It is shown that narrowing the difference between the high and low resistor values in the Kirchhoff Law–Johnson Noise (KLJN) key exchange strongly affects security against a recently introduced binary classifier-based wire resistance attack. Using time domain simulations of a non-ideal KLJN loop with finite cable resistance, we generate large ensembles of secure (HL/LH) bits and evaluate the mean-square noise voltages at Alice’s and Bob’s ends. For each bit, these mean-square values form a point in a two-dimensional classifier plane, where the separation between the HL and LH point clouds characterizes the information available to an eavesdropper (Eve). We quantify Eve’s success probability p by a simple decision rule based on the sign of the difference between the measured mean-square voltages. For strongly asymmetric resistors (for example [Formula: see text] kOhm and [Formula: see text] kOhm) and realistic wire resistances, the HL and LH clouds are fully separable and Eve’s p approaches 1, which confirms that the classifier attack can practically recover all secure bits. As the low resistor value approaches the high one (for example [Formula: see text] kOhm and [Formula: see text] kOhm) at the same cable resistance, the HL and LH clouds increasingly overlap and the measured p drops close to 0.7, approaching the ideal limit [Formula: see text] as [Formula: see text] approaches [Formula: see text]. A surprising phenomenon is that, in this classifier-based scenario, increasing the wire resistance can decrease the information leak. This counterintuitive effect is strikingly the opposite of the behavior in the classical Bergou–Scheuer–Yariv wire resistance attack, where the mean-square voltages at the two ends of the wire are simply compared.
It is shown that narrowing the difference between the high and low resistor values in the Kirchhoff Law–Johnson Noise (KLJN) key exchange strongly affects security against a recently introduced binary classifier-based wire resistance attack. Using time domain simulations of a non-ideal KLJN loop with finite cable resistance, we generate large ensembles of secure (HL/LH) bits and evaluate the mean-square noise voltages at Alice’s and Bob’s ends. For each bit, these mean-square values form a point in a two-dimensional classifier plane, where the separation between the HL and LH point clouds characterizes the information available to an eavesdropper (Eve). We quantify Eve’s success probability p by a simple decision rule based on the sign of the difference between the measured mean-square voltages. For strongly asymmetric resistors (for example [Formula: see text] kOhm and [Formula: see text] kOhm) and realistic wire resistances, the HL and LH clouds are fully separable and Eve’s p approaches 1, which confirms that the classifier attack can practically recover all secure bits. As the low resistor value approaches the high one (for example [Formula: see text] kOhm and [Formula: see text] kOhm) at the same cable resistance, the HL and LH clouds increasingly overlap and the measured p drops close to 0.7, approaching the ideal limit [Formula: see text] as [Formula: see text] approaches [Formula: see text]. A surprising phenomenon is that, in this classifier-based scenario, increasing the wire resistance can decrease the information leak. This counterintuitive effect is strikingly the opposite of the behavior in the classical Bergou–Scheuer–Yariv wire resistance attack, where the mean-square voltages at the two ends of the wire are simply compared.
Machine learning is shown not to increase Eve’s successful bit-guessing probability in the binary-classifier-based wire-resistance attack on the Kirchhoff Law-Johnson Noise (KLJN) key exchange when the resistor gap is narrow. Using time-domain simulations of a non-ideal KLJN loop with [Formula: see text] [Formula: see text]k[Formula: see text], [Formula: see text] [Formula: see text]k[Formula: see text] and wire resistances from 0 to 3200[Formula: see text] [Formula: see text], we generate 3000 secure bits per wire resistance and evaluate the mean-square noise voltages at both ends over 100-sample bit exchange periods (BEPs). This probability p is evaluated for the sign rule, the nearest-line rule, logistic regression, support vector machine (SVM), k-nearest neighbors, random forest and gradient boosting. The sign rule gives [Formula: see text] at[Formula: see text]50 [Formula: see text] and 0.682 at 3200[Formula: see text] [Formula: see text]. The nearest-line rule, logistic regression and SVM stay close to it and the other classifiers perform worse. For Gaussian noise, the likelihood ratio between the HL and LH cases depends only on the difference of the two mean-square voltages, so the sign rule is Bayes-optimal. The resulting [Formula: see text] reproduces the measured values and predicts that about 900 to 1200 samples per bit are needed for [Formula: see text]. This difference carries about 0.19 bit per secure bit, 1.5 times the hard-decision value.
The statistical fluctuations of the mean-square noise voltages measured at Alice’s and Bob’s ends in the KLJN scheme are used to implement a binary classifier for a new type of wire-resistance-based attack. The data are plotted on a two-dimensional graph, where the x- and y-axes represent the mean-square voltages at Alice’s and Bob’s ends, respectively. When the wire resistance is non-zero, the data form distinct lines for the LH and HL cases, allowing Eve to extract the secure bits with nearly 100% success. Further analysis shows that swapping the x and y axes for the LH data reproduces the curve for the HL case, effectively reducing the number of independent measurements by half. These results suggest that machine learning tools could exploit this property for enhanced detection performance, although such methods are unnecessary here since the LH and HL cases are completely separable. The only effective defense against this attack remains the traditional approach: properly increase the noise temperature on the side with lower resistance, or equivalently, scale down the noise temperature on the higher-resistance side. All claims are confirmed through computer simulations.
The unconditionally (that is, information theoretically) secure Kirchhoff-law-Johnson-noise (KLJN) key exchanger operates with two pairs of resistors, one at Alice's side and another one at Bob's end.Whenever the cable resistance is not negligible, the original scheme leaks information to the eavesdropper (Eve).The typical way to reduce this leak is treating the exchanged key by privacy amplification.In the present work our goal is to reduce the leak earlier, during the bit exchange level.So far, the common belief has been that the resistance values in Alice's/Bob's pairs should be very different for an easy separation of the bit values to yield low bit error probability in the key exchange.However, such situation is helping Eve during all the currently known types of attacks owing to higher information leak.We explore the possibility of enhancing the security by narrowing the difference between the resistance values.The impact on security and the cost of reduced communication speed are demonstrated for the Bergou-Scheuer-Yariv attack.
In (Nature) Science Report 5 (2015) 13653, Vadai, Mingesz and Gingl (VMG) introduce a new Kirchhoff-law-Johnson-noise (KLJN) secure key exchanger that operates with 4 arbitrary resistors (instead of 2 arbitrary resistance values forming 2 identical resistor pairs in the original system). They state that in this new, VMG-KLJN, non-equilibrium system with nonzero power flow, the security during the exchange of the two (HL and LH) bit values is as strong as in the original KLJN scheme. Moreover, they claim that, at practical conditions, their VMG-KLJN protocol “supports more robust protection against attacks”. First, we investigate the power flow and thermal equilibrium issues of the VMG-KLJN system with 4 arbitrary resistors. Then we introduce a new KLJN protocol that allows the arbitrary choice of 3 resistors from the 4, while it still operates with zero power flow during the exchange of single bits by utilizing a specific value of the 4th resistor and a binary temperature set for the exchanged (HL and LH) bit values. Then we show that, in general, the KLJN schemes with more than 2 arbitrary resistors (including our new protocol mentioned above) are prone to 4 new passive attacks utilizing the parasitic capacitance and inductance in the cable, while the original KLJN scheme is naturally immune against these new attacks. The core of the security vulnerability exploited by these attacks is the different line resistances in the HL and LH cases. Therefore, on the contrary of the statement and claim cited above, the practical VMG-KLJN system is less secure than the original KLJN scheme. We introduce another 2, modified, non-equilibrium KLJN systems to eliminate the vulnerability against some - but not all - of these attacks. However the price for that is the loss of arbitrariness of the selection of the 4th resistor and the information leak still remains greater than zero.
The information-theoretically (unconditionally) secure Kirchhoff-law-Johnson-noise (KLJN) bit exchange protocol uses two identical resistor pairs with high (H) and low (L) resistance values, driven by Gaussian noise generators emulating Johnson noise with a high common temperature. The resulting mean-square noise voltage on the wire connecting Alice and Bob has three levels: low (L/L), intermediate (H/LorL/H), and high (H/H), and secure key sharing is achieved at the intermediate level (L/HorH/L). This paper introduces the Flip-KLJN scheme, where a pre-agreed intermediate level, such asH/L, triggers a flip of the bit map value during the bit exchange period. For Eve, the bit map flips appear random. Thus, the formerly discardedH/HandL/Lsituations can also have a pre-agreed bit value mapping, which flips together with the original bit mapping. Thus, Flip-KLJN doubles the key rate and ensures that all three levels on the wire are indistinguishable for Eve. Bit error probabilities are addressed through analytic calculations and computer simulations.
We demonstrate the security vulnerability of an ideal Kirchhoff–Law–Johnson–Noise key exchanger against transient attacks. Transients start when Alice and Bob connect a wire to their chosen resistor at the beginning of each clock cycle. A transient attack takes place during a short duration of time, before the transients reflected from the end of Alice and Bob mix together. The information leak arises from the fact that Eve (the eavesdropper) monitors the cable and analyzes the transients during this time period. We will demonstrate such a transient attack, and then, we introduce a defense protocol to protect against the attack. Computer simulations demonstrate that after applying the defense method the information leak becomes negligible.