Unconditionally Secure Key Distribution Against Active Adversaries

In this report we will give an overview of current results and protocols for unconditional secure keyagreement based on access to shared randomness, see [2, 3, 4], and determine how useful these are in practice, and finally from this propose new open problems that could be looked at in the future. We assume that the reader is familiarly with the concepts of entropy and the passive key-generation algorithm as described in [6]. The key-agreement problem is fundamental to cryptography, it is the problem of generating a secret key between the parties Alice and Bob, which initially does not share a key, while the adversary Eve has access to the communications channel between Alice and Bob. We can now make different assumptions on Eve, in public-key cryptography as proposed by Diffie and Hellman, we assume 1) Eve have limited computational resources, say she is unable to solve the factoring or discrete log problem in feasible time. Furthermore 2) Eve is a passive adversaries, that is she can only read what is sent through the communications channel between Alice and Bob, this way the communication between Alice and Bob is authenticated. In [2, 3, 4] these two assumptions are dropped, so Eve has infinite computing power and has complete control over the communications channel used by Alice and Bob, so we want unconditional secure key-agreement protocols by use of an insecure an unauthenticated channel. Unconditional security has not been widely used in practice due to Shannons discouraging result, see [5], which basically says that we need a key at least as long as the message we want to send, to achieve unconditional security, and we need a new key for each message we want to send. But it turns out, as we will see in this report, that if the parties wanting to communicate have access to correlated random bits that has a property called non-simulatability, then they can agree on a secret key by use of a insecure and unauthentic channel which an adversary has complete control over, provided, of course, that the adversary does not block all communications through the channel. Then they can use this key as a one-time pad and achieve a communications channel with unconditional security.