The Fraud Prevention Source

Shared Secret Key Authentication

Over two thousand years ago, cryptosystems, secret communication procedures, were used to conceal the meanings in written messages and keep communications secret. These same cryptosystems can also be used to authenticate or verify information, which is not secret - information printed on financial instruments, tickets, product packages and containers, identification cards and even biometric information.

Let's look at a very simple 'shared secret key' cryptosystem to see how information can be authenticated or verified.

A Simple 'Shared Secret Key' Cryptosystem

In this simple 'shared secret key' cryptosystem there is an algorithm and a key, which is known only to the creator and authenticator of information - thus, a 'shared secret key' cryptosystem. The algorithm and the 'shared secret key' are used to encrypt (scramble) information and decrypt (unscramble) information.

The two components of this simple cryptosystem include the:

	Algorithm	- Letter Substitution
	Key	- Substitution of each letter with the letter    twenty places to the right as follows:

Information: abcdefghijklmnopqrstuvwxyz Substitute: uvwxyzabcdefghijklmnopqrst

In this simple system, during encryption the letter a is substituted or written with the letter u, b with v, c with w, etc.. When decrypting a written u is substituted or read as a, v as b, w as c, etc.. This simple cryptosystem can be used to authenticate or verify information which is not secret. For example, an identification card's descriptive information, "6 feet, 200 pounds and brown hair", can be encrypted and that encrypted information, along with the actual descriptive information, can be printed on the identification card.

When the identification card is used, the descriptive information, "6 feet, 200 pounds and brown hair", can be verified or authenticated by first decrypting the encrypted information printed on the identification card. This decrypted information can then be compared to the descriptive information printed on the card and of course to the actual appearance of the card's bearer.

In this example only the identification card issuer and authenticator know the algorithm and key. Thus, fraud perpetrators, who wish to create counterfeit or altered identification cards but don't know the algorithm and key, cannot create the correct encrypted information necessary to pass the authenticator's verification tests.

However, if fraud perpetrators knew that a letter substitution system was used as the algorithm, they could guess at the 'shared secret key' and their chance of success would be 1 in 25 (For example, the letter a can be equal to any letter except a). Also, if someone, in possession of encrypted and decrypted information, knew or guessed that a letter substitution system was used, he or she could determine the 'shared secret key' by simple 'trial and error' calculations because there are only 25 possible 'shared secret keys'. It would not take very long to test the 25 possible keys in this system.

So this simple system could be used as an authentication system, but it would have minimal security, because it would be relatively easy for fraud perpetrators to determine the correct 'shared secret key' and then create counterfeit information that would pass the system's authentication tests.

More Advanced Cryptosystems

To reduce the risks of 'trial and error' calculations discovering a 'shared secret key', it is necessary to use a cryptosystem with more possible keys. In general cryptosystems with a large number of possible keys are more secure than those with a small number, because it requires more time to test a large number of possible keys by 'trial and error' than to test a small number by 'trial and error'.

The more time it takes to test all the possible keys, the more secure is the system. For example, in a low security cryptosystem all the possible keys might be tested in a few minutes, while in a high security cryptosystem it might take ten years to test all the possible keys.

Today with the abundance of inexpensive powerful computers almost anyone can test a large number of possible keys by 'trial and error' in a very short period of time. Therefore, cryptosystems, which may have been secure before the computer age, are not secure today. Any modern cryptosystem must involve computers - or must be developed with computers in mind.

Modern 'Shared Secret Key' Cryptosystems

As computers became ubiquitous, encryption algorithms, like the Letter Substitution, were replaced by complex mathematical systems. Probably the best known 'shared secret key' cryptosystem of the computer age is the Data Encryption Standard (DES), which was developed in the late 1970s. DES, initially used to secure computer databases and electronic communications, was promulgated by the National Institute of Standards and Technology (NIST) to provide a system for the protection of the confidentiality and integrity of the federal government's sensitive unclassified computer information. Later it was also used for security in many other areas, such as the electronic transfer of funds and Internet security.

In the DES cryptosystem all information is converted into bits (0 or 1) and each key is 56 bits long. Because each key is 56 bits long there are more than 70 quadrillion (2 to the 56^th power) possible keys which can be used with the DES algorithm.

Since it would have taken many years to discover a correct DES key by 'trial and error', this algorithm was considered very secure in the late 1970s. DES remained the security standard for years, but by the 1990s, as cheaper computers became more powerful, it became less costly and time consuming to discover a correct DES key by 'trial and error'.

More secure (longer key lengths with more possible keys) algorithms were developed and in the fall of 2000, the Advanced Encryption Standard (AES) was selected by the National Institute of Standards and Technology (NIST) as the replacement for DES. AES has optional Key lengths of 128, 192 and 256 bits so that the number of possible Keys is 2 to the 128^th, 192nd and 256^th power when using AES. With current technology it would take many, many years to discover by 'trial and error' a correct AES Key.

Thus, mathematicians have responded to advances in computer technology by creating more secure (longer key lengths with more possible keys) cryptosystems which make 'trial and error' key determination almost impossible.

Authentication with Modern 'Shared Secret Key' Cryptosystems

Modern 'shared secret key' cryptosystems can be used to authenticate information about financial instruments, tickets, packages and containers, which carry products, and identification cards. For example, they can be used to authenticate a seat's location printed on an event ticket.

Here is one way this can be accomplished. An event ticket's seat and row numbers, along with the seat's section, and a 'shared secret key' could be entered into an encryption algorithm like AES. The output would be 'encrypted information', which is a scrambled representation of the seat's location printed on the ticket. This 'encrypted information' could then be printed on the ticket or the 'encrypted information' could be truncated to create an Authentication Code. For example, the 'encrypted information' could be converted to a 4 digit Authentication Code which could be printed on the ticket.

Then when a ticket was presented for entrance to the event the event ticket's seat and row numbers, along with the seat's section could be entered into the same encryption algorithm along with the same 'shared secret key', which was used to create the ticket's 'encrypted information' or Authentication Code. This calculated information could then be compared to the information printed on the ticket. If the calculated information was not equal to the information printed on the ticket, then it would be assumed that the ticket was a counterfeit.

In this system fraud perpetrators cannot create counterfeit tickets, which will pass authentication, because they don't know the 'shared secret key' and they cannot determine the 'shared secret key' by trial and error in any reasonable time period. Without the 'shared secret key' they cannot create the correct 'encrypted information' or Authentication Code necessary to pass the authentication tests.

For additional security a Hash Function, which is a one way function whose output cannot be decrypted, can be used to calculate a digest or fingerprint of the information that will be authenticated. See Ethent Views: "Hash Function Authentication" for a discussion of Hash Functions.

Ethent Views

2) The 'shared secret key' size should be sufficiently long so that 'trial and error' calculations are not feasible and no more than 2 parties should know the 'shared secret key'. If a large number of parties know the 'shared secret key' there is substantial risk that one of the parties will use the 'shared secret key' to create fraudulent documents, which will pass the authentication tests.