Quantum key distribution in optical transport networks

Features of multiplexing quantum and information channels of an optical transport network

Author – Fedor Kiselev

Senior Researcher at SMARTS-Quanttelecom LLC

In today’s world, where information has become the most valuable resource, the security of data transmission becomes critical. Asymmetric cryptography methods, on which most modern security systems are based, may be subject to new threats associated with the development of quantum computers.

At this time, quantum cryptography comes onto the scene, promising a revolution in the field of information security. Quantum cryptography uses the laws of quantum physics to create completely secure methods of transmitting information. At the center of this technology is a process known as quantum key distribution (QKD). This process allows two parties to form correlated bit sequences, which can later be used by the parties as symmetric cryptographic keys. Moreover, no attacker will be able to intercept or copy this key without detection.

Thanks to the unique properties of quantum particles, such as photons, which change their state when attempting to measure or copy, thereby signaling an attempt to interfere.

Thus, quantum key distribution offers a solution that is resistant to attacks even using quantum computers. Moreover, even if a full-fledged quantum computer never appears, the technology will be beneficial in terms of automating the process of generating and distributing symmetric keys, thereby eliminating the human factor from it.

As a rule, the transfer of quantum states in the process of quantum key distribution between two remote subscribers occurs over the so-called “dark” fiber. This is a fiber optic communication channel where there are no channels or optical signals other than quantum. This is due to the fact that any illumination will make an additional contribution to the quantum error rate, which in turn will lead to a drop in the rate of secret key generation or even to the complete impossibility of key distribution due to exceeding the distillation threshold for the quantum error rate.

However, this approach leads to additional operating and capital costs when creating quantum networks, since it is necessary to lay new “dark” fibers under quantum channels, or to separate them in already laid cables. Due to the rapid growth of traffic transmitted over optical transport networks, ensuring the possibility of simultaneous distribution of quantum and information channels is becoming an increasingly urgent task.

Let’s take a closer look at what’s stopping us from combining quantum and information channels in one optical fiber?

Typically, the quantum channel is placed in the C-band (in particular, at a wavelength of 1550 nm), since it is in this range that the optical fiber introduces the least attenuation into the transmitted signal (approximately 0.2 dB/km).

Table 1 – Symbols and names of ranges of the telecommunication window of the optical fiber.

As numerous studies show, the main influence on a quantum channel, when combined with information channels using DWDM (Dense Wavelength Divison Multiplexing) technology, is exerted by optical noise caused by nonlinear interactions of the electromagnetic field of information channels with optical fiber, as well as linear crosstalk caused by non-ideality filters and demultiplexers.

The largest contribution among all types of noise comes from noise from spontaneous Raman scattering (SRS). This is the effect of spontaneous inelastic scattering of light on the phonons of an optical fiber, which leads to the appearance of a broadband spectrum.

.As can be seen from Figure 2, this noise fills the entire C-band, which makes it impossible to completely get rid of this noise in this range.

The main method of combating optical noise arising from the presence of information channels in the same range as the quantum one is to reduce the optical power of the signals. The threshold to which the power can be reduced is determined by the characteristics of the communication channel, the sensitivity of the detector of the data transceiver module, as well as the threshold value of the optical signal-to-noise ratio (OSNR), at which error correction algorithms can still work at a level that satisfies the telecommunications standard.

There are several other strategies to reduce the impact of this noise.

Firstly, you can use spectral filters with a narrow bandwidth. Very often, QKD systems use filters on fiber Bragg gratings (FBGs), which can provide a bandwidth of less than 10 GHz, instead of the usual 100 GHz for DWDM technology. Thus, we can reduce the influence of noise from SRS by approximately 10 times, significantly limiting its power in the bandwidth of the quantum channel.

Secondly, you can gate the single photon detector (SPD). This strategy is similar in concept to spectral filtering – the shorter the gate time, the less noise from the SRS gets to the DOP.

Finally, it is possible to optimize the spectral arrangement of information and quantum channels. In Figure 2, the SRS noise spectrum is unevenly distributed, which means that for each set of information channels in a standard DWDM frequency grid there will be an optimal location for a quantum channel with minimal noise.

This strategy also helps control the noise from four-wave mixing (FWM) that occurs when a channel is placed in a uniform frequency grid. When optimizing, you can use well-known algorithms that solve combinatorial problems well; for example, in our work we used the simulated annealing algorithm.

Unfortunately, even the combined use of all these strategies limits the possibility of simultaneous propagation of quantum and information channels in a single optical fiber in the C-band telecommunications window. Backbone optical transport networks, as a rule, are loaded with a large number of information channels, the signal strength of which must be maintained in order to transmit information over a greater distance.

In such conditions, the only option left is to move the wavelength of the quantum channel to 1310 nm.

Among the disadvantages, optical losses in a fiber optic communication channel at a given wavelength are approximately twice as high as losses in the C-band, which also approximately halves the potential range of such a system. However, the noise from SRS – the main source of nonlinear noise in the channel – is reduced by 2-3 orders of magnitude, and other nonlinear noise completely disappears. This creates a situation in which the 1310 nm quantum channel operates as if in a dark fiber.

Figure 5 shows a theoretical calculation of the key generation rate of a quantum sideband key distribution (QSKD) system with a quantum channel at 1310 nm and 1550 nm, in dark fibers and in the presence of 40 information channels located in a standard DWDM frequency grid with an inter-channel distance of 100 GHz.

The method of combining quantum and information channels in one optical fiber, based on the location of the quantum channel in the O-band, is actively used in China. The works of Chinese scientists Mao and Wang, published in 2017 and 2018, respectively, show the successful operation of QKD systems built on a protocol with trap states in backbone quantum networks at distances of over 60 km.

In addition to dealing with nonlinear noise, it is also necessary to take into account the design features of optical transport networks. One of the key elements in optical networks is the optical amplifier. In telecommunications applications there are 3 types:

Erbium amplifier (EDFA) – works based on the process of stimulated emission of radiation, which occurs when electrons in erbium atoms transition from higher energy levels to lower ones. When an optical signal passes through the active medium of the amplifier, erbium ions interact with this signal, resulting in an increase in its intensity.

Raman amplifier – works based on the phenomenon of Raman scattering, which occurs when light interacts with the molecules of an optical fiber (usually the fiber communication channel itself acts as the amplifying medium). When an optical signal passes through an optical fiber, some of the photons interact with the molecules of the medium, undergoing the process of inelastic scattering; as a result, in the presence of the signal, the process of energy transfer from the pump field to the optical signal occurs.

Semiconductor Optical Amplifier (SOA) – works by stimulated emission from a semiconductor material. It increases the intensity of an optical signal by exciting electrons in a semiconductor, which then emit additional photons in the same frequency spectrum as the input signal, amplifying it. SOA is used to amplify the signal level before entering the optical fiber and supports all signal formats with a wavelength of 1310 nm.

Erbium amplifiers are used in a variety of situations and on the main line, they can be used as booster, preamplifiers or built-in amplifiers. Each such amplifier is a source of additional Amplified Spontaneous Emission (ASE) noise, which affects the OSNR of the data channels and can also be a problem for the quantum channel. To completely avoid the influence of this noise, it is recommended to use EDFAs only as preamplifiers after the quantum channel has been dropped from the Add/Drop line by the multiplexer. SOAs are most often used as preamplifiers, so they can also be suitable for these applications.

In the case of Raman amplifiers, the situation is much more complicated, since the active medium is the fiber optic communication channel itself, and the amplification process does not occur continuously over the entire length of the fiber optic channel. As a result, spontaneous Raman noise from a high-power pump (typically located at a wavelength of 1480 nm) is present throughout the entire communication link and significantly affects the operation of a quantum channel, even located at a wavelength of 1310 nm.

Finally, to successfully combine the quantum and information channels, the correct filters and multiplexers must be used, and used in the correct sequence.

To ensure the least losses for a quantum channel in the network, it is necessary, if possible, to enter the quantum channel last into the line and exit it first. This way, it will be possible to reduce the number of optical connections along the path of the quantum channel and, thereby, minimize its losses. At the same time, to ensure the necessary isolation of a quantum channel, it is often necessary to use a cascade of filters (Figure 7) to achieve high extinction values ​​of information channels (more than 100 dB).

The technology of combining quantum and information channels in one optical fiber, although not a panacea for solving complex technical and economic problems when building quantum key distribution networks, is a major step towards the development of this industry in the Russian Federation and the world.

It expands the design capabilities of network architectures, especially in locations where separating dark fiber or installing new fiber is associated with high capital or operating costs. This issue is quite relevant both when building trunk communication lines and when creating quantum networks in cities and industrial facilities.