Claes G. Granqvist

Laszlo Bela Kish, Claes-Goran Granqvist

We introduce two new Kirchhoff-law-Johnson-noise (KLJN) secure key distribution schemes which are generalizations of the original KLJN scheme. The first of these, the Random-Resistor (RR-) KLJN scheme, uses random resistors with values chosen from a quasi-continuum set. It is well-known since the creation of the KLJN concept that such a system could work in cryptography, because Alice and Bob can calculate the unknown resistance value from measurements, but the RR-KLJN system has not been addressed in prior publications since it was considered impractical. The reason for discussing it now is the second scheme, the Random-Resistor-Random-Temperature (RRRT-) KLJN key exchange, inspired by a recent paper of Vadai, Mingesz and Gingl, wherein security was shown to be maintained at non-zero power flow. In the RRRT-KLJN secure key exchange scheme, both the resistances and their temperatures are continuum random variables. We prove that the security of the RRRT-KLJN scheme can prevail at non-zero power flow, and thus the physical law guaranteeing security is not the Second Law of Thermodynamics but the Fluctuation-Dissipation Theorem. Alice and Bob know their own resistances and temperatures and can calculate the resistance and temperature values at the other end of the communication channel from measured voltage, current and power-flow data in the wire. However, Eve cannot determine these values because, for her, there are four unknown quantities while she can set up only three equations. The RRRT-KLJN scheme has several advantages and makes all former attacks on the KLJN scheme invalid or incomplete.

L. B. Kish, C. G. Granqvist

A simple and general proof is given for the information theoretic (unconditional) security of the Kirchhoff-law-Johnson-noise (KLJN) key exchange system under practical conditions. The unconditional security for ideal circumstances, which is based on the Second Law of Thermodynamics, is found to prevail even under slightly non-ideal conditions. This security level is guaranteed by the continuity of functions describing classical physical linear, as well as stable non-linear, systems. Even without privacy amplification, Eve's probability for successful bit-guessing is found to converge towards 0.5 - i.e., the perfect security level - when ideal conditions are approached.

Yessica Saez, Laszlo B. Kish, Robert Mingesz et al.

We classify and analyze bit errors in the current measurement mode of the Kirchhoff-law-Johnson-noise (KLJN) key distribution. The error probability decays exponentially with increasing bit exchange period and fixed bandwidth, which is similar to the error probability decay in the voltage measurement mode. We also analyze the combination of voltage and current modes for error removal. In this combination method, the error probability is still an exponential function that decays with the duration of the bit exchange period, but it has superior fidelity to the former schemes.