GB2390960A

GB2390960A - Polygraphic encryption optionally with deniable encryption

Info

Publication number: GB2390960A
Application number: GB0216456A
Authority: GB
Inventors: Michael Jefferson Lawrence
Original assignee: Individual
Current assignee: Individual
Priority date: 2002-07-16
Filing date: 2002-07-16
Publication date: 2004-01-21
Also published as: GB0216456D0

Abstract

A polygramic method of encrypting text is disclosed. Characters are encrypted in pairs (or digraphs) using a table containing enciphered characters. If there are N possible clear text characters then an N by N table is required, so that there is a different enciphered character for every possible clear text pair (or digram). The elements of the table may be generated randomly. One clear text character from the pair provides the column index of the cipher character, and other clear text character provides the row index. The second clear text character is positioned an integer number of characters from the first character. The integer is preselected, in one embodiment randomly. Each character is assigned to two different character pairs. The cipher characters use a different font to the clear text characters. Preferably each character of the cipher text font is obtained by cutting and pasting together two characters of the clear text font. Character pairs which appear frequently may be assigned more than one cipher text character to minimise vulnerability to frequency attacks. A deniable encryption system may be used. This involves encrypting cover text, using different encryption parameters to those used for encrypting the clear text, to produce cipher text identical to the cipher text for the clear text. An eavesdropper can be mislead into thinking the covertext is the clear text associated with intercepted cipher text.

Description

in GAP ENCRYPTION

Keyless encryption method, decryption method, cryptographic communication in one direction only and encryption dence with plausible deniability of encryption The facility to 'crack' an encryption scheme has, in the past, been based on each character of a message being the object that is to be encrypted. In this 'gap encryption' method, it is the gaps and the relationships between the characters that is being encrypted, so there is no 'brute force' method available to check for the existence of words or phrases in any decryption attempt based on random substitution of letters, no matter how complex. The characters representing the gaps do not map onto letters, only letter relationships.

The invention relates to the encryption of text or other messages using gaps and without the use of a key. Instead of a key, the message must be displayed in the correct fonts in order to be decrypted. The encrypted message and necessary fonts can be sent by different communication methods, without the need for two-way communication to take place. A second set of fonts can be provided which display a different message to the original, allowing deniable encryption. A further level of protection can be added which includes a computer-specific key for building the fonts necessary for decryption so that the message can only be displayed on a single computer. Encryption and decryption normally use keys. The most common method has a public and a private key which are used together for the two parts of the process. The length of the keys limits how secured the encryption is and the encryption is of the characters that make up the message. The use of brute force to try all possible character permutations will result in message decryption over a time period that is bounded since this process can be automated to look for letter/word/phrase combinations or prime numbers as factors in the keys. The programme that decrypts the message for the recipient is an executable file, and thus must be trusted to perform its task properly. The object of the invention is to simplify the use of encryption and to raise the level of security of encrypted communications. A programme can be written and distributed which manages the complete encryption process automatically, and the result is simple text files which can be received and used without hidden virus or other computer security concerns by the recipient. Instead of a key, the use of display fonts provides for much higher levels of security since instead of the common 128 length number keys, the number of font characters necessary successfully to decrypt a gap encrypted communication can be in excess of 50,000.

This invention does away with the need for a key in the encryption process and has a novel decryption process. It provides such a large number of possible characters, whose attributes are relations between letters in a message, rather than letters themselves, and randomises the gap repeat value, gap frequencies and the resulting font characters so that it is unlikely that a long enough message encrypted by gap

encryption could be cracked, and the process includes the lengthening of messages to reach that size.

Instead of taking as each unit for encryption the letters or characters within a message, the gap between the characters is taken as that unit to be manipulated, together with the relationship of each two sequential characters across the gap. In this way, the number of characters needing to be encrypted is not limited to the number or normal contents of a font or fonts, but is the square of that number. Where one font is used to display a message, there are 224 characters in the font map. When one font is encrypted using this 'gap' method, there will be 224 x 224 = 50, 176 characters required in the new font set if all possible combinations of two characters occur within the message. However, the message to be encrypted is not limited to a single font, so the possible number of characters required may be much larger. To display the message after encryption requires new fonts.

A character in the new font set is composed of the right-half of the lefthand letter and the left half of the right-hand letter around a gap.

As a metaphor, if a section of text is thought of as a wall covered with tiles, normally each tile will contain one letter at its centre. With gap encryption, the letters occur on the joins between the tiles, with the gaps between the letters at the centre of the tiles.

This method could be made more complex by moving the tiles so that each gap is taken both between the lines of text as well as the letters. This would mean that each tile would contain the bottom right-hand quarter of the upper-line left-hand letter in the top left corner, the bottom lefthand quarter of the upper-line right-hand letter in the top right corner, the top right-hand quarter of the lower-line left-hand letter in the bottom left corner and the top left-hand quarter of the lower-line righthand letter in the bottom right corner. For simplicity, this application will continue the description

using the simple two-letter gap method and assuming that the gap is only between sequential characters. A gap repeat value can be set randomly at the start of the process at any value between O and n, where O represents the case for sequential characters and n represents a moderately high number. The gap repeat value sets how many letters are in the gap between the two letters to be encrypted together into one character in the new font. Where the gap repeat value is any value other than zero, a farther step in the decryption process is required to re-map the pixels in the sent font set onto a further font set with a zero gap value. The size of the gap repeat value used to encrypt the message can be sent to the receiver to use in the decryption process separately to any of the other communication.

For fonts, each new font character uses the original font character maps at individual pixel level to make its new font characters, which are a composite of two (or more) parts of the original font characters. For ease of understanding hereafter, the message displayed in old fonts will be called 'plain text' and the message displayed in new fonts 'enciphered text'.

The method of encryption involves building a table of all the two-letter combinations that actually occur within me text [Stage 1] that is to be encrypted. (A "letter" includes any possible character such as a number, "space", "end of line", "end

of paragraph" etc) Since the order of the letters is important, the characters are not commutative. Where a piece of text or message is too small for gap encryption to adequately hide its meaning, the method also adds a sufficiently large volume of random further text to the original plain text [Stage 2]. The addition of garbage to the original message also serves to cover the parallel encryption of the second message, which needs to make sense only as far as its own end. So the original message will make sense until the garbage is reached, and the second message will also make sense until the garbage is reached. It is preferable that the second message is kept reasonable short.

The computer, acting as an encryption device, sets a random gap value and assigns each of the two-letter combinations a character in the new font set using the table [Stage 3]. The plain text is converted into enciphered text as each letter combination is examined and entered into the table [Stage 4]. The frequency of the letter-

combination characters in the new font set is then calculated and a randomly selected rank between 15th and 20'h (or some convenient numbers) most frequently occurring is chosen as the base frequency [Stage 5]. Each of the characters with higher frequency than the base frequency is divided into units of the base frequency and a new character in the new font set is assigned to each new unit [Stage 6]. In this way, the highest frequency two-letter combinations in the original text can be represented by multiple characters in the new font set. For example, the two-letter combination "st" may be represented by Font A, character 24, Font B. character 135, Font C, character 2 etc. The representation in the enciphered text of a high-frequency plain text two-

letter combination which has multiply-occurring new font characters is then randornised [Stage 7]. The specific positions of the two-letter combinations on the axes of the table, although not their relationships, are then randomised so that any regularity in the formation of the table is eliminated [Stage 8]. Then the font characters within the new font set are randomly assigned to each of the two-letter combinations that appear in the table [Stage 9]. The enciphered message is now finalized [Stage 10] . The new font set is created using the old fonts in which the plain text was displayed, cutting and pasting pixels to make the required letter-

combinations, which are named in accordance with the final table [Stage 1 1]. The sender can now send the encrypted message and fonts to the receiver [Stage 12].

Each encryption of a message or piece o f text is individual for that message. So the fonts generated for one message will not be the same for a second message, even the same message sent twice. However, the order in which the component parts of the message appear when encrypted from plain text to enciphered text are not altered by this method.

A second message, which may be publicly displayed if required, is also provided bythe sender as part of the deniable encryption scheme. Once the Stage 9 randomization has been completed, the font characters are rearranged into a second font set so that the second message is what is displayed using the second font set. Any missing or unsuitable lettercombinations are assigned new font characters.

The decryption of the message or text involves only the display of the enciphered text using the appropriate new fonts [Stage 13]. Both the plain text and enciphered text are identical in sequence of components. The message and both the font sets may

be sent from sender to receiver by different methods at different times. Without the correct new fonts, what will appear to a viewer will be the enciphered text displayed using only the fonts that the computer automatically substitutes. Since the enciphered text is a representation of the relationships between the letters, this will be garbage.

Without the fonts, the message is meaningless and the method has achieved cryptographic communication. The use of the second font set allows the receiver and sender to offer a readable version of the enciphered text which is different to the original message.

A further enhancement to the above method involves the use of a computer programme at the sender's end to communicate with the receiver, who is being authorised to read the original text or message, and to use aspects of the receiver's computer as a key within the build specifications of the display of the fonts. An

example might be by generating a long number where the position or size of part of the number has significance in the font-building process. This is separate to the earlier gap encryption process, which decides how the plain text and enciphered text are related. This is a second order encryption where the position of individual pixels is controlled by the key number. If the key number is wrong, then even though the positions of the new font characters is known, what they represent is not discernible.

There are then two possible routes to displaying the message correctly in this enhanced method. One involves embedding aspects of the font-building programme into the original text message (making it an Exe file, for exernple) and using the key number to generate the new fonts, which must then be held in, and accessed from, volatile memory. In this route, the fonts have to be generated each time the message needs to be displayed, and there is no permanent record of the font set. The second route involves making a separate font-building programme, sent separately to the receiver, which builds the correct new fonts based on the number. The first route makes the enciphered text only readable on the specific machine for which the number has been generated. The second route allows the three components (enciphered text, font-building programme and key number) to be copied to another machine and displayed correctly.

By manipulation of the mapping of the cutting and pasting of pixels from the old fonts to the characters in the new fonts, the clarity of the letters that are displayed can be adjusted so that the display itself, or a print out of the displayed text, cannot be read by an optical character reader, although it can be read by a human.

The number of fonts necessary to display a message existing in plain text in only one font is limited to about 225 fonts (224 for combinations plus extra for high frequency multiply-occurring characters). As the size of the message grows, the number of fonts in the font set will not grow as fast. So the size of the enciphered text and required fonts will not be unbounded. In contrast to normal encryption methods, once above a reasonable size, the size of the text to be encrypted does not reduce the difficulty of decryption but increases it as the number of lettercombinations grows towards the maximum around 225 fonts, in this case.

There is some error-robustness in the overall method since any plain or enciphered text that is processed or communicated badly (or any font characters not transmitted

well) would not corrupt all of the message when displayed. Some of the message would be readable unless the number of errors became significant.

Example

The following example shows the numbered stages within the method, although the identity of the new font characters must, precisely because of the difficulty of showing here in the text of this application a 'gap font', be described as two characters em 'cat' in plain text becomes the four new font characters '_/c cia all It_' in enciphered text (where '_' means "space").

Stage 1 - The message The cat sat down.

Stage 2 - It is not long enough, so garbage is added to bring it to the nunimum message length - here just 20 characters for simplicity, although this is not long enough in normal usage.

The cat sat down.xyt

Stage 3 - The gap value is set randomly; for simplicity in this example it is set to zero.

The message is turned into two-letter combinations (the gaps between the combinations are shown here for clarity only) _IT T/h h/e e/_ _/c c/a all tl_ _ls sla all It_ _/d d/o o/w win n/..Ix xly ylt tl_ Stage 4 - A frequency table is compiled of the letter-combinations.

I et/2nd _ a c d e h n o s t w x y T letter _ 1 1 1 1

a 2 c 1 d 1 e 1 h 1 n I o 1 s 1 t 3 w 1 x 1 Y 1

T 1

Stage 5 - The frequency spectrum is ranked in order and the highest frequencies are reduced to units of an appropriate ranking. Here there are only two multiply-occurring two-letter characters. These are t/_ that appears 3 times, and aft that appears twice.

Stage 6 - Multiply-occurring combinations are given extra characters in the new font.

For this example, 2 is taken as the base frequency and t/_ is cut into 2 separate appearances of t/_, one of which will be given different characters in the new fonts.

The frequency table is now 1st/2nd a c d E h n o s t w x Y T. Ietter _ 1 1 1 1

a 2 c 1 d 1 e 1 h 1 n I o 1 s 1 t 2 1 w 1 x 1 Y 1

T 1

The extra occurrence of t/ is renamed %/1 and given an extra character in the table representing the new font set. e table is now I et/2nd _ a c d E h n o s t w x Y T.] leKer 1 1 1 1

a 2 c 1 d 1 e 1 h 1 n I o 1 s 1 t 2 w 1 x 1 Y 1

T I

% 1

Stage 7 - The multiply occurring characters are randomly ordered in the plain text message. Here the extra character for the additional unit of t/_ is renamed %t] and one of the three occurrences of t/_ is randomly substituted for. The message in enciphered text is now _/T T/h kite e/_ _tc c/a alt %11 _Is sla alt tl_ _/d dio o/w wtn n/../x xty y/t tl_ Stage 8 - The axes of the table itself are now randomised.

1 st/2na _ a t d x h n o s c w e y T. ] letter 1 1 1 1

n I c 1 d 1 e 1 h 1 a 2 % 1

s 1 t 2 w 1 x - 1 Y 1 T 1

o 1

Stage 9 - The font characters are assigned randomly. Here FaC I would mean 'Font a, Character 1', but with only one font in the example, since there are only 21 characters needed in the new font, only the identities 'C 1', 'C2' for characters 1 and 2 in the new font will be used.

1 et/2nd _ a t d x h n o s c w e y T.] letter Cl C13 C19 C18 n C3 C C12

d C2 e C'4 h C7 a C10 % C'S

S C9 t cat W C11

X C14

y C16 T C17

C6 O CS

The second (public) message is turned into two-letter combinations and the required characters to build the second message are mapped onto the table using the enciphered text character sequence. Where there are missing or 'wrong' characters, these are composed and given the correct character identities. So the enciphered text is the same for both messages, but the fonts display different messages. If the second message was doe the enciphered text for both messages would read C18 C17 C7 C4Cl9Cl2 C10 CS C13 C9 CIOCI5 Cl C2 C5 Cll C3 C6 C14 C16C15 and the font character C18 would have to map onto the two letter combination _/d, the character C17 would have to map onto d/o, C7 onto o/e and C4 onto e/_. The remainder of the characters are then randomly assigned so that after "doe" is garbage.

Stage 10 - The original message has now been encrypted in the new font as enciphered text and reads C18 C17 C7 C4 Cl9 C12 C10 C8 C13 C9 C10 C15 Cl C2 C5 C1 1 C3 C6 Cl4 C16 C15

This is plain text and will make no sense unless read with the appropriate fonts which contain parts of each letter either side of the gaps that have been encrypted.

Stage 11 - The new font set is created using the old fonts in which the plain text was displayed, cutting and pasting pixels to make the required letter-combinations, which are named in accordance with the final Stage 9 table.

Stage 12 - The message and the fonts are sent by the sender to the receiver.

Stage 13 - Displaying the enciphered text using the new fonts decrypts the message.

The message reads C18 C17 C7 C4 Cl9 C12 C10 C8 C13 C9 ClO C15 Cl C2 C5 Cl l C3 C6 Cl4 C16 C15 The garbage at the end can be ignored on reading.

M Lawrence 14 July 2002

Claims

1 Encryption or decryption of text or other messages without the use of keys by using character relationships across gaps rather than mapping characters onto characters.

2 A method of producing deniable encryption by using multiple font sets, each of which produces different messages.

3 Use under Claim 1 which does not involve two-way communication.

4 Use in Claim I of multiple letter or character relationships, one to another, instead of the letters or characters themselves in encryption or decryption.

5 Use in Claim 1 of individualized fonts for encryption and decryption.

6 Use in Claim 1 of cutting and pasting font character pixels for creation of new hybrid fonts for encryption and decryption.

7 Creation as part of Claim 1 of a computer-specific key for building the fonts necessary for encryption and decryption so that a message can only be displayed on a single computer.

8 Use in Claim 1 of random gap repeat values for encryption and decryption.

9 Use in Clann 1 of flattening of frequency spectrum in gap encryption for multiply-occurring characters.

10 Adding as part of Claim 1 garbage to increase number of character combinations to defeat cracking in gap encryption and to enhance encryption deniability.

11 Under Claim 1, the randomised assignment of font identities to letter" combinations in gap encryption.

12 Claim 1 encryption method where no message is encrypted the same twice.

13 Use in Claim 1 of font display to foil optical character reading of screen or print outs.

14 Use in Claim 1 of a large but bounded number of unknown variables in encryption and decryption without keys, for any extremely long length of message using a small number of fonts for original display.