## RSA

This trapdoor encryption system was introduced by Rivest, Shamir and Adleman (1978). It has, so far, withstood all known attacks.

The system is simplicity itself. Each user of the system makes two numbers, eU and nU public and keeps a number dU secret. In order for A to send a message to B, A looks up B's public values and, if the message is m (written as a number), then A blocks the message into pieces of size < nB and sends c = meB mod nB. Then B decodes by m = cdB mod nB. The security of the system lies in the choices of the public and private keys. To understand these choices we need to consider some number theory.

For any integer n, Euler's Totient Function, (n) is the number of integers greater than or equal to 1 which are relatively prime to n. It can be shown that:

Ex: (12) = |{1,5,7,11}| = 4.

12(1-1/2)(1-1/3) = 12(1/2)(2/3) = 2(2) = 4.

Euler's Theorem: If gcd (a,n) = 1 then

Corollary: If n is a product of distinct primes then for any integer t.

Pf: Let p be any prime that divides n. If gcd(a,p) = 1, then is valid by Euler's Theorem. On the other hand, if a0 mod p, then the statement is trivially true. Since the congruence holds for each prime dividing n, it also holds for n.

For the RSA choices, each user selects two prime numbers (about 100 digits long) p and q and sets nU = pq. Note that (nU) = (p-1)(q-1). [p and q are no longer used, but must be kept secret]. Next, eU is selected subject to 1 < eU < (nU) and gcd(eU, (nU)) = 1. Finally, dU is calculated (using the extended Euclidean Algorithm) so that eU dU1 mod (nU). We now see that

by the corollary.

Finding p and q can be done with a fast primality tester.

The practical user of RSA must be on guard against some common pitfalls, known as protocol failures. In these cases, how a message gets encoded to a numerical equivalent may defeat the cryptosystem.

The RSA scheme can be used for signatures in the usual way.

The only known way to break the system is to find (nU) which is almost equivalent to factoring nU. The Rabin variation is a version of RSA in which it can be shown that the security is equivalent to the difficulty of factoring.

So the security rests (perhaps) on the difficulty of factoring large numbers. To avoid those situations where fast factoring algorithms exist one should select p and q so that

1. p and q are not too close (one should be a few decimal digits longer)
2. p-1 and q-1 have a small gcd and both have at least one large prime factor.
One must also be aware of the current state of the art in factoring large numbers. The best general algorithms at the moment are: