Open Mind Tree

Suppose Bob wants to send Alice a message via some insecure medium, and, to ensure that such messages are unreadable by anybody else, Bob and Alice have agreed previously that any messages they exchange will be encoded using the Caesar Cipher. Note that the medium itself doesn't matter at all: it could be a hand-penned message written on parchment and carried on horseback or it could be ASCII text sent as an e-mail.

The message that Bob wishes to send is called plaintext, the process of encoding it known as encryption and the resulting encoded message called ciphertext. To create ciphertext from plaintext using the Caesar Cipher is a simple task for Bob to perform. He merely replaces each letter in the message with the letter in the alphabet three places to its right - wrapping around back to 'A' after the letter 'Z'. So, a letter 'A' in the plaintext becomes a letter 'D' in the ciphertext, a 'B' becomes an 'E', and so on, all the way up to 'Z' which is replaced by a 'C'. Having encrypted his message, Bob then sends the ciphertext on its way to Alice, who, on receiving it, applies the reverse process - known as decryption - and thus obtains the original plaintext message. To decrypt the encoded message, she replaces each letter in the ciphertext with the letter in the alphabet three places to its left. Hey presto, the original message is recreated!

Now, the Caesar Cipher is trivial to break, and, once broken, all future communication between Alice and Bob would be insecure. It doesn't take a huge leap of the imagination, however, for Alice and Bob to make things slightly more difficult for their enemies. With each message they wish to encrypt, they could vary the amount which they shift the letters by - rather than always shifting by a fixed amount of three. Here they are generalizing from the Caesar Cipher to a whole class of ciphers, a class defined by the cryptographic algorithm 'replace each letter in the plaintext with the letter n places to its right in the alphabet'. The number n in each application of this algorithm is known as the cryptographic key. To be able to decrypt the ciphertext now you need to know both the algorithm and the key (the value of n which was used to encrypt the message).

This 'shift right' cipher system is a simple example of what is called a block cipher. This is because the cipher system divides the plaintext message up into fixed size blocks and operates on each of these blocks in turn. In our case, the block size is determined by how we choose to represent each letter in the plaintext. If we were to use the familiar 8-bit ASCII encoding for text, for example, the block size would be 8 bits. Modern block ciphers use larger block sizes than this and, depending on the algorithm, will work with a fixed block size or one of series of supported block sizes; moreover, the operation performed on each block will be much more complex than our simple 'shift right' operation and will typically involve several 'rounds' or applications of the encrypting operation and may involve feedback from a previously encrypted block. The basic principle is the same, though.

This problem of key management was solved in 1976 with the invention of public key cryptography by Whitfield Diffie and Martin Hellman at Stanford University.Most modern e-mail clients support PGP or GnuPG for encryption and signing. In public key cryptography, two mathematically related keys are generated for each party wishing to communicate - a public key and a private key. A message encoded using a particular public key can only be decrypted using the corresponding private key.

For Bob to send a message to Alice using public key cryptography, Bob encrypts the message using Alice's public key. Bob can now transmit the ciphertext to Alice safe in the knowledge that only Alice's private key can be used to decrypt and read it. Since her private key is used for decryption, Alice needs to keep this safe; but because she doesn't need to give this private key to anybody for them to be able to communicate securely with here, this is relatively easy to do. She need to take no such precautions with here public key. In fact, anybody can have access to it. She can post it to here friends in e-mails or even publish it on the web.

Public key algorithms rely for their security on the fact that certain types of mathematical function are easy to perform in one direction but are difficult to do in reverse. For example, the popular RSA algorithm (named after its inventors Ronald Rivest, Adi Shamir and Leonard Adlemann) makes use of the factorization problem: while it's easy to multiply two large prime numbers together, it is vastly more difficult to factor that product back into two large primes. It is exactly this problem that would need to be solved for the code-breaker to be able to deduce an RSA private key form the corresponding public key. Other public key algorithms take advantage of other computationally hard problems, such as the discrete logarithm problem or elliptic curves.

What, then, is the solution to security and privacy on the Internet? The answer is public key cryptography.Web sites that employ the HTTPS protocol use SSL for authentication and encryption.It is public key cryptography that makes e-commerce on the web possible. It is used, for example, in the Secure Socket Layer (SSL) protocol which enables secure access to we pages. It is public key cryptography, as used in software like PGP (Pretty Good Privacy), that can safeguard the privacy of your e-mails and data. And it is public key cryptography which permits the creation of digital signatures, so that the authenticity and integrity of digital messages can be verified.

Most applications of cryptography on the Internet, such as SSL, work as an additional layer over existing Internet protocols. A more recent development, IPSec (Internet Protocol Security), is a secure replacement for the Internet Protocol (IP) using public key cryptography (IP is one of the fundamental Internet protocols and is responsible for the packaging and routing of information transmitted via the Internet). An increasingly popular application for IPSec today is Virtual Private Networking, the creation of a secure, private network on top of an untrusted, public network like the Internet. All data sent over a VPN is encrypted by IPSec, so the application software used to communicate over a VPN doesn't have to use or support cryptography for the data to remain secure.

The first step in creating a digital signature is to use a cryptographic hash function - such as the popular MD5 algorithm - to generate a message digest or "fingerprint" of the message or file to be signed. This message digest is a sequence of bits (128 bits in the case of the MD5 algorithm) which uniquely or very nearly uniquely represents the message. A good hash function is designed to make it very difficult to generate another message which would hash to the same sequence of bits. Thus any modification of the message should result in a different message digest.

Next, the message digest is encrypted with the author's private key to create the message signature. Both message and signature are then distributed.

Suppose Alice wishes to verify that a digitally signed message she has received was indeed from Bob. She uses Bob's public key to decrypt the message's digital signature and so reveal the message digest. She then computes a fresh message digest for the message that she received and compares it to the one that was encoded in the signature. If they don't match, either the message wasn't actually from Bob or the message has been tampered with in some way.

One measure of an algorithm's strength is the size of its keyspace, the set of all possible keys that can be used with that algorithm. For example, in our simple 'shift right' algorithm discussed earlier, the keyspace is the set of numbers from 1 to 25 (although we can shift by 26 letters, the resulting ciphertext would be identical to the plaintext, so we won't count that). If an algorithm's keyspace is too small, then it is possible to break the encryption by brute force: we simply try each possible key in turn and see if the ciphertext decrypts to a meaning plaintext message. With our simple example, this type of attack is trivial to perform, even without computer power, and the cipher is hopelessly insecure.

More sophisticated algorithms can also be open to brute force attack, however, especially with the readily available power of today's microprocessors. Algorithms that may have once been considered secure are safe no longer. For example, the DES algorithm - which has a keyspace of 2⁵⁶ (that is 2 multiplied by itself 56 times) which is roughly 7x10¹⁶ keys (or 7 followed by sixteen zeroes) - can be broken by brute force in a matter of hours by specialized devices. It was also broken by the distributed.net project - which utilizes the otherwise unused CPU cycles of hundreds of average PCs via the Internet - in under 24 hours in January 1999. The inexorable increase in processor power of the desktop PC makes this type of exhaustive search more and more accessible.

The term "strong cryptography" is applied to algorithms that, even when the algorithm is known, are not prone to brute force attack. Currently, for block ciphers a key size of 80 bits (a keyspace of 2⁸⁰) and above is considered secure for everyday purposes, while for public key systems such as RSA a safe key size is at least 1024 bits. These key sizes are considered safe simply because there is currently insufficient computing power in the world to exhaustively search such large keyspaces in a practical time.

To crack strong cryptography, more subtlety is required: one must look for ways to reduce the number of keys that it is necessary search. This typically requires complex mathematical analysis, or cryptanalysis, of the particular algorithm to find its weakness, and several algorithms have been shown to be insecure in this way.

Traditional measures have been to limit the exportation of strong cryptographic technologies, but with the advent of e-commerce this is no longer a practical or effective remedy. Alternative proposals - such as a key escrow scheme (where to be able to legally use public key cryptography, you must register and hand over a copy of your private key to the government) or "backdoors" to encryption, such as the Clipper chip proposed by the Clinton administration in the US - have been shelved thanks to the lobbying of civil liberties groups such as the Electronic Frontier Foundation (http://www.eff.org). After all, why should we trust the government not to misuse such powers? What's more, any such central repository of keys - as proposed by the escrow scheme - would be a prime target for criminal attack. However, with the recent outbreaks of terrorist activity, many fear that both the US and the UK governments will once more try to push through legislation to limit the use of strong cryptography.

The fact is that everybody has a legitimate use of strong cryptography, even it is only to secure their purchases over the Internet. The next time you log in to Amazon.com to buy the latest bestseller, have a think about how cryptography is being used to protect your privacy and to protect you against fraud. Do you really want to surrender that protection to the government?