Unconditionally secure communication via wire

Information technology relies on two major information carriers: electrons and photons. The first uses wires and silicon microprocessors with device sizes in the sub-100nm range, and the second carries very high bandwidth data in optical fibers. Yet electrons and photons behave very differently. In addition, practical applications of electrons generally assume the no-wave limit (i.e., current flows in a wire like water flows in a tube). Photons, in contrast, are mostly handled as propagating waves. Interfacing between the two technologies is nontrivial.

The question whether electrical or optical is better is a perennial one. To complicate things, there is no single right answer. For high-speed communication, optics is the hands-down favorite; moreover, photons can carry information even without fiber, through air (when weather and pollution permit) or empty space. When it comes to processing data, however, electricity wins.^1–3 Since the 1970s people have speculated that (classical-physical) optical computers would best supercomputers based on speed. A second, more recent prediction is that for certain tasks, quantum computers—which are based on waves (both electronic and photonic)—will in turn trump optical computers. That said, for general-purpose computing, electrons and holes in silicon chips remain the technology of choice. The issue here is that not only are optical and quantum solutions expensive, but theoretical predictions about their heat dissipation and error rates make these methods unsuited for practical use.^1–3

The subject of this article is an aspect of information technology for which the ‘Electrical or optical?’ question recently became relevant. The problem is unconditionally secure data communications, and up to now, the solution to it was clearly optical. Data communication security is an essential part of today's world. We expect it to be there, functioning, while our computers are connected to the Internet.

Take the example of the software tools we use when we log on to the bank via the Web. Before a secure data exchange can begin, the two communicators (let's call them Alice and Bob) must generate and share a joint secret (secure) encryption key through the communication channel, while an eavesdropper (say, Eve) is supposedly monitoring the related data (see Figure 1). Using current software approaches, however, this task is mathematically impossible. The methods are only ‘computationally safe.’ That is, Eve can decode the data, but it takes too long. If Eve had a genuinely powerful algorithm or a sufficiently fast computer with standard algorithms, she could extract the secure key and decrypt the communicated data with reasonable speed. Because new algorithms and computing solutions are continually being generated, existing software-based secure communication is a potential time bomb.

Figure 1. The two communicators (Alice and Bob) must generate and share a joint secure key through the communication channel while the eavesdropper (Eve) is monitoring the related data. This is an impossible task with existing software methods, which are only ‘computationally safe’ and thus potential time bombs.

So-called quantum key distribution (developed by Stephen Wiesner in the 1970s; Charles H. Bennett and Gilles Brassard in 1984; and Artur Ekert in 1990) proposed a solution that is claimed to be unconditionally secure.^4–6 The information bits are carried by single photons (see Figure 2). In this scheme, security is based on the ‘no-cloning-theorem’ of quantum physics. The idea is that a single photon cannot be copied without noise (error). If Eve captures and measures the photon, it gets destroyed, and she must regenerate and reinject it into the channel. If not, Alice and Bob will consider the bit invalid. Owing to the no-cloning rule, however, while Eve is restoring the bit, she introduces noise, and the error rate in the channel will be greater than without eavesdropping. After analyzing a number of transmitted bits and their errors, Alice and Bob will discover the eavesdropping. Such detective work notwithstanding, no quantum communicator is secure against the advanced type of the man-in-the-middle-attack, where Eve breaks the channel and installs two quantum communicators. She uses one to communicate with Alice, pretending that she is Bob, and the other to communicate with Bob, pretending she is Alice.

Figure 2. Generic quantum communication arrangement. Detecting the eavesdropper requires a statistics of bit errors, which calls for a sufficiently large number of bits. The communication of only a few bits is not secure.

Recently, we proposed an unconditionally secure classical-physical communication scheme, the Kirchhoff-loop-Johnson-(-like)-noise (KLJN) communicator.^7–12 This is a statistical-physics competitor of quantum communicators. It contains two identical pairs of resistors (see Figure 3). The logic-low (L) and logic-high (H) resistors, R₀ and R₁, are randomly selected at the beginning of each clock period and are driven by their own Johnson-like (thermal) noise voltages or alternatively by their electrically enhanced versions with a pre-agreed factor (Johnson-like noise). The practical realizations contain additional elements such as filters and amplitude control units. A secure key bit is generated and exchanged when the resistor values at the two ends differ. The role of the thermal noise is to determine the total resistance in the loop without giving information to Eve about the actual location of R₀ and R₁. When the total loop resistance becomes known (this information is available to the public, too), and when it is the sum of R₀ and R₁, Alice and Bob can calculate the resistance value at the other side since they know their own resistance values. The security of the communicator is based on the robustness of classical information, classical statistical physics, the second law of thermodynamics, and the impossibility of constructing a perpetual motion machine.⁹ It is naturally protected against the man-in-the-middle attack, and active eavesdropping is detected immediately,⁹ much more quickly than the time needed to transfer a single bit.^9,10 Statistics of bit errors are not needed. The communication of even a single bit is secure.

Figure 3. The Kirchhoff-loop-Johnson(-like)-noise (KLJN) wire communication arrangement with two pairs of resistors (R₀ and R₁) and their noise voltage generators. To detect the eavesdropper (represented, e.g., by the current generator in the middle), the instantaneous current and voltage data measured at the two ends is broadcasted and compared. The eavesdropping is detected immediately, within a much shorter period than the time needed to transfer a single bit. Statistics of bit errors are not needed. The communication of even a single bit is secure.

Note that both the quantum and classical claims about unconditional security outlined above refer to idealized systems (at the level of mathematical models). In practical applications, no physical system is ideal, and there are always parasite elements and spurious effects. For this reason, in practical applications, neither the quantum nor the KLJN system is totally secure. Still, knowing their mathematical model makes it possible to maximize security and other performance parameters depending on the physical and financial constraints. The ultimate test of security is necessarily experimental: before marketing, a communicator must pass all the known breaking methods.

Many quantum communicators have reportedly been built, up to a range of 200km. The majority employ optical fibers, and some of the most advanced and are able to communicate through air.⁶ A substantial drawback, however, is that testing different breaking ideas costs more than building the communicators themselves. As a result, quantum security is still mostly theoretical, a fact that is reflected in the plethora of theory-based papers on proposed breaking methods. Considered in terms of percentages, roughly 80% of the experimental testing remains to be done for these quantum devices to be marketable on a large scale.

Figure 4. The KLJN wire communicator network element (communicator pair). Its fidelity is 99.98%, and its security properties are superior to those of existing quantum communicators.

The KLJN idea is young, and so far only one system has been built and tested^8,12 (see Figure 4). We verified it experimentally against all the proposed attack types, and in every case the active eavesdropping was discovered during communication of a single bit. Although the speed of this communicator is half that of the most advanced quantum communicator,⁶ there are straightforward ways to increase the speed in future designs. The device cost a few hundred dollars, and in integrated form we expect its fabrication price to be similar to that of an Ethernet card in a PC. The system is network-ready¹¹ and robust against vibrations, dust, aging, and so on. With proper filters, existing wires such as power lines, phone lines, and Internet lines can be used as channels.

In conclusion, to determine which method is best for totally secure communication, ask yourself whether you can use a wire. If you can, the solution is electrical with a KLJN arrangement. If you cannot use a wire, an optical approach using quantum communication is preferable.