Chapter 3: Classical Encryption Techniques PDF

84 CHAPTER 2 / INTRODUCTION TO NUMBER THEORY The congruence relation is used to define residue classes. Those numbers that have the same remainder r when divided by m form a residue class (mod m). There are m residue classes (mod m). For a given remainder r, the residue class to which it belongs c...

84 CHAPTER 2 / INTRODUCTION TO NUMBER THEORY The congruence relation is used to define residue classes. Those numbers that have the same remainder r when divided by m form a residue class (mod m). There are m residue classes (mod m). For a given remainder r, the residue class to which it belongs consists of the numbers r, r { m, r { 2m, c According to our definition, the congruence a K b (mod m) signifies that the numbers a and b differ by a multiple of m. Consequently, the congruence can also be expressed in the terms that a and b belong to the same residue class (mod m). PART TWO: SYMMETRIC CIPHERS CHAPTER Classical Encryption Techniques 3.1 Symmetric Cipher Model Cryptography Cryptanalysis and Brute-Force Attack 3.2 Substitution Techniques Caesar Cipher Monoalphabetic Ciphers Playfair Cipher Hill Cipher Polyalphabetic Ciphers One-Time Pad 3.3 Transposition Techniques 3.4 Rotor Machines 3.5 Steganography 3.6 Key Terms, Review Questions, and Problems 85 86 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES LEARNING OBJECTIVES After studying this chapter, you should be able to: ◆ ◆ ◆ ◆ ◆ ◆ Present an overview of the main concepts of symmetric cryptography. Explain the difference between cryptanalysis and brute-force attack. Understand the operation of a monoalphabetic substitution cipher. Understand the operation of a polyalphabetic cipher. Present an overview of the Hill cipher. Describe the operation of a rotor machine. Symmetric encryption, also referred to as conventional encryption or single-key encryption, was the only type of encryption in use prior to the development of publickey encryption in the 1970s. It remains by far the most widely used of the two types of encryption. Part One examines a number of symmetric ciphers. In this chapter, we begin with a look at a general model for the symmetric encryption process; this will enable us to understand the context within which the algorithms are used. Next, we examine a variety of algorithms in use before the computer era. Finally, we look briefly at a different approach known as steganography. Chapters 4 and 6 introduce the two most widely used symmetric cipher: DES and AES. Before beginning, we define some terms. An original message is known as the plaintext, while the coded message is called the ciphertext. The process of converting from plaintext to ciphertext is known as enciphering or encryption; restoring the plaintext from the ciphertext is deciphering or decryption. The many schemes used for encryption constitute the area of study known as cryptography. Such a scheme is known as a cryptographic system or a cipher. Techniques used for deciphering a message without any knowledge of the enciphering details fall into the area of cryptanalysis. Cryptanalysis is what the layperson calls “breaking the code.” The areas of cryptography and cryptanalysis together are called cryptology. 3.1 SYMMETRIC CIPHER MODEL A symmetric encryption scheme has five ingredients (Figure 3.1): ■ ■ ■ Plaintext: This is the original intelligible message or data that is fed into the algorithm as input. Encryption algorithm: The encryption algorithm performs various substitutions and transformations on the plaintext. Secret key: The secret key is also input to the encryption algorithm. The key is a value independent of the plaintext and of the algorithm. The algorithm will produce a different output depending on the specific key being used at the time. The exact substitutions and transformations performed by the algorithm depend on the key. 3.1 / SYMMETRIC CIPHER MODEL Secret key shared by sender and recipient Secret key shared by sender and recipient K K Transmitted ciphertext X Y = E(K, X ) Plaintext input Figure 3.1 ■ ■ 87 Encryption algorithm (e.g., AES) X = D(K, Y ) Decryption algorithm (reverse of encryption algorithm) Plaintext output Simplified Model of Symmetric Encryption Ciphertext: This is the scrambled message produced as output. It depends on the plaintext and the secret key. For a given message, two different keys will produce two different ciphertexts. The ciphertext is an apparently random stream of data and, as it stands, is unintelligible. Decryption algorithm: This is essentially the encryption algorithm run in reverse. It takes the ciphertext and the secret key and produces the original plaintext. There are two requirements for secure use of conventional encryption: 1. We need a strong encryption algorithm. At a minimum, we would like the algorithm to be such that an opponent who knows the algorithm and has access to one or more ciphertexts would be unable to decipher the ciphertext or figure out the key. This requirement is usually stated in a stronger form: The opponent should be unable to decrypt ciphertext or discover the key even if he or she is in possession of a number of ciphertexts together with the plaintext that produced each ciphertext. 2. Sender and receiver must have obtained copies of the secret key in a secure fashion and must keep the key secure. If someone can discover the key and knows the algorithm, all communication using this key is readable. We assume that it is impractical to decrypt a message on the basis of the ciphertext plus knowledge of the encryption/decryption algorithm. In other words, we do not need to keep the algorithm secret; we need to keep only the key secret. This feature of symmetric encryption is what makes it feasible for widespread use. The fact that the algorithm need not be kept secret means that manufacturers can and have developed low-cost chip implementations of data encryption algorithms. These chips are widely available and incorporated into a number of products. With the use of symmetric encryption, the principal security problem is maintaining the secrecy of the key. Let us take a closer look at the essential elements of a symmetric encryption scheme, using Figure 3.2. A source produces a message in plaintext, X = [X1, X2, c , XM]. The M elements of X are letters in some finite alphabet. Traditionally, the alphabet usually consisted of the 26 capital letters. Nowadays, 88 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES X Cryptanalyst K Message source X Encryption algorithm Y = E(K, X) Decryption algorithm X Destination K Secure channel Key source Figure 3.2 Model of Symmetric Cryptosystem the binary alphabet {0, 1} is typically used. For encryption, a key of the form K = [K1, K2, c , KJ] is generated. If the key is generated at the message source, then it must also be provided to the destination by means of some secure channel. Alternatively, a third party could generate the key and securely deliver it to both source and destination. With the message X and the encryption key K as input, the encryption algorithm forms the ciphertext Y = [Y1, Y2, c , YN]. We can write this as Y = E(K, X) This notation indicates that Y is produced by using encryption algorithm E as a function of the plaintext X, with the specific function determined by the value of the key K. The intended receiver, in possession of the key, is able to invert the transformation: X = D(K, Y) An opponent, observing Y but not having access to K or X, may attempt to recover X or K or both X and K. It is assumed that the opponent knows the encryption (E) and decryption (D) algorithms. If the opponent is interested in only this particular message, then the focus of the effort is to recover X by generating a plainn . Often, however, the opponent is interested in being able to read text estimate X future messages as well, in which case an attempt is made to recover K by generatn. ing an estimate K 3.1 / SYMMETRIC CIPHER MODEL 89 Cryptography Cryptographic systems are characterized along three independent dimensions: 1. The type of operations used for transforming plaintext to ciphertext. All encryption algorithms are based on two general principles: substitution, in which each element in the plaintext (bit, letter, group of bits or letters) is mapped into another element, and transposition, in which elements in the plaintext are rearranged. The fundamental requirement is that no information be lost (i.e., that all operations are reversible). Most systems, referred to as product systems, involve multiple stages of substitutions and transpositions. 2. The number of keys used. If both sender and receiver use the same key, the system is referred to as symmetric, single-key, secret-key, or conventional encryption. If the sender and receiver use different keys, the system is referred to as asymmetric, two-key, or public-key encryption. 3. The way in which the plaintext is processed. A block cipher processes the input one block of elements at a time, producing an output block for each input block. A stream cipher processes the input elements continuously, producing output one element at a time, as it goes along. Cryptanalysis and Brute-Force Attack Typically, the objective of attacking an encryption system is to recover the key in use rather than simply to recover the plaintext of a single ciphertext. There are two general approaches to attacking a conventional encryption scheme: ■ Cryptanalysis: Cryptanalytic attacks rely on the nature of the algorithm plus perhaps some knowledge of the general characteristics of the plaintext or even some sample plaintext–ciphertext pairs. This type of attack exploits the characteristics of the algorithm to attempt to deduce a specific plaintext or to deduce the key being used. ■ Brute-force attack: The attacker tries every possible key on a piece of ciphertext until an intelligible translation into plaintext is obtained. On average, half of all possible keys must be tried to achieve success. If either type of attack succeeds in deducing the key, the effect is catastrophic: All future and past messages encrypted with that key are compromised. We first consider cryptanalysis and then discuss brute-force attacks. Table 3.1 summarizes the various types of cryptanalytic attacks based on the amount of information known to the cryptanalyst. The most difficult problem is presented when all that is available is the ciphertext only. In some cases, not even the encryption algorithm is known, but in general, we can assume that the opponent does know the algorithm used for encryption. One possible attack under these circumstances is the brute-force approach of trying all possible keys. If the key space is very large, this becomes impractical. Thus, the opponent must rely on an analysis of the ciphertext itself, generally applying various statistical tests to it. To use this 90 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES Table 3.1 Types of Attacks on Encrypted Messages Type of Attack Ciphertext Only Known to Cryptanalyst ■ ■ Known Plaintext Encryption algorithm Ciphertext Encryption algorithm Ciphertext ■ One or more plaintext–ciphertext pairs formed with the secret key ■ ■ Chosen Plaintext Encryption algorithm Ciphertext ■ Plaintext message chosen by cryptanalyst, together with its corresponding ciphertext generated with the secret key ■ ■ Chosen Ciphertext Encryption algorithm Ciphertext ■ Ciphertext chosen by cryptanalyst, together with its corresponding decrypted plaintext generated with the secret key ■ ■ Chosen Text Encryption algorithm Ciphertext ■ Plaintext message chosen by cryptanalyst, together with its corresponding ciphertext generated with the secret key ■ Ciphertext chosen by cryptanalyst, together with its corresponding decrypted plaintext generated with the secret key ■ ■ approach, the opponent must have some general idea of the type of plaintext that is concealed, such as English or French text, an EXE file, a Java source listing, an accounting file, and so on. The ciphertext-only attack is the easiest to defend against because the opponent has the least amount of information to work with. In many cases, however, the analyst has more information. The analyst may be able to capture one or more plaintext messages as well as their encryptions. Or the analyst may know that certain plaintext patterns will appear in a message. For example, a file that is encoded in the Postscript format always begins with the same pattern, or there may be a standardized header or banner to an electronic funds transfer message, and so on. All these are examples of known plaintext. With this knowledge, the analyst may be able to deduce the key on the basis of the way in which the known plaintext is transformed. Closely related to the known-plaintext attack is what might be referred to as a probable-word attack. If the opponent is working with the encryption of some general prose message, he or she may have little knowledge of what is in the message. However, if the opponent is after some very specific information, then parts of the message may be known. For example, if an entire accounting file is being transmitted, the opponent may know the placement of certain key words in the header of the file. As another example, the source code for a program developed by Corporation X might include a copyright statement in some standardized position. If the analyst is able somehow to get the source system to insert into the system a message chosen by the analyst, then a chosen-plaintext attack is possible. An example of this strategy is differential cryptanalysis, explored in Appendix S. 3.1 / SYMMETRIC CIPHER MODEL 91 In general, if the analyst is able to choose the messages to encrypt, the analyst may deliberately pick patterns that can be expected to reveal the structure of the key. Table 3.1 lists two other types of attack: chosen ciphertext and chosen text. These are less commonly employed as cryptanalytic techniques but are nevertheless possible avenues of attack. Only relatively weak algorithms fail to withstand a ciphertext-only attack. Generally, an encryption algorithm is designed to withstand a known-plaintext attack. Two more definitions are worthy of note. An encryption scheme is unconditionally secure if the ciphertext generated by the scheme does not contain enough information to determine uniquely the corresponding plaintext, no matter how much ciphertext is available. That is, no matter how much time an opponent has, it is impossible for him or her to decrypt the ciphertext simply because the required information is not there. With the exception of a scheme known as the one-time pad (described later in this chapter), there is no encryption algorithm that is unconditionally secure. Therefore, all that the users of an encryption algorithm can strive for is an algorithm that meets one or both of the following criteria: ■ ■ The cost of breaking the cipher exceeds the value of the encrypted information. The time required to break the cipher exceeds the useful lifetime of the information. An encryption scheme is said to be computationally secure if either of the foregoing two criteria are met. Unfortunately, it is very difficult to estimate the amount of effort required to cryptanalyze ciphertext successfully. All forms of cryptanalysis for symmetric encryption schemes are designed to exploit the fact that traces of structure or pattern in the plaintext may survive encryption and be discernible in the ciphertext. This will become clear as we examine various symmetric encryption schemes in this chapter. We will see in Part Two that cryptanalysis for public-key schemes proceeds from a fundamentally different premise, namely, that the mathematical properties of the pair of keys may make it possible for one of the two keys to be deduced from the other. A brute-force attack involves trying every possible key until an intelligible translation of the ciphertext into plaintext is obtained. On average, half of all possible keys must be tried to achieve success. That is, if there are X different keys, on average an attacker would discover the actual key after X/2 tries. It is important to note that there is more to a brute-force attack than simply running through all possible keys. Unless known plaintext is provided, the analyst must be able to recognize plaintext as plaintext. If the message is just plain text in English, then the result pops out easily, although the task of recognizing English would have to be automated. If the text message has been compressed before encryption, then recognition is more difficult. And if the message is some more general type of data, such as a numerical file, and this has been compressed, the problem becomes even more difficult to automate. Thus, to supplement the brute-force approach, some degree of knowledge about the expected plaintext is needed, and some means of automatically distinguishing plaintext from garble is also needed. Hiva-Network.Com 92 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES 3.2 SUBSTITUTION TECHNIQUES In this section and the next, we examine a sampling of what might be called classical encryption techniques. A study of these techniques enables us to illustrate the basic approaches to symmetric encryption used today and the types of cryptanalytic attacks that must be anticipated. The two basic building blocks of all encryption techniques are substitution and transposition. We examine these in the next two sections. Finally, we discuss a system that combines both substitution and transposition. A substitution technique is one in which the letters of plaintext are replaced by other letters or by numbers or symbols.1 If the plaintext is viewed as a sequence of bits, then substitution involves replacing plaintext bit patterns with ciphertext bit patterns. Caesar Cipher The earliest known, and the simplest, use of a substitution cipher was by Julius Caesar. The Caesar cipher involves replacing each letter of the alphabet with the letter standing three places further down the alphabet. For example, plain: meet me after the toga party cipher: PHHW PH DIWHU WKH WRJD SDUWB Note that the alphabet is wrapped around, so that the letter following Z is A. We can define the transformation by listing all possibilities, as follows: plain: a b c d e f g h i j k l m n o p q r s t u v w x y z cipher: D E F G H I J K L M N O P Q R S T U V W X Y Z A B C Let us assign a numerical equivalent to each letter: a b c d e f g h i j k l m 0 1 2 3 4 5 6 7 8 9 10 11 12 n o p q r s t u v w x y z 13 14 15 16 17 18 19 20 21 22 23 24 25 Then the algorithm can be expressed as follows. For each plaintext letter p, substitute the ciphertext letter C:2 C = E(3, p) = (p + 3) mod 26 A shift may be of any amount, so that the general Caesar algorithm is C = E(k, p) = (p + k) mod 26 (3.1) 1 When letters are involved, the following conventions are used in this book. Plaintext is always in lowercase; ciphertext is in uppercase; key values are in italicized lowercase. We define a mod n to be the remainder when a is divided by n. For example, 11 mod 7 = 4. See Chapter 2 for a further discussion of modular arithmetic. 2 3.2 / SUBSTITUTION TECHNIQUES 93 where k takes on a value in the range 1 to 25. The decryption algorithm is simply p = D(k, C) = (C - k) mod 26 (3.2) If it is known that a given ciphertext is a Caesar cipher, then a brute-force cryptanalysis is easily performed: simply try all the 25 possible keys. Figure 3.3 shows the results of applying this strategy to the example ciphertext. In this case, the plaintext leaps out as occupying the third line. Three important characteristics of this problem enabled us to use a bruteforce cryptanalysis: 1. The encryption and decryption algorithms are known. 2. There are only 25 keys to try. 3. The language of the plaintext is known and easily recognizable. In most networking situations, we can assume that the algorithms are known. What generally makes brute-force cryptanalysis impractical is the use of an algorithm that employs a large number of keys. For example, the triple DES algorithm, PHHW PH DIWHU WKH WRJD SDUWB KEY 1 oggv og chvgt vjg vqic rctva 2 nffu nf bgufs uif uphb qbsuz 3 meet me after the toga party 4 ldds ld zesdq sgd snfz ozqsx 5 kccr kc ydrcp rfc rmey nyprw 6 jbbq jb xcqbo qeb qldx mxoqv 7 iaap ia wbpan pda pkcw lwnpu 8 hzzo hz vaozm ocz ojbv kvmot 9 gyyn gy uznyl nby niau julns 10 fxxm fx tymxk max mhzt itkmr 11 ewwl ew sxlwj lzw lgys hsjlq 12 dvvk dv rwkvi kyv kfxr grikp 13 cuuj cu qvjuh jxu jewq fqhjo 14 btti bt puitg iwt idvp epgin 15 assh as othsf hvs hcuo dofhm 16 zrrg zr nsgre gur gbtn cnegl 17 yqqf yq mrfqd ftq fasm bmdfk 18 xppe xp lqepc esp ezrl alcej 19 wood wo kpdob dro dyqk zkbdi 20 vnnc vn jocna cqn cxpj yjach 21 ummb um inbmz bpm bwoi xizbg 22 tlla tl hmaly aol avnh whyaf 23 skkz sk glzkx znk zumg vgxze 24 rjjy rj fkyjw ymj ytlf ufwyd 25 qiix qi ejxiv xli xske tevxc Figure 3.3 Brute-Force Cryptanalysis of Caesar Cipher 94 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES Figure 3.4 Sample of Compressed Text examined in Chapter 7, makes use of a 168-bit key, giving a key space of 2168 or greater than 3.7 * 1050 possible keys. The third characteristic is also significant. If the language of the plaintext is unknown, then plaintext output may not be recognizable. Furthermore, the input may be abbreviated or compressed in some fashion, again making recognition difficult. For example, Figure 3.4 shows a portion of a text file compressed using an algorithm called ZIP. If this file is then encrypted with a simple substitution cipher (expanded to include more than just 26 alphabetic characters), then the plaintext may not be recognized when it is uncovered in the brute-force cryptanalysis. Monoalphabetic Ciphers With only 25 possible keys, the Caesar cipher is far from secure. A dramatic increase in the key space can be achieved by allowing an arbitrary substitution. Before proceeding, we define the term permutation. A permutation of a finite set of elements S is an ordered sequence of all the elements of S, with each element appearing exactly once. For example, if S = {a, b, c}, there are six permutations of S: abc, acb, bac, bca, cab, cba In general, there are n! permutations of a set of n elements, because the first element can be chosen in one of n ways, the second in n - 1 ways, the third in n - 2 ways, and so on. Recall the assignment for the Caesar cipher: plain: a b c d e f g h i j k l m n o p q r s t u v w x y z cipher: D E F G H I J K L M N O P Q R S T U V W X Y Z A B C If, instead, the “cipher” line can be any permutation of the 26 alphabetic characters, then there are 26! or greater than 4 * 1026 possible keys. This is 10 orders of magnitude greater than the key space for DES and would seem to eliminate brute-force techniques for cryptanalysis. Such an approach is referred to as a monoalphabetic substitution cipher, because a single cipher alphabet (mapping from plain alphabet to cipher alphabet) is used per message. There is, however, another line of attack. If the cryptanalyst knows the nature of the plaintext (e.g., noncompressed English text), then the analyst can exploit the regularities of the language. To see how such a cryptanalysis might proceed, we give a partial example here that is adapted from one in [SINK09]. The ciphertext to be solved is 3.2 / SUBSTITUTION TECHNIQUES 95 UZQSOVUOHXMOPVGPOZPEVSGZWSZOPFPESXUDBMETSXAIZ VUEPHZHMDZSHZOWSFPAPPDTSVPQUZWYMXUZUHSX EPYEPOPDZSZUFPOMBZWPFUPZHMDJUDTMOHMQ As a first step, the relative frequency of the letters can be determined and compared to a standard frequency distribution for English, such as is shown in Figure 3.5 (based on [LEWA00]). If the message were long enough, this technique alone might be sufficient, but because this is a relatively short message, we cannot expect an exact match. In any case, the relative frequencies of the letters in the ciphertext (in percentages) are as follows: P 13.33 H Z 11.67 D S 8.33 E U 8.33 V O 7.50 X M 6.67 5.83 F 3.33 B 1.67 C 5.00 W 3.33 5.00 Q 2.50 4.17 T 4.17 A 0.00 G 1.67 K 0.00 Y 1.67 L 0.00 2.50 I 0.83 N 0.00 1.67 J 0.83 R 0.00 Comparing this breakdown with Figure 3.5, it seems likely that cipher letters P and Z are the equivalents of plain letters e and t, but it is not certain which is which. The letters S, U, O, M, and H are all of relatively high frequency and probably 12.702 14 9.056 12 0 A B C D E F G H J K L M N O Figure 3.5 Relative Frequency of Letters in English Text P Q R S T U V W X Y 0.074 0.150 1.974 2.360 2.758 0.095 0.978 1.929 2.406 I 0.772 0.153 2.015 2.228 1.492 2 2.782 4 4.025 4.253 6 6.327 5.987 7.507 6.749 6.094 8 6.996 8.167 Relative frequency (% ) 10 Z 96 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES correspond to plain letters from the set {a, h, i, n, o, r, s}. The letters with the lowest frequencies (namely, A, B, G, Y, I, J) are likely included in the set {b, j, k, q, v, x, z}. There are a number of ways to proceed at this point. We could make some tentative assignments and start to fill in the plaintext to see if it looks like a reasonable “skeleton” of a message. A more systematic approach is to look for other regularities. For example, certain words may be known to be in the text. Or we could look for repeating sequences of cipher letters and try to deduce their plaintext equivalents. A powerful tool is to look at the frequency of two-letter combinations, known as digrams. A table similar to Figure 3.5 could be drawn up showing the relative frequency of digrams. The most common such digram is th. In our ciphertext, the most common digram is ZW, which appears three times. So we make the correspondence of Z with t and W with h. Then, by our earlier hypothesis, we can equate P with e. Now notice that the sequence ZWP appears in the ciphertext, and we can translate that sequence as “the.” This is the most frequent trigram (three-letter combination) in English, which seems to indicate that we are on the right track. Next, notice the sequence ZWSZ in the first line. We do not know that these four letters form a complete word, but if they do, it is of the form th_t. If so, S equates with a. So far, then, we have UZQSOVUOHXMOPVGPOZPEVSGZWSZOPFPESXUDBMETSXAIZ t a e e te a that e e a a VUEPHZHMDZSHZOWSFPAPPDTSVPQUZWYMXUZUHSX e t ta t ha e ee a e th t a EPYEPOPDZSZUFPOMBZWPFUPZHMDJUDTMOHMQ e e e tat e the t Only four letters have been identified, but already we have quite a bit of the message. Continued analysis of frequencies plus trial and error should easily yield a solution from this point. The complete plaintext, with spaces added between words, follows: it was disclosed yesterday that several informal but direct contacts have been made with political representatives of the viet cong in moscow Monoalphabetic ciphers are easy to break because they reflect the frequency data of the original alphabet. A countermeasure is to provide multiple substitutes, known as homophones, for a single letter. For example, the letter e could be assigned a number of different cipher symbols, such as 16, 74, 35, and 21, with each homophone assigned to a letter in rotation or randomly. If the number of symbols assigned to each letter is proportional to the relative frequency of that letter, then single-letter frequency information is completely obliterated. The great mathematician Carl Friedrich Gauss believed that he had devised an unbreakable cipher using homophones. However, even with homophones, each element of plaintext affects only one element of ciphertext, and multiple-letter patterns 3.2 / SUBSTITUTION TECHNIQUES 97 (e.g., digram frequencies) still survive in the ciphertext, making cryptanalysis relatively straightforward. Two principal methods are used in substitution ciphers to lessen the extent to which the structure of the plaintext survives in the ciphertext: One approach is to encrypt multiple letters of plaintext, and the other is to use multiple cipher alphabets. We briefly examine each. Playfair Cipher The best-known multiple-letter encryption cipher is the Playfair, which treats digrams in the plaintext as single units and translates these units into ciphertext digrams.3 The Playfair algorithm is based on the use of a 5 * 5 matrix of letters constructed using a keyword. Here is an example, solved by Lord Peter Wimsey in Dorothy Sayers’s Have His Carcase:4 M O N A R C H Y B D E F G I/J K L P Q S T U V W X Z In this case, the keyword is monarchy. The matrix is constructed by filling in the letters of the keyword (minus duplicates) from left to right and from top to bottom, and then filling in the remainder of the matrix with the remaining letters in alphabetic order. The letters I and J count as one letter. Plaintext is encrypted two letters at a time, according to the following rules: 1. Repeating plaintext letters that are in the same pair are separated with a filler letter, such as x, so that balloon would be treated as ba lx lo on. 2. Two plaintext letters that fall in the same row of the matrix are each replaced by the letter to the right, with the first element of the row circularly following the last. For example, ar is encrypted as RM. 3. Two plaintext letters that fall in the same column are each replaced by the letter beneath, with the top element of the column circularly following the last. For example, mu is encrypted as CM. 4. Otherwise, each plaintext letter in a pair is replaced by the letter that lies in its own row and the column occupied by the other plaintext letter. Thus, hs becomes BP and ea becomes IM (or JM, as the encipherer wishes). The Playfair cipher is a great advance over simple monoalphabetic ciphers. For one thing, whereas there are only 26 letters, there are 26 * 26 = 676 digrams, 3 This cipher was actually invented by British scientist Sir Charles Wheatstone in 1854, but it bears the name of his friend Baron Playfair of St. Andrews, who championed the cipher at the British foreign office. 4 The book provides an absorbing account of a probable-word attack. CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES so that identification of individual digrams is more difficult. Furthermore, the relative frequencies of individual letters exhibit a much greater range than that of digrams, making frequency analysis much more difficult. For these reasons, the Playfair cipher was for a long time considered unbreakable. It was used as the standard field system by the British Army in World War I and still enjoyed considerable use by the U.S. Army and other Allied forces during World War II. Despite this level of confidence in its security, the Playfair cipher is relatively easy to break, because it still leaves much of the structure of the plaintext language intact. A few hundred letters of ciphertext are generally sufficient. One way of revealing the effectiveness of the Playfair and other ciphers is shown in Figure 3.6. The line labeled plaintext plots a typical frequency distribution of the 26 alphabetic characters (no distinction between upper and lower case) in ordinary text. This is also the frequency distribution of any monoalphabetic substitution cipher, because the frequency values for individual letters are the same, just with different letters substituted for the original letters. The plot is developed in the following way: The number of occurrences of each letter in the text is counted and divided by the number of occurrences of the most frequently used letter. Using the results of Figure 3.5, we see that e is the most frequently used letter. As a result, e has a relative frequency of 1, t of 9.056/12.702 ≈ 0.72, and so on. The points on the horizontal axis correspond to the letters in order of decreasing frequency. Figure 3.6 also shows the frequency distribution that results when the text is encrypted using the Playfair cipher. To normalize the plot, the number of occurrences of each letter in the ciphertext was again divided by the number of occurrences of e in the plaintext. The resulting plot therefore shows the extent to which the frequency distribution of letters, which makes it trivial to solve substitution 1.0 0.9 Plaintext 0.8 Normalized relative frequency 98 Playfair 0.7 0.6 0.5 0.4 Vigenère 0.3 0.2 Random polyalphabetic 0.1 0 1 2 3 4 5 6 1 7 8 9 10 10 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Frequency ranked letters (decreasing frequency) Figure 3.6 Relative Frequency of Occurrence of Letters 3.2 / SUBSTITUTION TECHNIQUES 99 ciphers, is masked by encryption. If the frequency distribution information were totally concealed in the encryption process, the ciphertext plot of frequencies would be flat, and cryptanalysis using ciphertext only would be effectively impossible. As the figure shows, the Playfair cipher has a flatter distribution than does plaintext, but nevertheless, it reveals plenty of structure for a cryptanalyst to work with. The plot also shows the Vigenère cipher, discussed subsequently. The Hill and Vigenère curves on the plot are based on results reported in [SIMM93]. Hill Cipher5 Another interesting multiletter cipher is the Hill cipher, developed by the mathematician Lester Hill in 1929. CONCEPTS FROM LINEAR ALGEBRA Before describing the Hill cipher, let us briefly review some terminology from linear algebra. In this discussion, we are concerned with matrix arithmetic modulo 26. For the reader who needs a refresher on matrix multiplication and inversion, see Appendix E. We define the inverse M -1 of a square matrix M by the equation M(M -1) = -1 M M = I, where I is the identity matrix. I is a square matrix that is all zeros except for ones along the main diagonal from upper left to lower right. The inverse of a matrix does not always exist, but when it does, it satisfies the preceding equation. For example, A = ¢ AA-1 = ¢ = ¢ 5 17 8 ≤ 3 A-1 mod 26 = ¢ (5 * 9) + (8 * 1) (17 * 9) + (3 * 1) 53 156 9 1 2 ≤ 15 (5 * 2) + (8 * 15) ≤ (17 * 2) + (3 * 15) 130 1 ≤ mod 26 = ¢ 79 0 0 ≤ 1 To explain how the inverse of a matrix is computed, we begin with the concept of determinant. For any square matrix (m * m), the determinant equals the sum of all the products that can be formed by taking exactly one element from each row and exactly one element from each column, with certain of the product terms preceded by a minus sign. For a 2 * 2 matrix, ¢ k11 k21 k12 ≤ k22 the determinant is k11k22 - k12k21. For a 3 * 3 matrix, the value of the determinant is k11k22k33 + k21k32k13 + k31k12k23 - k31k22k13 - k21k12k33 - k11k32k23. If a square 5 This cipher is somewhat more difficult to understand than the others in this chapter, but it illustrates an important point about cryptanalysis that will be useful later on. This subsection can be skipped on a first reading. 100 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES matrix A has a nonzero determinant, then the inverse of the matrix is computed as [A-1]ij = (det A)-1( -1)i + j(Dji), where (Dji) is the subdeterminant formed by deleting the jth row and the ith column of A, det(A) is the determinant of A, and (det A)-1 is the multiplicative inverse of (det A) mod 26. Continuing our example, det ¢ 5 17 8 ≤ = (5 * 3) - (8 * 17) = -121 mod 26 = 9 3 We can show that 9-1 mod 26 = 3, because 9 * 3 = 27 mod 26 = 1 (see Chapter 2 or Appendix E). Therefore, we compute the inverse of A as A = ¢ 5 17 A-1 mod 26 = 3¢ 8 ≤ 3 3 -17 -8 3 ≤ = 3¢ 5 9 18 9 ≤ = ¢ 5 27 54 9 ≤ = ¢ 15 1 2 ≤ 15 THE HILL ALGORITHM This encryption algorithm takes m successive plaintext letters and substitutes for them m ciphertext letters. The substitution is determined by m linear equations in which each character is assigned a numerical value (a = 0, b = 1, c , z = 25). For m = 3, the system can be described as c1 = (k11p1 + k21p2 + k31p3) mod 26 c2 = (k12p1 + k22p2 + k32p3) mod 26 c3 = (k13p1 + k23p2 + k33p3) mod 26 This can be expressed in terms of row vectors and matrices:6 k11 (c1 c2 c3) = (p1 p2 p3) £ k21 k31 k12 k22 k32 k13 k23 ≥ mod 26 k33 or C = PK mod 26 where C and P are row vectors of length 3 representing the plaintext and ciphertext, and K is a 3 * 3 matrix representing the encryption key. Operations are performed mod 26. 6 Some cryptography books express the plaintext and ciphertext as column vectors, so that the column vector is placed after the matrix rather than the row vector placed before the matrix. Sage uses row vectors, so we adopt that convention. Hiva-Network.Com 3.2 / SUBSTITUTION TECHNIQUES 101 For example, consider the plaintext “paymoremoney” and use the encryption key 17 K = £ 21 2 17 18 2 5 21 ≥ 19 The first three letters of the plaintext are represented by the vector (15 0 24). Then (15 0 24)K = (303 303 531) mod 26 = (17 17 11) = RRL. Continuing in this fashion, the ciphertext for the entire plaintext is RRLMWBKASPDH. Decryption requires using the inverse of the matrix K. We can compute det K = 23, and therefore, (det K)-1 mod 26 = 17. We can then compute the inverse as7 K -1 4 = £ 15 24 9 17 0 15 6≥ 17 This is demonstrated as 17 £ 21 2 17 18 2 5 4 21 ≥ £ 15 19 24 9 17 0 15 443 6 ≥ = £ 858 17 494 442 495 52 442 1 780 ≥ mod 26 = £ 0 365 0 0 1 0 0 0≥ 1 It is easily seen that if the matrix K-1 is applied to the ciphertext, then the plaintext is recovered. In general terms, the Hill system can be expressed as C = E(K, P) = PK mod 26 P = D(K, C) = CK-1 mod 26 = PKK-1 = P As with Playfair, the strength of the Hill cipher is that it completely hides single-letter frequencies. Indeed, with Hill, the use of a larger matrix hides more frequency information. Thus, a 3 * 3 Hill cipher hides not only single-letter but also two-letter frequency information. Although the Hill cipher is strong against a ciphertext-only attack, it is easily broken with a known plaintext attack. For an m * m Hill cipher, suppose we have m plaintext–ciphertext pairs, each of length m. We label the pairs Pj = (p1jp1j c pmj) and Cj = (c1jc1j c cmj) such that Cj = PjK for 1 … j … m and for some unknown key matrix K. Now define two m * m matrices X = (pij) and Y = (cij). Then we can form the matrix equation Y = XK. If X has an inverse, then we can determine K = X-1Y. If X is not invertible, then a new version of X can be formed with additional plaintext–ciphertext pairs until an invertible X is obtained. Consider this example. Suppose that the plaintext “hillcipher” is encrypted using a 2 * 2 Hill cipher to yield the ciphertext HCRZSSXNSP. Thus, we know that (7 8)K mod 26 = (7 2); (11 11)K mod 26 = (17 25); and so on. Using the first two plaintext-ciphertext pairs, we have 7 The calculations for this example are provided in detail in Appendix E. 102 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES ¢ 7 17 2 7 ≤ = ¢ 25 11 8 ≤K mod 26 11 The inverse of X can be computed: ¢ 7 11 8 -1 25 ≤ = ¢ 11 1 22 ≤ 23 so K = ¢ 25 1 22 7 ≤¢ 23 17 2 549 ≤ = ¢ 25 398 600 3 ≤ mod 26 = ¢ 577 8 2 ≤ 5 This result is verified by testing the remaining plaintext–ciphertext pairs. Polyalphabetic Ciphers Another way to improve on the simple monoalphabetic technique is to use different monoalphabetic substitutions as one proceeds through the plaintext message. The general name for this approach is polyalphabetic substitution cipher. All these techniques have the following features in common: 1. A set of related monoalphabetic substitution rules is used. 2. A key determines which particular rule is chosen for a given transformation. VIGENÈRE CIPHER The best known, and one of the simplest, polyalphabetic ciphers is the Vigenère cipher. In this scheme, the set of related monoalphabetic substitution rules consists of the 26 Caesar ciphers with shifts of 0 through 25. Each cipher is denoted by a key letter, which is the ciphertext letter that substitutes for the plaintext letter a. Thus, a Caesar cipher with a shift of 3 is denoted by the key value 3.8 We can express the Vigenère cipher in the following manner. Assume a sequence of plaintext letters P = p0, p1, p2, c , pn - 1 and a key consisting of the sequence of letters K = k0, k1, k2, c , km - 1, where typically m 6 n. The sequence of ciphertext letters C = C0, C1, C2, c , Cn - 1 is calculated as follows: C = C0, C1, C2, c , Cn - 1 = E(K, P) = E[(k0, k1, k2, c , km - 1), (p0, p1, p2, c , pn - 1)] = (p0 + k0) mod 26, (p1 + k1) mod 26, c ,(pm - 1 + km - 1) mod 26, (pm + k0) mod 26, (pm + 1 + k1) mod 26, c , (p2m - 1 + km - 1) mod 26, c Thus, the first letter of the key is added to the first letter of the plaintext, mod 26, the second letters are added, and so on through the first m letters of the plaintext. For the next m letters of the plaintext, the key letters are repeated. This process 8 To aid in understanding this scheme and also to aid in it use, a matrix known as the Vigenère tableau is often used. This tableau is discussed in a document at box.com/Crypto7e. 3.2 / SUBSTITUTION TECHNIQUES 103 continues until all of the plaintext sequence is encrypted. A general equation of the encryption process is Ci = (pi + ki mod m) mod 26 (3.3) Compare this with Equation (3.1) for the Caesar cipher. In essence, each plaintext character is encrypted with a different Caesar cipher, depending on the corresponding key character. Similarly, decryption is a generalization of Equation (3.2): pi = (Ci - ki mod m) mod 26 (3.4) To encrypt a message, a key is needed that is as long as the message. Usually, the key is a repeating keyword. For example, if the keyword is deceptive, the message “we are discovered save yourself” is encrypted as key: plaintext: ciphertext: deceptivedeceptivedeceptive wearediscoveredsaveyourself ZICVTWQNGRZGVTWAVZHCQYGLMGJ Expressed numerically, we have the following result. key 3 4 2 4 15 19 8 21 4 3 4 2 4 15 plaintext 22 4 0 17 4 3 8 18 2 14 21 4 17 4 ciphertext 25 8 2 21 19 22 16 13 6 17 25 6 21 19 key 19 8 21 4 3 4 2 4 15 19 8 21 4 plaintext 3 18 0 21 4 24 14 20 17 18 4 11 5 ciphertext 22 0 21 25 7 2 16 24 6 11 12 6 9 The strength of this cipher is that there are multiple ciphertext letters for each plaintext letter, one for each unique letter of the keyword. Thus, the letter frequency information is obscured. However, not all knowledge of the plaintext structure is lost. For example, Figure 3.6 shows the frequency distribution for a Vigenère cipher with a keyword of length 9. An improvement is achieved over the Playfair cipher, but considerable frequency information remains. It is instructive to sketch a method of breaking this cipher, because the method reveals some of the mathematical principles that apply in cryptanalysis. First, suppose that the opponent believes that the ciphertext was encrypted using either monoalphabetic substitution or a Vigenère cipher. A simple test can be made to make a determination. If a monoalphabetic substitution is used, then the statistical properties of the ciphertext should be the same as that of the language of the plaintext. Thus, referring to Figure 3.5, there should be one cipher letter with a relative frequency of occurrence of about 12.7%, one with about 9.06%, and so on. If only a single message is available for analysis, we would not expect an exact match of this small sample with the statistical profile of the plaintext language. Nevertheless, if the correspondence is close, we can assume a monoalphabetic substitution. 104 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES If, on the other hand, a Vigenère cipher is suspected, then progress depends on determining the length of the keyword, as will be seen in a moment. For now, let us concentrate on how the keyword length can be determined. The important insight that leads to a solution is the following: If two identical sequences of plaintext letters occur at a distance that is an integer multiple of the keyword length, they will generate identical ciphertext sequences. In the foregoing example, two instances of the sequence “red” are separated by nine character positions. Consequently, in both cases, r is encrypted using key letter e, e is encrypted using key letter p, and d is encrypted using key letter t. Thus, in both cases, the ciphertext sequence is VTW. We indicate this above by underlining the relevant ciphertext letters and shading the relevant ciphertext numbers. An analyst looking at only the ciphertext would detect the repeated sequences VTW at a displacement of 9 and make the assumption that the keyword is either three or nine letters in length. The appearance of VTW twice could be by chance and may not reflect identical plaintext letters encrypted with identical key letters. However, if the message is long enough, there will be a number of such repeated ciphertext sequences. By looking for common factors in the displacements of the various sequences, the analyst should be able to make a good guess of the keyword length. Solution of the cipher now depends on an important insight. If the keyword length is m, then the cipher, in effect, consists of m monoalphabetic substitution ciphers. For example, with the keyword DECEPTIVE, the letters in positions 1, 10, 19, and so on are all encrypted with the same monoalphabetic cipher. Thus, we can use the known frequency characteristics of the plaintext language to attack each of the monoalphabetic ciphers separately. The periodic nature of the keyword can be eliminated by using a nonrepeating keyword that is as long as the message itself. Vigenère proposed what is referred to as an autokey system, in which a keyword is concatenated with the plaintext itself to provide a running key. For our example, key: plaintext: ciphertext: deceptivewearediscoveredsav wearediscoveredsaveyourself ZICVTWQNGKZEIIGASXSTSLVVWLA Even this scheme is vulnerable to cryptanalysis. Because the key and the plaintext share the same frequency distribution of letters, a statistical technique can be applied. For example, e enciphered by e, by Figure 3.5, can be expected to occur with a frequency of (0.127)2 ≈ 0.016, whereas t enciphered by t would occur only about half as often. These regularities can be exploited to achieve successful cryptanalysis.9 VERNAM CIPHER The ultimate defense against such a cryptanalysis is to choose a keyword that is as long as the plaintext and has no statistical relationship to it. Such a system was introduced by an AT&T engineer named Gilbert Vernam in 1918. 9 Although the techniques for breaking a Vigenère cipher are by no means complex, a 1917 issue of Scientific American characterized this system as “impossible of translation.” This is a point worth remembering when similar claims are made for modern algorithms. 3.2 / SUBSTITUTION TECHNIQUES Key stream generator 105 Key stream generator Cryptographic bit stream (ki) Cryptographic bit stream (ki) Plaintext (pi) Ciphertext (ci ) Plaintext (pi) Figure 3.7 Vernam Cipher His system works on binary data (bits) rather than letters. The system can be expressed succinctly as follows (Figure 3.7): ci = pi ⊕ ki where pi = ith binary digit of plaintext ki = ith binary digit of key ci = ith binary digit of ciphertext ⊕ = exclusive@or (XOR) operation Compare this with Equation (3.3) for the Vigenère cipher. Thus, the ciphertext is generated by performing the bitwise XOR of the plaintext and the key. Because of the properties of the XOR, decryption simply involves the same bitwise operation: pi = ci ⊕ ki which compares with Equation (3.4). The essence of this technique is the means of construction of the key. Vernam proposed the use of a running loop of tape that eventually repeated the key, so that in fact the system worked with a very long but repeating keyword. Although such a scheme, with a long key, presents formidable cryptanalytic difficulties, it can be broken with sufficient ciphertext, the use of known or probable plaintext sequences, or both. One-Time Pad An Army Signal Corp officer, Joseph Mauborgne, proposed an improvement to the Vernam cipher that yields the ultimate in security. Mauborgne suggested using a random key that is as long as the message, so that the key need not be repeated. In addition, the key is to be used to encrypt and decrypt a single message, and then is discarded. Each new message requires a new key of the same length as the new message. Such a scheme, known as a one-time pad, is unbreakable. It produces random output that bears no statistical relationship to the plaintext. Because the ciphertext 106 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES contains no information whatsoever about the plaintext, there is simply no way to break the code. An example should illustrate our point. Suppose that we are using a Vigenère scheme with 27 characters in which the twenty-seventh character is the space character, but with a one-time key that is as long as the message. Consider the ciphertext ANKYODKYUREPFJBYOJDSPLREYIUNOFDOIUERFPLUYTS We now show two different decryptions using two different keys: ciphertext: ANKYODKYUREPFJBYOJDSPLREYIUNOFDOIUERFPLUYTS key: pxlmvmsydofuyrvzwc tnlebnecvgdupahfzzlmnyih plaintext: mr mustard with the candlestick in the hall ciphertext: ANKYODKYUREPFJBYOJDSPLREYIUNOFDOIUERFPLUYTS key: pftgpmiydgaxgoufhklllmhsqdqogtewbqfgyovuhwt plaintext: miss scarlet with the knife in the library Suppose that a cryptanalyst had managed to find these two keys. Two plausible plaintexts are produced. How is the cryptanalyst to decide which is the correct decryption (i.e., which is the correct key)? If the actual key were produced in a truly random fashion, then the cryptanalyst cannot say that one of these two keys is more likely than the other. Thus, there is no way to decide which key is correct and therefore which plaintext is correct. In fact, given any plaintext of equal length to the ciphertext, there is a key that produces that plaintext. Therefore, if you did an exhaustive search of all possible keys, you would end up with many legible plaintexts, with no way of knowing which was the intended plaintext. Therefore, the code is unbreakable. The security of the one-time pad is entirely due to the randomness of the key. If the stream of characters that constitute the key is truly random, then the stream of characters that constitute the ciphertext will be truly random. Thus, there are no patterns or regularities that a cryptanalyst can use to attack the ciphertext. In theory, we need look no further for a cipher. The one-time pad offers complete security but, in practice, has two fundamental difficulties: 1. There is the practical problem of making large quantities of random keys. Any heavily used system might require millions of random characters on a regular basis. Supplying truly random characters in this volume is a significant task. 2. Even more daunting is the problem of key distribution and protection. For every message to be sent, a key of equal length is needed by both sender and receiver. Thus, a mammoth key distribution problem exists. Because of these difficulties, the one-time pad is of limited utility and is useful primarily for low-bandwidth channels requiring very high security. The one-time pad is the only cryptosystem that exhibits what is referred to as perfect secrecy. This concept is explored in Appendix F. 3.3 / TRANSPOSITION TECHNIQUES 107 3.3 TRANSPOSITION TECHNIQUES All the techniques examined so far involve the substitution of a ciphertext symbol for a plaintext symbol. A very different kind of mapping is achieved by performing some sort of permutation on the plaintext letters. This technique is referred to as a transposition cipher. The simplest such cipher is the rail fence technique, in which the plaintext is written down as a sequence of diagonals and then read off as a sequence of rows. For example, to encipher the message “meet me after the toga party” with a rail fence of depth 2, we write the following: m e m a t r h t g p r y e t e f e t e o a a t The encrypted message is MEMATRHTGPRYETEFETEOAAT This sort of thing would be trivial to cryptanalyze. A more complex scheme is to write the message in a rectangle, row by row, and read the message off, column by column, but permute the order of the columns. The order of the columns then becomes the key to the algorithm. For example, Key: Plaintext: Ciphertext: 4 3 1 2 5 6 7 a t t a c k p o s t p o n e d u n t i l t w o a m x y z TTNAAPTMTSUOAODWCOIXKNLYPETZ Thus, in this example, the key is 4312567. To encrypt, start with the column that is labeled 1, in this case column 3. Write down all the letters in that column. Proceed to column 4, which is labeled 2, then column 2, then column 1, then columns 5, 6, and 7. A pure transposition cipher is easily recognized because it has the same letter frequencies as the original plaintext. For the type of columnar transposition just shown, cryptanalysis is fairly straightforward and involves laying out the ciphertext in a matrix and playing around with column positions. Digram and trigram frequency tables can be useful. The transposition cipher can be made significantly more secure by performing more than one stage of transposition. The result is a more complex permutation that is not easily reconstructed. Thus, if the foregoing message is reencrypted using the same algorithm, 108 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES Key: Input: Output: 4 3 1 2 5 6 7 t t n a a p t m t s u o a o d w c o i x k n l y p e t z NSCYAUOPTTWLTMDNAOIEPAXTTOKZ To visualize the result of this double transposition, designate the letters in the original plaintext message by the numbers designating their position. Thus, with 28 letters in the message, the original sequence of letters is 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 After the first transposition, we have 03 10 17 24 04 11 18 25 02 09 16 23 01 08 15 22 05 12 19 26 06 13 20 27 07 14 21 28 which has a somewhat regular structure. But after the second transposition, we have 17 09 05 27 24 16 12 07 10 02 22 20 03 25 15 13 04 23 19 14 11 01 26 21 18 08 06 28 This is a much less structured permutation and is much more difficult to cryptanalyze. 3.4 ROTOR MACHINES The example just given suggests that multiple stages of encryption can produce an algorithm that is significantly more difficult to cryptanalyze. This is as true of substitution ciphers as it is of transposition ciphers. Before the introduction of DES, the most important application of the principle of multiple stages of encryption was a class of systems known as rotor machines.10 The basic principle of the rotor machine is illustrated in Figure 3.8. The machine consists of a set of independently rotating cylinders through which electrical pulses can flow. Each cylinder has 26 input pins and 26 output pins, with internal wiring that connects each input pin to a unique output pin. For simplicity, only three of the internal connections in each cylinder are shown. If we associate each input and output pin with a letter of the alphabet, then a single cylinder defines a monoalphabetic substitution. For example, in Figure 3.8, if an operator depresses the key for the letter A, an electric signal is applied to 10 Machines based on the rotor principle were used by both Germany (Enigma) and Japan (Purple) in World War II. The breaking of both codes by the Allies was a significant factor in the war’s outcome. 3.4 / ROTOR MACHINES Direction of motion A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 24 25 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 21 3 15 1 19 10 14 26 20 8 16 7 22 4 11 5 17 9 12 23 18 2 25 6 24 13 Fast rotor 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 20 1 6 4 15 3 14 12 23 5 16 2 22 19 11 18 25 24 13 7 10 8 21 9 26 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 109 Direction of motion 8 18 26 17 20 22 10 3 13 11 4 23 5 24 9 12 25 16 19 6 15 21 2 7 1 14 Medium rotor Slow rotor (a) Initial setting A B C D E F G H I J K L M N O P Q R S T U V W X Y Z A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 23 24 25 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 13 21 3 15 1 19 10 14 26 20 8 16 7 22 4 11 5 17 9 12 23 18 2 25 6 24 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 20 1 6 4 15 3 14 12 23 5 16 2 22 19 11 18 25 24 13 7 10 8 21 9 26 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 8 18 26 17 20 22 10 3 13 11 4 23 5 24 9 12 25 16 19 6 15 21 2 7 1 14 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z Fast rotor Medium rotor Slow rotor (b) Setting after one keystroke Figure 3.8 Three-Rotor Machine with Wiring Represented by Numbered Contacts the first pin of the first cylinder and flows through the internal connection to the twenty-fifth output pin. Consider a machine with a single cylinder. After each input key is depressed, the cylinder rotates one position, so that the internal connections are shifted accordingly. Thus, a different monoalphabetic substitution cipher is defined. After 26 letters of plaintext, the cylinder would be back to the initial position. Thus, we have a polyalphabetic substitution algorithm with a period of 26. A single-cylinder system is trivial and does not present a formidable cryptanalytic task. The power of the rotor machine is in the use of multiple cylinders, in which the output pins of one cylinder are connected to the input pins of the next. Figure 3.8 shows a three-cylinder system. The left half of the figure shows a position in which the input from the operator to the first pin (plaintext letter a) is routed through the three cylinders to appear at the output of the second pin (ciphertext letter B). With multiple cylinders, the one closest to the operator input rotates one pin position with each keystroke. The right half of Figure 3.8 shows the system’s configuration after a single keystroke. For every complete rotation of the inner cylinder, the middle cylinder rotates one pin position. Finally, for every complete rotation of the middle cylinder, the outer cylinder rotates one pin position. This is the same type of operation seen with an odometer. The result is that there are 26 * 26 * 26 = 17,576 different substitution alphabets used before the system Hiva-Network.Com 110 CHAPTER 3 / CLASSICAL ENCRYPTION TECHNIQUES repeats. The addition of fourth and fifth rotors results in periods of 456,976 and 11,881,376 letters, respectively. Thus, a given setting of a 5-rotor machine is equivalent to a Vigenère cipher with a key length of 11,881,376. Such a scheme presents a formidable cryptanalytic challenge. If, for example, the cryptanalyst attempts to use a letter frequency analysis approach, the analyst is faced with the equivalent of over 11 million monoalphabetic ciphers. We might need on the order of 50 letters in each monalphabetic cipher for a solution, which means that the analyst would need to be in possession of a ciphertext with a length of over half a billion letters. The significance of the rotor machine today is that it points the way to a large class of symmetric ciphers, of which the Data Encryption Standard (DES) is the most prominent. DES is introduced in Chapter 4. 3.5 STEGANOGRAPHY We conclude with a discussion of a technique that (strictly speaking), is not encryption, namely, steganography. A plaintext message may be hidden in one of two ways. The methods of steganography conceal the existence of the message, whereas the methods of cryptography render the message unintelligible to outsiders by various transformations of the text.11 A simple form of steganography, but one that is time-consuming to construct, is one in which an arrangement of words or letters within an ap