Modern cryptosystems have a narrow application area, mainly because of a strict restriction of overrunning during the encryption of so-called unicity distance for the key. It is possible to essentially increase the unicity distance by using a synthesis of a constructed/artificial/planned language in the application area, where an alphabet of a high dimension is used. There was made a synthesis of highly secured cryptosystem. It has such feature as the security of text information with an increased unicity distance. This distance is set into a tabular form which assumes that this text information is taken from the thesaurus of predefined application area. There were gotten mathematical expressions that display the procedure of the encryption and decryption. There were considered the features of construction of the generator of pseudorandom sequences. Also, there was defined the level of security of cryptosystem with a different levels of resource capabilities. There was given a graph of correlation between the indicator of security and the code length of the encryption key. There were also defined the conditions of providing a necessary security regime in cloud systems.
As the procedure of text messages encrypting is directly connected with a usage of the
alphabet symbols of any natural language, it is well known [
Z H1(K) ,
H (K) (1) (2) (3) where n1 stands for the length of a message, which is written by a natural language (for example, English, Ukrainian, Russian etc.), that contains an alphabet В1 . While the entropy of the key, which is used for the security of information when using the integrated alphabet of the display language for this exact information, is calculated as
n
H (K ) log2 iÏ1B (s 1)
where B stands for the alphabet of a tabular form, i.e. the amount of possible
combinations of table rows of a semantic dictionary, which is created artificially as a
result of statistical and semantic analysis of the subject area, n is the number of rows
in tabular form, and s stands for the number of columns in tabular form.
The paper [
H1(K ) log2 (B1n1 ) ,
EUD can be implemented for different types of substitution ciphers and is mainly
used in polyalphabetic ciphers, such as Vigenere cipher and Beaufort cipher [
This number is determined by using exclusive disjunction of a number of the true element and the pseudorandom number derived from the PRNG. The result of the usage of an encryption device is the ciphered text of the table, which is transmitted to the receiving side via an open communication channel.
Fig.1. Variant of the flowchart of EUD construction The application system administrator submits the received encrypted text to the decryption device on the receiving side. Also, he sets the same initial state of the PRNG as the initial state of the transmitter's generator using the known key. The decryptor does the same transformations as the encryptor. The result of the decryption device is the output table that was formed on the transmitting side.
If the characteristics of the synthesized EUD are defined as it is shown in [
As see, for the increase of distance of unicity the dictionary of language of reflection
of information is plugged in composition of this system, linguistic units from that are
placed in the set table form [
In order to explain the mechanism of replacing the true element of the open text sequence with the masking element of the closed sequence in the encryption procedure, we have to consider the following example. Let’s suppose that the dictionary of the subject area consists of n subdictionaries. All elements of every subdictionary are numbered. Let’s also assume that the location of the first true element of the introductory sequence in the structure of the defined by the k-th subdictionary by encoder is of the serial number Ninp, and the first random number given by the generator of the PRNG is equal to Ninp, where 0 Nran p к , p к stands for the dimension (i.e., the total number of words) of the k-th subdictionary. Then the ordinal location number in the structure of the k-th subdictionary of "not true" (i.e., masking) linguistic element, which is set as the first element in the output encrypted sequence instead of the true first linguistic element, will be equal to
Nki sec mod pk (Nki inp mod pk (Nki ran ) pk ) . (4) The expression (1) is fair only if provided by the coincidence of the dimension of the applied subdictionary with the scope of random sequences generated by the PRNG. (Scope means the difference between the maximum and minimum pseudorandom numbers in the sequence). In general, when the size of the PRNG is equal to the dimension of the largest subdictionary, the encryption equation, which in fact is a formula for determining the sequence number of the masking word placement in the thesaurus subdictionary, has the following form:
Nsec mod pk (Ninp Nran ) , (5) where Ninp stands for the serial number of the location in the k-th subdictionary of the i-th true linguistic element taken from the source that requires to be encrypted, the open sequence of linguistic units, where i is the sequence number of the location of this element in an open sequence; Nsec stands for a serial number in the k-th subdictionary of the masking element, which is placed into the i-th position (instead of the true element) of the output encrypted sequence of linguistic units; Nran is a pseudorandom integer generated by a PRNG on the i-th step of the generation in order to encrypt the i-th true element with a serial number Ninp; pк is the dimension of the kth subdictionary; k is the sequence number of the subdictionary in the dictionary of the language of the applied region.
In this case the equation of decryption of information is a formula for determining the sequence number of the location in the selected subdictionary of the i-th true linguistic unit placed in the output of the decrypted sequence instead of the masking linguistic unit taken from the original encrypted sequence. The decryption equation has the following form:
Ninp mod pk (Nsec Nran ) , (6) where Ninp stands for the serial number of the linguistic unit (according to the numeration in the k-th subdictionary), which is identical to the true unit reflected in the original open sequence of linguistic units; Nsec is a serial number of the encrypted linguistic unit (according to the numbering in the k-th subdictionary) taken from the encrypted sequence that is provided for processing, at the current step of the decryption procedure; Nran is a pseudorandom integer generated by the PRNG to decrypt the encrypted unit with the serial number Nsec; pк is the dimension (i.e., the total number of words) of the k-th subdictionary; k is the serial number of the subdictionary which is selected at the current step of the decryption procedure. Formula (6) is valid if the width of the PRNG is equal to the dimension of the subdictionary, which has the largest number of linguistic elements among the plurality of all subdictionaries of the used domain vocabulary.
It should be noted that PRNGs that are used in the information security systems have to meet the following requirements: high cryptofirmness; good statistical properties, generated pseudorandom sequences, which according to their statistical properties, should not differ from truly random sequences in the uniquely determined conditions; a big period of generated random sequence; efficient hardware and software implementations.
Fig. 2 shows the scheme of work of the PRNG in the EUD.
The output of PRNG First pseudorandom sequence
Second pseudorandom
sequence
i-th pseudorandom
sequence
Т 2Т (і -1)Т іТ
Fig. 2. The work of the PRNG in the EUD without providing of a random way of substitution
of symbols
t
This picture also shows that knowing the size of the period of random sequence T, the analyst
can make a filter and “catch” symbols with a period T. And if you apply the encryption
procedure, where each next table is encrypted from the first value of the pseudorandom
sequence of numbers, then a random character of symbols substitution is not provided.
Fig. 3 shows how the PRNG provides the random way of substitution by encrypting
tables with text information with a help of the EDU. In Fig. 3 there is evidence that
each next table is encrypted by a sequence of pseudorandom numbers without
resetting the PRNG to its initial state. The amount of pseudorandom numbers used
during encryption is getting fixed after every encryption session. Also, every next
table is encrypted not from the initial state of the PRNG, but from the point of the
sequence where it was fixed on the previous encryption act. Therefore, the random
nature of the substitution is provided and, consequently, the attack type like the
probabilistic analysis of the repetition rate of the characters in the source text is
neutralized [
The output of PRNG t
T thoefe1nsctrtyapbtlieon thoefe2nncdrytpabtiloen …... thoefejn-sctrytapbtiloen Fig. 3. The operation of PRNG in the EUD with the provision of random substitutions Firmness of EUD 1. Analysis of EUD security, provided that at least one sample of the output (open) text message and a sample of the correspondence with the cipher gram are known. Initial conditions and resources of the malefactor: 1) the analyst knows at least one sample of the original open table and correspondent table encrypted in/with it; 2) the participants of the secret information exchange do not change the key of the cipher during the time when the attack occurs; 3) the analyst has the means of encryption and decryption of the EUD . He knows the structure of the key of the cipher, the structure of the given table form and the dictionary of the applied area, the structure of which is coordinated with the structure of the table form; 4) analysis of the security of the algorithm of cipher transitions on the basis of intercepted cipher grams was not conducted or did not have a positive result.
The components of the equipment that are used by the analyst to carry out an attack on the cipher under the above conditions are shown in Fig. 1. Analytics actions. An analyst decrypts the known encrypted sample of a known output table. Then he makes attempts to determine the key of the cipher by directly scanning all possible values of this key until an appropriate open output message is received. Indicator of firmness: K1 is the maximum possible number of search for the cipher key (which is equal to the number of possible values of the cipher key).
K1 xk , (7) where x is the basis of the language alphabet of the cipher key, k is the code length of the cipher key.
Conclusion regarding the above attack model: the analyst has the opportunity to determine the fact of successfully completing the attack by comparing the decrypted sample of the known cipher gram with its open source output message; in this case the observance or non-observance of restrictions on the unicity distance does not affect the security of the EUD, and the security of the coincides with the security of the algorithm of cipher transformations that were used in accordance with the scheme shown in Fig. 1. 2. Analysis of EUD security in the absence of correspondent samples of output and encrypted text messages and non-compliance with restrictions on the unicity distance Initial conditions and resources of the malefactor: a) an analyst is provided with a sufficient volume of intercepted ciphertext (which was obtained at intervals when the key of the cipher was not changed) to make reasonable statistical conclusions about the probability of the appearance of its individual elements; b) the absence of any corresponded pair of samples of the source and encrypted information, i.e. the analysis of firmness can only be performed with the help of intercepted ciphertext; c) an analyst has the ability to determine that the condition of observing the distance of unity is not fulfilled, and, therefore, the attack on the cipher does not lose its meaning; d) the scheme of the generator of pseudorandom sequences (PRNG) provides the random way of substitution; e) the possibility to get information about the statistical properties of the dictionary of the application area, which is consistent with the structure of the given table form; f) the analyst has the PRNG encryption and decryption means, as well as the dictionary of the applied area. He also knows the structure of the cipher key and the structure of the tabular form.
The flowchart of equipment used to carry out an attack on the code in the above conditions is similar to that shown in Fig. 1, but the amount of work that the analyst has to perform is much wider.
Analytics actions Preparatory stage. 1) Pre-receiving of information about the statistical properties of secret data exchange, which consists of: a) obtaining a statistically complete sample of cipher grams on a problem within the given thesaurus of the applied area; b) processing of the sample in order to obtain an ideal statistical function of the distribution of semantic units, which make up the information of secret exchange to solve a problem.
Attack stage. 1) decryption of intercepted samples of encrypted text by the direct search of all values of the cipher keys until all samples of the distribution function of the probability of appearance of speech elements at the output of the decoder are obtained. 2) received samples of the distribution function are compared to the function that was gotten at the preparatory stage. The next step is making the decision about the most likely variant of the key. Firmness indicator is statistical: K2 K1 V , (8) Where K1 stands for the index of algorithm security, V is the size of the statistical information of a secret exchange.
Conclusion regarding the above attack model: The firmness of the EUD is V times higher than the firmness of the used cipher transformation method and is statistical unit of measurement. Let’s consider the graph of the dependence of security the parameter K on the key of the cipher k to the two above-mentioned attack models (Fig. 4). Firstly, it leads to a trivial result: with the length of the cipher key always rising, the stability of the security system is increasing. Secondly, in any case, the stability of the EUD for attacks of type № 1 is much lower than the security of this system comparing to type of attacks №2. Thirdly, the sustainability of the EUD is increasing with the increase in the sample size (which is marked with a symbol V ). Let’s consider the dependence of the stability parameter K on the parameter a, where a is the basis of the alphabet of the key language of information. According to the graph on the Fig. 4, we can see that the increasing of the value of parameter a leads to the increasing of the index of stability K with a constant (unchanged) value of the code length of a cipher key k. For example, if we consider the group of curves in relation to the attack model №1, then we can see that with the bit rate k = 8, the firmness index is equal to the following values: K = 256 for a = 2, К = 1·108 for а = 10, К = 1,4064·1012 for a = 128, К = 1,8447·1019 for а = 256. Therefore, it is recommended to increase the basis of the alphabet of the language of the key information.
Initial conditions and resource possibilities of the malefactor (that case when the EUD parameters provide conditions for maintenance of a perfectly secret system): 1) the analyst has the means of encryption / decryption of the EUD , including dictionary of the applied area, but there is no possibility of intercepting the amount of cipher grams that exceed/overrun the unity distance, because the security system satisfies the conditions of quite proof; 2) the absence of any a priori information about the original text information and encrypted text which is correspondent with this information; 3) there is no need to obtain any priori information about the statistical properties of confidential data (because the conditions for maintaining a perfectly secret system are provided).
The attack model (if the EUD`s parameters match the conditions of a perfectly secret
system) - does not need to be defined. Stability indicator is not defined, since there are
no conditions for determining the moment of successful completion/ending of the
attack. In this case, the distance of unity U is used as an efficiency indicator of the EUD.
The criterion for the correct functioning of the constructed EUD is the observance of
the unicity distance (the condition that the amount of cipher encrypted with the same
key will not exceed the uniсity distance), which is defined as [
U log2 (K ) , (9)
D
where U stands for the unicity distance, K is the maximum number of ciphers
variations, D is the redundancy of accepted language for displaying messages.
To fulfill the condition of observing the distance of unity, it is necessary to correctly
determine the length of the key k in the correlation with the length of the message n,
considering that [
k log2 N ,
(10)
where k stands for the code length of the key, N is the number of possible
values/meanings of the message with the length n. So the dependence of the entropy
(the length of it in bits) of the cipher key on the length of the message can be written
as (3) (the output of the formula was given in [
Conclusion regarding the above attack model: EUD has the properties of an absolutely secret system if all the above written conditions were performed.
Conclusions
1. In this paper a cryptosystem was synthesized that executes the offered method [