Comments On 'A New Transient Attack On The Kish Key Distribution System'

📝 Abstract

A recent IEEE Access Paper by Gunn, Allison and Abbott (GAA) proposed a new transient attack against the Kirchhoff-law-Johnson-noise (KLJN) secure key exchange system. The attack is valid, but it is easy to build a defense for the KLJN system. Here we note that GAA’s paper contains several invalid statements regarding security measures and the continuity of functions in classical physics. These deficiencies are clarified in our present paper, wherein we also emphasize that a new version of the KLJN system is immune against all existing attacks, including the one by GAA.

💡 Analysis

A recent IEEE Access Paper by Gunn, Allison and Abbott (GAA) proposed a new transient attack against the Kirchhoff-law-Johnson-noise (KLJN) secure key exchange system. The attack is valid, but it is easy to build a defense for the KLJN system. Here we note that GAA’s paper contains several invalid statements regarding security measures and the continuity of functions in classical physics. These deficiencies are clarified in our present paper, wherein we also emphasize that a new version of the KLJN system is immune against all existing attacks, including the one by GAA.

📄 Content

Metrol. Meas. Syst., Vol. accepted for publication (May, 2016)

1

COMMENTS ON “A NEW TRANSIENT ATTACK ON THE KISH KEY DISTRIBUTION SYSTEM” Laszlo B. Kish 1), Claes G. Granqvist 2)

1. Texas A&M University, Department of Electrical and Computer Engineering, College Station, TX 77843-3128, USA
2. Department of Engineering Sciences, The Ångström Laboratory, Uppsala University, P.O. Box 534, SE-75121 Uppsala, Sweden Abstract
   A recent IEEE Access Paper by Gunn, Allison and Abbott (GAA) proposed a new transient attack against the Kirchhoff-law– Johnson-noise (KLJN) secure key exchange system. The attack is valid, but it is easy to build a defense for the KLJN system. Here we note that GAA’s paper contains several invalid statements regarding security measures and the continuity of functions in classical physics. These deficiencies are clarified in our present paper, wherein we also emphasize that a new version of the KLJN system is immune against all existing attacks, including the one by GAA.
   Keywords: Measurement theory; Information security; Foundations of physics; Engineering over-simplifications.


   
Research and development on unconditionally secure communication and key exchange have a history of progress via attacks and debates and, for example, this type of evolution has taken place for quantum key distributions (QKDs) [1,2, and references therein]. The present paper concerns the classical statistical-physics- based Kirchhoff-law–Johnson-noise (KLJN) key distribution system, delineated in Figure 1, which is no exception to the tradition of the research area, and the creation of the KLJN schemes [3,4] immediately triggered attacks [5-7]. The various attacks [5–16] have led to useful discussions [17–23], including corrections of flaws in the attacks [19–23] and developments of new defense protocols [5,10,11,13,24,25] as well as protocols that have increased immunity against attacks in general [24–27]. Furthermore, KLJN schemes that are totally immune to a certain attack have been presented [13,28–30] as has a new system that is immune to all existing attacks [31]. Responses to the attacks have included plain denials of their validity [18,21–23], and in some cases experimental results that purportedly supported an attack have been found flawed [23]. The debates sometimes represent a standoff between opposing parties with different scientific backgrounds, which is a typical feature of science debates on breakthrough results in physics, as observed already by Max Planck [32].
Recently, Gunn, Allison and Abbott (GAA) published an interesting paper [15] with the first attack utilizing transients at the beginning of the bit-exchange. Their idea is impressively simple and involves monitoring the mean-square voltage before the front of the transient reaches the other end of the communication cable. We note that this approach requires a very short sampling time—less than 10% of the correlation time for the noise [14]—and the relative change of the voltage is typically small during this period.
In a simple illustration of the key effect of GAA’s approach, we assume that Eve monitors the voltage on the cable while its capacitance C is charged up by a DC voltage via a resistor R. According to the Johnson–Nyquist formula [3], the voltage noise spectrum can be written as S( f ) = 4kTR —where f is frequency, k is Boltzmann’s constant and T is temperature—which means that the larger resistance has a higher mean-square voltage. Thus the DC voltage scales with R , whereas the RC time constant scales linearly with the resistance and the rate scales inversely with this time constant. If Alice and Bob use no precaution and abruptly switch the resistors (with their generators) to the line, then the mean absolute value of the rate-of-change for cable voltage at the entry point will scale as ∼1/ R . It is also obvious from the above considerations that a linear ramping-up of the noise amplitude is not helpful, at least not if the communicating parties perform the ramping in a symmetrical fashion as in the first experimental demonstration [11] of the KLJN scheme.

L.B. Kish, C.G. Granqvist, Comments on “A New Transient Attack on the Kish Distribution System”

2

FIGURE 1. Outline of the KLJN scheme without defense circuitry [3] against active (invasive) attacks and attacks utilizing non-idealities. The RL and RH resistors, identical pairs at Alice and Bob, represent the Low (L) and High (H) bit-values. The corresponding (band-limited) white noise spectra SL and SH form identical pairs at the two ends, but they belong to independent Gaussian stochastic processes. Both parties are at the same temperature Teff, so the net power flow is zero. The LH and HL bit-situations of Alice and Bob produce identical voltage and current noise spectra, Su and Si, in the wire, implying that they represent a secure bit exchange. The total loop resistance Rloop is publicly

This content is AI-processed based on ArXiv data.