for each one to be encrypted Data block has its own base key. The sequence of the base key becomes besides for decryption never reused.

The described below, is a new one-time key generation method is suitable for use in conjunction with all known block encryption algorithms. The one-time key generation method works independently from the block encryption algorithm used, i.e. the core functionality the block encryption algorithm used by the one-time key generation method changed at any point.

description to the invention

The single-key generation method to be protected generates a continuous sequence of basic keys for the symmetric block encryption method employed. For each plaintext data block of an encryption session, the method generates an independent base key, transfers it to the used block encryption algorithm, and encrypts it. The key sequence is unique for each session and depends solely on the initialization parameters of the one-time key generation method. The new one-time key generation method is not a new encryption algorithm, but merely an extension to existing symmetric block encryption algorithms. [7], ( 2 )

The One-time key generation method for symmetrical Block encryption algorithms suitable and can be used on many commercially available microprocessors or Microcontrollers in conjunction with a symmetric block encryption algorithm be implemented.

The One-time key generation method based on fractal calculations, a pseudo-random number generator and a list of selection functions, also referred to below as selection functions.

For each new encryption session, a separate fractal-computed byte field is generated and the pseudorandom number generator is reinitialized. For each data block to be encrypted, a new base key is created using the selection function determined. The initialization is similar in the sender and receiver during session opening and is similar to negotiating a session key.

By the use of fractal calculation methods has the one-time-key method very big Key space, so one Basic key sequence except for decryption never be reused and also very large amounts of data encoded can be without that the calculated one-time key sequence is repeated. An attacker, however, must the entire base key sequence of a session to decrypt the entire ciphertext can. The knowledge of a single base key from the base key sequence allows decryption in each case only one data block, the logged-in ciphertext information. Through the unique sequence of basic keys, combined with one secure encryption algorithm, protects the method against attack methods of cryptanalysis based on a Minimum amount of ciphertext data consistent with a key were edited.

Around To get full access to the plaintext, be with the one-time key method both initial key information as well as information on the method of key generation needed. Both Information can in the new one-time key generation process independently of each other to get managed. This increases the security of the whole system. A single user of the Systems alone has no over the complete Information to the ciphertext with a similar, but not uniquely initialized and maintained system to decrypt.

All Properties ensure that through this mechanism a very long-period sequence of basic keys a one-time key sequence is generated in the individual base key also several times at irregular intervals may occur. Extremely long basic key sequences, at who are not allowed to repeat the sequence of the basic key sequence, can provided by recalculation of the fractal field.

At the location of the fixed session key (base key) occurs from the user's perspective, the knowledge of the initial values for the one-time key generation method. This includes Starting points and jump values in the fractal calculation of the byte field, the start value for the random number generator, data to mask the current random number value as well as the knowledge of all selection functions. This information functionally correspond to the base key when not extended Block encryption method and are strict just like the fixed session key (base key) to keep secret. The initial values must be both in the source system in the encryption as well as in the target system before the start of decryption be present so that both sender and receiver the same, unique Once key sequence produce.

The Imitating the one-time key sequence may only with knowledge of all initial values and calculation methods and never due to statistical analysis of the ciphertext data from an encryption session possible be.

Everyone generated base key from the one-time key sequence comes with the key extension function the used block encryption algorithm extended to the necessary number of round keys.

Of the entire, security-relevant key consists of the start-up and jump values for calculating the fractal field, the starting values of the random number generator, the knowledge of number and content the selection functions and other fixed, however also during maintenance intervals of changeable maskings. The Variety of dependencies and parameterizations allow it, with partially consistent Parameterization special target systems or a group of target systems safety-related from each other by small changes in the masking delineate.

The Security of the one-time-key procedure can be done by distributing the knowledge of key shares to different ones Responsibilities in independent areas of responsibility. This concerns in particular the separation from the responsibility for the production of initial values (session specific key information) from the responsibility for The set of selection functions used (maintainable share the key information).

demarcation to the prior art with one-time key method

As in the patent DE 101 29 285 C2 described very generally, the goal of a one-time encryption method is to extend the usability and increase the security of an encryption system by using a key sequence instead of a fixed key.

In the method described therein also pseudorandom keys (S0 ... Si) and subkeys (TS0 ... TSi), as well as any number of key calculation functions SFi + 1 (S0, ..., Si, D0, ..., Di , TS1, ... TSi + 1). Next to any number of key fields Furthermore, the method allows any number of encryption algorithms (VAi (S0,..., Si, D0,..., Di, TS1,... TSi + 1)).

As described in the specification DE 101 29 285 C2 at point [0002], initial data is needed at the beginning of a session to prepare the one-time key generator. For this purpose, with the known methods such as Diffie-Hellmann or RSA initial data between transmitter and receiver can be exchanged. As a rule, the session key is exchanged.

These Procedures also come for the new one-time key generation method for transmission the initial data between sender and receiver in question. With the initial data Here are start and jump values in the calculation of the fractal field as well as the starting values for set the pseudo-random number generator. The initial data include additionally Information about the intended base key and data block length for the used Block encryption method.

• – Das neue Verfahren ermöglichst eine inhaltlich und zeitlich unabhängige Berechnung der Basisschlüsselfolge. Die Berechnung des Schlüssels kann parallel zur Ver- bzw. Entschlüsselung erfolgen, da es in allen Anteilen unabhängig vom Blockverschlüsselungsalgorithmus lauffähig und unabhängig von Ergebnisdaten der letzten Verschlüsselung ist. Damit eignet sich das Verfahren in besonderer Weise auch zur Implementation auf Multiprozessor- oder Multicore-Systemen. In Abweichung dazu wird im Patent DE 101 29 285 C2 unter Patentanspruch 1.ii bzw. unter [0008] eine Abhängigkeit der Schlüsselberechnung von allen Schlüsseln (S0 – Si), allen Teilschlüsseln (TS1 – TSi+1) und von allen Daten (D0 – Di) beschrieben.
• – In neuen Einmalschlüsselgenerierungsverfahren werden an zentraler Stelle fraktale Berechnungsmethoden eingesetzt, um sichere Schlüssel und einmalige Schlüsselfolgen zu gewährleisten. (neu)
• – Die auch zu Laufzeit gewährleistete Austauschbarkeit von Schlüsselteilen (Selektionsfunktionen) ist ein wesentlicher Bestandteil des Einmalschlüsselgenerierungsverfahrens und läßt sich auch auf modernen Mikrokontrollerarchitekturen umsetzten. Hierdurch wird eine vom Nutzer unabhängige Wartbarkeit des Gesamtschlüssels ermöglicht. (neu)
• – Es werden geringere Datenmengen zwischen dem verschlüsselnden und dem empfangenden System übermittelt, da die Basisschlüssel der Einmalschlüsselsequenz im vorliegenden Einmalschlüsselgenerierungsverfahren nicht zusammen mit den Daten übertragen werden müssen, sondern in Sender und Empfänger gleichermaßen mit Hilfe der Initialdaten berechnet werden. Das Verfahren nach DE 101 29 285 C2 übermittelt zusammen mit jedem verschlüsselten Datensatz auch den verschlüsselten Teilschlüssel.
• – Im neuen Einmalschlüsselgenerierungsverfahren ist es im Gegensatz zum Verfahren aus DE 101 29 285 C2 (Teile [0008] und [0030]) nicht notwendig, weder die gesamte Menge noch die korrekte Reihenfolge der übermittelten, verschlüsselten Daten zu kennen, um den Datenstrom zu entschlüsseln. Fehlende Datenblöcke unterbrechen nicht die Entschlüsselung der nachfolgenden Datenblöcke.
• – Die Variablentypisierung bei der Berechnung des fraktalen Feldes, die Art und Funktion des Pseudozufallszahlengenerators, die Anzahl und Komplexität der Selektionsfunktionen und die Größe des Fraktalen Feldes steuern die Sicherheit des neuen Einmalschlüsselgenerierungsverfahrens. (neu)
• – Das neue Einmalschlüsselgenerierungsverfahrens ist vollständig unabhängig vom eingesetzten Blockverschlüsselungsverfahren und verändert oder beeinträchtigt dieses nicht.
• – Das neue Einmalschlüsselgenerierungsverfahrens kann auch dann eingesetzt werden, wenn nur ein Datenblock zu verschlüsseln ist. (siehe DE 101 29 285 C2 , [0001])

Notwithstanding that in the patent DE 101 29 285 C2 The following new features and important functional deviations can be found in the described one-time key generation procedure:

• - The new method enables a content-independent and temporally independent calculation of the basic key sequence. The calculation of the key can be done in parallel to the encryption or decryption, as it is executable in all shares regardless of the block encryption algorithm and independent of result data of the last encryption. Thus, the method is particularly suitable for implementation on multiprocessor or multicore systems. By way of derogation, the patent DE 101 29 285 C2 in claim 1.ii or under [0008] a dependence of the key calculation of all keys (S0 - Si), all subkeys (TS1 - TSi + 1) and of all data (D0 - Di) described.
• - New single-key generation techniques use fractal computational methods at a central location to ensure secure keys and unique key sequences. (New)
• - The exchangeability of key parts (selection functions) guaranteed at runtime is an essential part of the single-key generation process and can also be implemented on modern microcontroller architectures. This allows a user-independent maintainability of the total key. (New)
• - Lesser amounts of data are transmitted between the encrypting and the receiving system, since the base key of the one-time key sequence in the present one-time key generation method need not be transmitted together with the data, but are calculated in the sender and receiver alike using the initial data. The procedure according to DE 101 29 285 C2 also transmits the encrypted subkey together with each encrypted record.
• In the new one-time key generation method, it is unlike the method DE 101 29 285 C2 (Parts [0008] and [0030]) do not need to know either the total amount or the correct order of the transmitted encrypted data to decrypt the data stream. Missing data blocks do not interrupt the decryption of the subsequent data blocks.
• The variable typing in the calculation of the fractal field, the nature and function of the pseudo-random number generator, the number and complexity of the selection functions and the size of the fractal field control the security of the new one-time-key generation method. (New)
• The new one-time key generation method is completely independent of the block encryption method used and does not change or impair it.
• The new one-time key generation method can also be used if only one data block is to be encrypted. (please refer DE 101 29 285 C2 , [0001]

Full Description of the pictures

1 : Overview of the block encryption procedure AES-Rijndael. In addition to the key extension function for the required 11-15 round keys, the core routines of the Rijndael encryption algorithm are shown.

2 : Embedding the One-Time Key Generation Method into a Block Encryption Method

3 Details of the single-key generation procedure: initialization of pseudo-random number generator and calculation of the fractal field, structure and structure of the fractal field, the set of selection functions and the pseudo-random number generator

4 : Dynamics of the calculation of point sequences in the complex number plane using the example of the mapping z2 = z1 ^ 2 with z = x + i * y and i ^ 2 = -1

Detailed description to claim

a) Initialization of the Once key generation process initialization

The Initial data include for the one-time key generation method the data for the Starting points and jump values for calculating the fractal field in the complex number plane, as well as the starting values for the random number generator. These Data is essentially a session key and be for exchanged every session between sender and receiver again.

An independent pseudo-random number generator is initialized and subsequently incremented in parallel to each new plaintext data block. A selection function is determined with the current value of the pseudo-random number generator. It is used to extract byte sequences from the fractal-computed byte field and to assemble them into a base key for the block encryption algorithm. ( 3 )

The Initialization data can for example, with its own, independent intrinsic key in the Sender encrypted and as a session key transmitted to the recipient become. The sender and receiver authenticate themselves via the Knowledge of the intrinsic key automatically. The initial data is used to calculate the fractal field in the receiver and the pseudo-random number generator set to the starting value. Subsequently can blockwise encryption the plain text data on the transmitter, the subsequent transfer of the ciphertext data to the recipient and blockwise decryption the ciphertext information blocks on the receiver kick off.

About the Initialization data can additionally Consistency information for controlling the transmission quality of the transmission link be transported.

There the initial data from random Numbers / byte sequences would exist in a "Brute Force" attack on the Initial data by trying all possible keys one automated testing on meaningful test samples and thus fail on a correct result of the decryption. An attack on the intrinsic key by trying out all possible ones key (Brute Force) seems impossible.

b) Calculation of the fractal field

The fractal field is based on special, set in the session opening Starting values and the fractal calculation method used (e.g. for the calculation of Mandelbrot and Juiliamengen) and remains as a basis for the Calculation of the base key sequence while received the meeting. The fractal calculation basis was for the single-key generation method selected to make sure that no weak base key possible are.

Fractal calculations in the complex plane have the property of mapping successive, jumpy point sequences that can span the entire complex plane and have parameter-dependent convergence and divergence properties, respectively. Apart from individual attraction points, no point is converted into itself. Successive points of a calculation are therefore unique. [6], ( 4 )

These Properties takes advantage of the one-time key generation method. outgoing From the start value, point sequences are calculated using iterated function systems of degree n> = 2 (e.g. Mandelbrot or Julia quantities) and manipulated such that statistical Abhägigkeiten be largely excluded in the point sequences. Subsequently the data resulting from this calculation will be pointwise taken as database in a memory field. These points correspond digital number representations with at least 4 bytes of data length per Point, where the number of bytes per point with the declaration the variables used in the program code is determined and influence on the Security of the targeted one-time key sequence.

Under fractal calculation methods for almond bread and Juliamengen. Fractal functions are on the one hand continuous, on the other hand but nowhere differentiable [6]. Depending on fixed starting points in the complex Gaussian plane further follow - up points will be in the complex level with different dynamics and convergence or divergence properties. Known graphic representations Mandelbrot and Juliamengen give the dynamics of divergence or Convergence properties in the calculation of point sequences in the complex number level vividly again. The greatest dynamics can be found here in the margins of the Mandelbrot and Juliamengen. The calculated Points in the complex plane are clearly reproducible sequences of value pairs. The starting values for the calculation should be due the special dynamics of the points, if possible in the areas of greatest dynamics (for example in the margins of the Mandelbrot or Juliamen gene). [6]

From the fractal calculation at least 4-byte floating-point numbers are obtained as value pairs and stored in a data field of at least 1024 bytes. In the case of a 4-byte Variable depth and a 1024-byte data field are 128 value pairs in the complex plane to calculate. This is done with a mixture of calculations to Mandelbrot and Juliamengen. After 4 points each, the start value for the calculation of the next 4 points is changed by the jump value from the initialization. Among other things, this method protects against the overflow of variables and reduces the possibility of statistical dependencies in the calculated base keys.

Because of the self-similarity The figures in the complex plane become the absolute values the calculated points are linked together after the calculation. This takes place with the background, the point sequences for the fractal field on to mix and thus an extended independence from the used fractal computational rules to achieve.

• Mandelbrot-Menge: z1 = z0^2 + c mit x0 = y0 = 0 und Variation von c
• Julia-Menge: z1 = z0^2 + c mit c = konstant und Variation von z

For the generation of the fractal field arbitrary complex valued functions of degree> = 2 can be used. Among other things, the calculation methods for Mandelbrot and Julia quantities are suitable, which are to illustrate the method in the following by way of example:

• Mandelbrot set: z1 = z0 ^ 2 + c with x0 = y0 = 0 and variation of c
• Julia set: z1 = z0 ^ 2 + c with c = constant and variation of z

there applies to the complex computational rules z = x + iy with x = real part and y = imaginary part of a Point in the complex Gaussian number plane. One point corresponds a value pair in the calculation of the fractal field. Particularly favorable dynamic Properties arise, for example, through the choice of starting values for the calculation in the border area of the Mandelbrot or Juliamengen. [6]

By Variation of the constant c (Mandelbrot) or by variation of the Starting point z0 (x0 + iy0) (Julia) over a given range the Gauschen complex level, can Score sequences as basis for the calculation of the fractal field can be obtained. There is too every constant c has its own amount of july and thus infinitely many Points. The size Amount of possible Point sequences, as well as the erratic sequence of successive, calculated Points is the reason for the selection of a fractal calculated field as a basis for the single-key generation process.

Especially Divergence, convergence and random properties (bifurcations) increase the suitability of fractal calculations for the generation of sequences of numbers for the fractal field.

The Combination of Mandelbrot and Julia quantities means in the calculation example, that both the constant c as well as the values for x0 and y0 in the course of the calculation of the fractal field can be varied. In the calculation will be Floating-point calculations with at least 4-byte variables carried out. The calculated value pairs are in fractal before storage Interrelated field and mathematically treated according to which causes a further confusion of numerical values. This should Self similarities and repetitive relative convergence properties of the calculated Point sequences in the complex number plane are additionally blurred, so that a field with as possible random, yourself results in non-repetitive values.

aim is the creation of pseudorandom byte sequences from the calculated points of the complex plane as the key base and thus the consistent exclusion of bad, easy too guessing basic key sequences. [3]

Example calculation of a fractal field in detail:

As a guideline for the calculation of the point sequence with the functions for the calculation of fractals, a starting point in the complex plane is defined. In the calculation, the variation of this point is allowed only in an environment of the starting value. However, other specifications or procedures are also possible here. The number of calculated byte values should not be less than 1024. The data for the fractal field is determined in three loops:
The first program loop with 4 passes changes the starting value of the calculation by the defined x-grade value in the real part (x-direction) of the starting point. In the embedded therein second program loop with 4 passes each of the imaginary part (iy direction) of the starting value is changed by the y-jump value. In the innermost, third program loop, four consecutive iterations are each performed with the fractal calculation methods. From two consecutive points of an iteration, each time 4 new values for the fractal field are calculated. The calculated values are each at least 4 bytes long, according to the variable depth used. For each two pairs of values, the real parts and imaginary parts are linked by XOR, in addition, the distance to the origin (radius) is calculated and determined by linking with the calculated radius two more values for the fractal field.

On In the above example, this is calculated from 64 fractals Points by reworking the points in the innermost loop 256 values, and at 4 bytes per point, therefore, 1024 bytes for the field generated.

c) Flock of selection functions

The second base for the one-time-key method represents a list of selection functions that can be used to build byte sequences taken from the fractal calculated byte field and assembled into a base key become. The mathematically arbitrarily structured selection functions grab over Set the index to the contents of the byte field and set a new, sufficiently long basic key together.

Around statistical independence the single key byte when selecting by the same selection function from the byte field additionally to increase, becomes a quasi-random one Offset parameter for the Selection functions by a configurable masking from the derived from the current value of the random number generator.

A additional increase The security of the entire system is achieved by the fact that the crowd of selection functions by one in maintenance intervals at runtime conducted Reprogramming a program segment can be changed. They are thus interchangeable with the disposable key and can also be serviced automatically. Number and complexity of the selection functions affect the security of the one-time-key procedure. The exchange can be independent from personal key parts done and by an independent Administration or by automatisms.

The ultimately valid pseudorandom Base key sequence is taken from the fractal field using selection functions. The number of selection functions must be at least 16. The Number of functions directly affects the security of the one-time key generation method.

controlled The selection of the selection function is characterized by a bit sequence the current value of a pseudo-random generator that is masked of the current random value is obtained. The masking poses one of the serviceable parameterizations. The byte sequence thus obtained is used as a base key for the subsequently called standard key extension methods (For example, the KeyScheduler at AES-Rijndael [2], Figure 1) of the block encryption algorithm used.

The Selection functions are to be set up as pointers to functions and can therefore about the Start address can be called in the program code. Structure and complexity of the selection functions should be similar be constructed and very similar runtime behavior exhibit. Thus an attack by measuring the transit time differences in the selection or during the execution the functions is largely prevented. [4]

Which which is called at least 16 functions controls part of the current random value of the pseudo-random number generator. The selection functions contain arbitrary mathematical instructions with which from the previously calculated fractals field leaps and bounds single bytes are selected. The selected number of bytes corresponds to the given key length to the base key for the inserted Block encryption algorithm.

The Selection functions determine values based on an index over the fractal field. The functions are made by masking the calculation results so limited that the index never calculated the size of the maximum permissible Index of the fractal field exceeds. To further increase the mixing of generated byte sequences, from the random number value additionally an offset value for added the index calculation.

The Selection functions are in a code area that is in maintenance intervals can be exchanged. This makes these functions interchangeable. It is in the implementation of the one-time key generation method to make sure that the Selection of the next Selection function over a direct addressed function call is executed in the program run, so that the runtime of the selection does not affect the function used can be closed.

The Selection functions are a particularly effective option determining key information and key calculation methods from the encrypted Complicate data, so off a collection of known pairs of plaintext and ciphertext data blocks simply on the key generation process deduced can be.

d) Pseudo-random number generator

Of the used pseudo-random number generator should have at least one Period of 2 ^ 60 [5], so that the thus generated random number sequences a sufficient number space include. The sequences of random numbers generated with it are masked and to select a selection function, as well as for derivation index starting values (offset values) for consideration by the selection function used when accessing the fractal field.

in the Frame of session opening will be the new starting values for determines the pseudo-random number generator. To be a flawless decryption to ensure, have to the seeds of the random number generator before the start of a encoding necessarily secured and restored before decryption become. The value of the random number generator will be after each encryption / decryption changed a data block.

Documents considered for the assessment of patentability:
patents DE 101 29 285 C2

Bibliography

• [1] AES Proposal, The Rijndael Block Cipher, Rijndael Documentation, Version 2, September 1999, http://csrc.nist.gov/CryptoToolkit/aes
• [2] J. Daemen, V. Rijmen: The Design of Rijndael, Springer Verlag 2002
• [3] Reinhard Wobst, Adventure Cryptology, Methods, Risks and Benefits of Data Encryption, Addison Wessley, Bonn, 3rd edition 2001
• [4] P. Kocher, J. Jaffe, B. Jun, Introduction to Differential Power Analysis and Related Attacks, Cryptography Research, San Francisco 1998
• [5] Henk C.A. van Tilborg, Fundamentals of Cryptology, Kluver Academic Publishers, Second Printing 2001
• [6] J. Dufner, A. Roser, F. Unseld, fractals and Julia sets, Publisher Harry Deutsch, 1998
• [7] H. Golawski, fractal encryption system, electronics 07/2005, p. 48 ff

Claims (5)

One-time key generation method for generating one-time basic key sequences for block encryption algorithms based on the interaction of three main features for manipulating digitally encoded data, characterized in that after the initial computation of a sufficiently large byte field using fractal computational methods and storage in a byte field, a set of interchangeable selection functions (selection functions ) Byte strings from an indexed byte field are assembled into a sequence of base keys using an independent pseudorandom number generator incremented per frame with a sufficiently large period to derive from the current random number value the selection function and index seeds with which the next base key from the fractal calculated byte field is derived. Method according to claim 1, characterized in that that in a first main feature a) with the help of iterated ones Function systems of degree> = 2 (hereinafter referred to as fractal functions), starting from a predetermined seed, a sequence of points in the Gaussian complex number plane calculated and stored in a data field b) the like calculated points in bytewise manipulation with simple, mathematical procedures are related to each other, so a first Confusion and diffusion of the calculated point sequences over the used calculation method is achieved and in the memory filed list with the calculated points subsequently as an indexed field of bytes (byte matrix, byte field, fractal Field), the byte field being used as the basis for subsequent compilation a sequence of bytes to a base key using a random selection function is used; c) with iterated function systems in the complex Pay level calculates scores for each new session in the sender and receiver dependent on from the initialization values of the one-time key generation method newly created and stored as a byte field. A method according to claim 1 and 2, characterized in that in a second main feature a) the selection functions include mathematical instructions that allow access to the contents of the byte array via the field index, b) one of these selection functions from the list of possible selection functions based on the current Numerical value of a pseudo-random number generator is selected, c) an offset value is taken from the current numerical value of the random number generator, whereby the quasi-random offset value shifts the start value of the index calculation in the byte field by the selection function, d) the next partial key of the time key is obtained by the byte matrix by means of the randomly selected selection function the number of bytes is taken which corresponds to the currently set key length, e) the selection functions are stored in one field of functions and the access to the next, random selected function via pointers to these functions, so that no conclusions about the used selection function in the Krypanalyse can be drawn from the runtime to find the function in the code, f) the selected with the respective selection function byte sequence is assembled to the next base key and using this Basic key the next block of data is encrypted or decrypted via the block encryption method used, g) all grouped in a group selection functions as long as possible processing time wherein the processing time during the selection of a byte sequence from the fractal field may not provide any indication of mathematical details of the selected cryptanalysis selection function, h) the set of selection functions as a whole can be remotely replaced during maintenance, and thus an automated change a key share is possible. Method according to claims 1, 2 and 3, characterized that in a third main feature a) an independent worker Pseudo-random number generator, which is at least over a period of 2 exp 60 (about 1.153 E18) has is used, b) from the generated random numbers by a first masking selects the next selection function the crowd of available Selection functions quasi-random is controlled c) from the generated random numbers by a second masking the quasi-random Selection of the index start value for the index calculations are guaranteed by the selection functions becomes, d) by suitable, dynamic masking of the random number value Group functions can be set up, The method of claim 2, wherein the functions for the calculation of Mandelbrot and Juliamengen.