1

Experimental quantum key distribution at a wavelength of ~850nm

V. Fernandez, D. Arroyo, M. J. Garcia, P. A. Hiskett, Robert J. Collins, Gerald S. Buller, AB. Orue

Summary Some basic considerations for the experimental realisation of a quantum key distribution (QKD) system are discussed in this paper. Among them are the wavelength of the photon source and the composition of the transmission channel. In particular, the advantages and disadvantages of using either ~ 850nm or ~ 1300/1550nm for the sources wavelength are analysed along with a comparison of the performance of optical fibre versus free space as the physical medium for the transmission channel. KeywordsQuantum key distribution, quantum cryptography, optical data communications, optical fibre, free space communications, data encryption.

implementation of a free space QKD system for metropolitan area applications is described. Such system is currently being designed by our group with the aim of improving current transmission rates. For this purpose a GHz-clocked photon source in conjunction with an optical synchronisation of Alice and Bob - which imposes no software speed constraints - is proposed. II. EXPERIMENTAL REALISATION OF QKD QKD is the only method that offers a verifiably secure distribution of cryptographic keys between two or more parties that wish to communicate in absolute secrecy. This secrecy is not guaranteed by the conjectured impossibility to reverse certain mathematical functions, such as RSA encryption, but by Quantum Mechanical laws; in particular the Heisenberg Uncertainty Principle and the No Cloning theorem [8]. The latter forbids making a perfect copy of an unknown quantum state, and since each bit of the key is encrypted by a single photon this prevents an eavesdropper to copy it. The Heisenberg Uncertainty Principle (HUP) states there are pairs of variables, known as conjugate variables that have associated quantum observables that do not commute. This means that the measurement of one necessarily disturbs the other. QKD uses pairs of these conjugate variables, such as rectilinear and diagonal bases of polarised light. In the first QKD proposed protocol, the BB84 [4], Alice (the transmitter) sends a random binary sequence to Bob (the receiver) encoded in single photons randomly using four polarisation states:, (which represent the states 0 and 1 using a rectilinear basis +) , (which represent the states 0 and 1 using a diagonal basis ). When a diagonally polarised state (i.e. diagonal basis) is measured using a vertical orientated polariser (i.e. a rectilinear basis) the result is ambiguous, since by the HUP it is impossible to determine whether the state was initially polarised vertically or diagonally. Therefore Bob, with no prior information of the basis set that Alice used, will measure incorrectly in approximately half of the cases. However in the next stage of the protocol Alice tells Bob by a public channel which basis she used to generate each of the photons. Bob keeps only the photons where he used the same basis than Alice and discards the rest. Alice and Bob should now share a sequence of bits. If an eavesdropper (Eve) tries to measure the states, she will necessarily introduce an error, exactly like Bob does. However, when Alice and Bob compare their basis,

I. INTRODUCTION

he realisation of a quantum computer [1][2] could impose a threat on public key cryptosystems [3], which currently protect the distribution of cryptographic keys. One proposed alternative method to distribute keys in a secure manner is quantum key distribution (QKD) [4] where the key is encrypted using quantum properties such as the polarisation or phase of single photons. QKD systems have technologically evolved considerably over the past few years. Practical implementations in both optical fibre and free space have achieved distances as long as 149 km [5] and 144 km [6], respectively. However the bit rates obtained for most of these systems (generally less than Kbits-1) are still far from being competitive with conventional key exchange rates. In addition, there are still no quantum encryption algorithms that could be safely used against a quantum computer attack. Therefore QKD must be used with the only encryption algorithm that has been mathematically proven: the one time pad [7]. Since this algorithm states that the cryptographic key must be as long as the message to ensure secrecy, we must look at ways to considerably improve QKD transmission rates to make this process efficient. Accordingly, some technical considerations affecting this bit rate are discussed in this paper, such as the photon sources wavelength or efficiency and the physical medium composing the transmission channel. Finally the
Veronica Fernandez, David Arroyo, and Maria Jose Garcia are with the Instituto de Fisica Aplicada, Consejo Superior de Investigaciones Cientificas, Serrano 144, 28006 Madrid, Spain e-mail: veronica.fernandez@iec.csic.es. Amalia B. Orue is with the rea de Cultura Cientfica, CSIC, Serrano 117, Madrid, Spain. Phil A. Hiskett is with Selex Galileo, Edinburgh, United Kingdom. Robert J. Collins and Gerald S. Buller are with Heriot-Watt University, Edinburgh, United Kingdom.

Bobs error is eliminated, since they only keep the photons which they both used the same basis. Therefore, if after this process there is still an error in Alice and Bobs sequence they must conclude that it was someone else who introduced it. A quantum channel is used to transmit the quantum states between Alice and Bob and a classical public channel is utilized to hold the public discussion where they exchange the basis set, correct errors and detect the presence of an eavesdropper. The quantum channel is usually composed of optical fibre or free space whereas the public channel can be any insecure standard channel, such as a telephone line. A. Single photon sources In order to keep the information attained by an eavesdropper to the minimum, Alice must send only one photon per bit. This is achieved using a single photon source, which generates an individual photon when required. One of the most promising candidates are quantum dots [9][10], which are excitons (electron-hole pairs) confined in the three spatial dimensions. When the excitons are confined to such small areas their energy levels are quantized. If a single quantum dot contains more than one exciton, they will be distributed in energy states according to the Pauli Exclusion Principle. However, as the electrostatic repulsive force is strong among them due to the reduced area they are embedded in, each of them will occupy a different energy level [9]. Therefore when they recombine they will emit at different wavelengths in such a way that the last exciton to recombine after an excitation pulse takes place, will have a unique wavelength. Therefore by spectrally filtering this last wavelength one can dispose of a continuously generated stream of single photons. However, although considerable progress is been achieved in the performance of single photon sources their efficiency is still far from the desired level. This is partly due to the complexity of achieving the desired density of states when fabricating a layer of quantum dots. In addition the collection of the single photon stream into an optical fibre is not very efficient. Moreover the temperatures necessary to achieve single photon emission are usually of only a few Kelvin, with the associated technological difficulties involved with this. For all of these reasons QKD systems use a more practical alternative: attenuated laser pulses, often called weak coherent pulses (WCP), which follow a Poisson distribution given by:

that 90% of the laser pulses are empty, considerably reducing the transmission rate of a QKD system. B. Advantages of using the first telecommunications window for the photon source of a QKD system There are typically three different wavelengths that are commonly used for the photon source of a QKD system. These correspond to the first, second and third telecommunication windows, located around ~ 800 nm, ~ 1300 nm and ~ 1550 nm wavelengths, respectively. The advantage of choosing the first window relies mainly in the possibility of taking advantage of the more mature technology of Silicon-based single photon detectors. They generally allow for a more efficient detection of single photons at shorter wavelengths and they are commercially off-the-shelf components. This permits the operation at high repetition rates, which in QKD systems with the impossibility of using quantum repeaters to amplify the optical signal due to the No cloning theorem [8] means the only possibility of increasing the key transmission rate. As for the transmission channel, free space is a good candidate when using the first telecommunication window as the wavelengths source, since a low absorption window is situated in the proximity of ~ 850 nm. However, if optical fibres are chosen instead, there are two main options. The first option requires the use of single-mode at ~ 850 nm optical fibres. However these fibres are not standard components, meaning that they are usually more expensive and not always compatible with other commonly-used standard optical components (single mode at ~ 1550 nm). Alternatively P. D. Townsend proposed [11] using standard telecommunication fibre in conjunction with a ~ 850 nm-wavelength source. In this case, Alice and Bobs stations would use single mode components at ~ 850 nm and standard telecommunications fibre for the quantum channel. Of course, standard telecommunication fibre is not single mode at ~ 850 nm, and in fact there are two linear polarisation modes propagating: LP01 and LP11 [12] (see Fig.1). However, after mode control techniques, such as fusion splicing of both single-mode fibres at ~ 850nm and ~ 1550nm, only 0.4% of the photons are launched into the secondary mode LP11 and 99.6% are launched into the fundamental mode LP01. However, a wavelength of ~ 850nm is characterised by an attenuation of ~ 2dB/km in a standard optical fibre. This value is considerably higher than those of ~ 1300 nm and ~ 1550 nm wavelengths, which are 0.35 dB/km and 0.2 dB/km, respectively. So why not use these wavelengths instead of ~ 850 nm? The reason for this is the lack of suitably efficient, single-photon detectors. Previous work has shown that prime candidates for these detectors are Ge and InGaAs/InP single-photon avalanche diode detectors. However they suffer from high dark-count rates, and need to be cooled down to cryogenic temperatures. In addition they show the deleterious effects of afterpulsing [13][14] where an avalanche current can fill mid-gap trap states in the material which emit carriers that subsequently cause further dark avalanches. This

P(n, )

n
n!

(1)

where P is the probability of finding n photons in a given laser pulse attenuated to a mean photon number . Since an attenuated laser follows a Poisson distribution there is always a nonzero probability of finding more than one photon in a given pulse and therefore has to be set to a value where this probability is sufficiently low to allow secure QKD. A mean photon number of ~ 0.1 is often considered a good compromise of security, as less than 0.5% of all emitted laser pulses contain more than 1 photon. However this also means

effect restricts the repetition rates to typically tens of KHz or a few MHz. The improvement and development of these detectors is still being investigated, but the afterpulsing effect in particular remains dominant.

Fig. 1. Mode profile of 850 nm wavelength when propagating in standard telecommunication fibre for two cases: a) before mode control and b) after mode control.

C. GHz-clocked QKD As mentioned in the previous section the use of the mature technology of Silicon-based single photon detectors in conjunction with the first telecommunications window allows the operation of a QKD system at high clock rates, typically of GHz, albeit over shorter distances (<20 km) due to a higher attenuation on standard telecommunications fibre. However this also constitutes an attractive alternative over systems that use ~ 1550 nm, especially for secure communications at an urban-distance range, such as for companies situated in the same city that are linked by short distances and require high transmission rates. Fig. 2 shows typical bit rates measured at Bob with fibre-based QKD systems using the wavelengths under discussion [15][16]. We can see differences of several orders of magnitude in the transmission rates of both systems.

Fig. 2. Comparison of the bit rates achieved for a QKD system that uses a wavelength of ~ 850nm and silicon SPADs [15] and a system that uses a wavelength of ~ 1550nm and InGaAs SPADs [16].

D. Free space or optical fibre as the transmission medium? The transmission channel of a QKD system must preserve the purity of the quantum states. However if we consider optical fibre as the transmission medium there are several phenomena that take place in this medium which degrade the polarization states. These phenomena are due to inhomogeneities associated to any material, either due to its atomic composition or to imperfections in the fabrication process. As a consequence, birefringence, polarisation mode

dispersion or polarisation dependent losses are phenomena that commonly take place and need to be considered in optical transmission by fibre. Birefringence has its origin in the presence of two or more optical axes in a material which causes two different refraction indexes. These axes are due to either the presence of two natural atomic distributions in the material or to external factors such as pressure, temperature, etc. As a consequence an electromagnetic wave will propagate at two different local velocities, which are commonly known as fast and slow axes. This will introduce a phase shift between the components of the light projected along these two axes which will in turn cause a variation in its polarisation state. Therefore polarised light propagating through optical fibre will suffer random variations in its polarisation state which, if not properly compensated, will degrade the quantum states and subsequently cause errors in the transmission. Moreover birefringence is associated to other phenomena that can degrade the performance of a QKD system. Some of them are polarisation mode dispersion and polarisation dependent loss. Due to the presence of two local velocities mentioned previously in the optical fibre, a polarised state will be divided into two orthogonal polarisation modes that will travel at different velocities causing the original pulse to be temporally broadened which will cause errors in the transmission. This ultimately limits the maximum repetition rate o transmission rate over an optical fibre. Another common effect that appears as a consequence of birefringence is polarisation dependent loss. The two orthogonal polarisation modes are differently absorbed in the optical fibre causing the relative angle of the states encoding the binary states of the cryptographic key to differ [17]. Since the security of the BB84 protocol or B92 protocol relies heavily on this angle special care must be paid to compensate for this effect. Moreover optical fibre considerably absorbs certain nearinfrared and infrared wavelengths, especially the first telecommunication window, ~ 850 nm, which is absorbed at a rate of 2.2 dB per km. On the other hand, free space is characterised for being a low birefringent and dispersive medium, besides offering a low-absorption window close to ~ 770 nm, allowing the use of the Si-based single photon detectors, which offers a series of previously mentioned advantages. In addition free-space communications generally offer more flexibility in terms of installation and lower cost than optical fibre. However, the most important advantage that the atmosphere uniquely offers is the possibility of secure global communication. Indeed, through the atmosphere two remote locations could communicate by using a low orbit satellite as secure relay station. Instead, QKD systems that use optical fibre are limited by both the absorption and the impossibility of using quantum repeaters. However it must be noted that free-space QKD systems have some associated problems such as weather dependency or the necessity of having an open channel between sender and receiver. In addition turbulence

present in the atmosphere due to variations in the density of the air can cause fluctuations in the alignment of the laser beam of these systems which need to be compensated for. E. Short distance urban-area free space QKD system Research on free-space QKD systems has been especially focused on increasing the transmission distance of the secure free-space links in locations situated far from urban areas. This is mainly aimed to develop satellite-based secure communications. However, much less attention has been paid in short-distance QKD applications in urban areas. This is probably due to higher contributors to turbulence from industrial equipment (ducts, pipes), machines (internal combustion engine) or from vehicles, which affect the laser beam alignment. In addition the laser light is further attenuated due to a higher absorption by pollutants (see Fig. 3). Despite this inconvenience short-range free-space systems offer several advantages. A satellite can be considered as a third party of a QKD protocol that needs to be established as a secure relay station. This is not a trivial task. Additionally long-distance free-space QKD systems are technologically considerably more challenging, and due to the higher distances involved the transmission bit rates are much lower. Therefore many companies situated in urban areas might find a short-distance free-space QKD system an alternative option since it can offer higher bit rate, lower cost and less complexity in the fabrication than long-distance free-space systems.

respectively. A beam expander will be used to generate a collimated and sufficiently large beam in order to decrease divergence effects. At Bob, a cassegrain telescope will be used to efficiently focus the beam and detect the signal from Alice. In addition a high speed Time Interval Analyser (TIA) with four channels will allow us to analyse the optical signal from Alice and establish the degree of security in the transmission. Especial care must be paid to one of the most critical parts of the system, which is the filtering of the solar background radiation from the sun. For this purpose, a combination of spatial, spectral and software filtering will be used. The timing synchronisation between Alice and Bobs stations will be performed by multiplexing a different wavelength than that used for the quantum states encrypting the key. This will permit faster synchronisation compared to systems using software phase locked loop driven by either the photo detection events received at Bob [18] or by the Global Positioning System (GPS) [6]. This type of synchronisation in conjunction with a GHz-driven source will result in bit transmission rates in the order of MBits-1. To the authors knowledge no QKD system has been built in a city that works at this clock frequency and distance. Current systems for this application are typically clocked at only a few MHz [19], and the secure key transmission rates are in the order of a few Kbits-1 or less. The authors have previously achieved high key transmission rates operating at high clock frequencies over optical fibre in a GHz clocked QKD system [15][17].

Fig. 3. Transmission coefficient T versus wavelength for a horizontal path of 1km for two types of aerosol, calculated using an atmospheric transmission simulation code (Modtran4.0).

Fig. 4. The transmitters station: Alice. ES is an electrical splitter, LD is a laser diode, M1 is a mirror, and D is a photodiode. This example illustrates the BB84 protocol. The black lines represent electrical cables and the grey ones optical paths.

For these reasons were building a short-distance QKD system as that shown in Figs. 4 and 5, capable of transmitting quantum keys securely between two locations in Madrid situated at 3 km of distance at high transmission rates. These two locations will be the Institute of Applied Physics (CSIC) and the telecommunications operator Telefonica. Alice will use a fast GHz pre-programmed pulse pattern generator to produce the cryptographic key that, in conjunction with commercially available Si single photon detectors (SPCMAQR-12) will allow the transmission rates in the order of Mbits-1. The pre-programmed pseudorandom sequence will feed four vertical laser diodes which will generate the four polarisation states required to implement the BB84 protocols,

Fig. 5. The receivers station: Bob. PBS is a polarising beam splitter and D1, D2, D3 and D4 are single photon detectors.

III. CONCLUSIONS In this paper we have discussed some technical considerations affecting the transmission bit rates of a QKD system. Among them were the wavelength and transmission medium, along with the advantages and disadvantages of current long-range free-space QKD systems. Finally a shortdistance high-speed QKD system has been proposed as an alternative option to current systems as it offers the possibility of higher transmission rates and a cheaper and easier-toimplement alternative for urban-span secure communications links. ACKNOWLEDGMENT We would like to thank the Ministerio of Educacin y Ciencia, proyecto TSI2007_62657 and CDTI, Ministerio de Industria, Turismo y Comercio (Spain), in collaboration with Telefnica I+D, Project SEGUR@ with reference CENIT-2007 2004. REFERENCES
[1] P. W. Shor, Algorithms for quantum computation: discrete logarithms and factoring, in Proc. of the Symposium on the Foundations of Computer Science, California, pp. 124134 (IEEE Computer Society Press, New York,1994) L. K. Grover, A fast quantum mechanical algorithm for database search, in Proc of the 28th Annual ACM Symposium on the Theory of Computing (STOC), pp. 212-219, 1996. R. L. Rivest, A. Shamir, and L. Adleman, A method for obtaining digital signatures and public-key cryptosystems, Communications of the ACM, 21, 1978. C. H. Bennett, and C. Brassard, Quantum cryptography: public key distribution and coin tossing, in Proceedings of IEEE International Conference on Computers, Systems, and Signal Processing, Bangalore, India, pp. 175-179 (IEEE, New York,1984) P. A. Hiskett, D. Rosenberg, C. G. Peterson, R. J. Hughes, S. Nam, A. E. Lita, A. J. Miller and J. E. Nordholt, Long-distance quantum key distribution in optical fibre, New Journal of Physics, vol. 8, 193, 2006. T. Schmitt-Manderbach et al. Experimental demonstration of freespace decoy-state quantum key distribution over 144 km, Physical Review Letters, vol. 98, 010504, 2007. G. S. Vernam, Cipher printing telegraph systems for secret wire and radio telegraphic communications, Journal of the American Institute of Electrical Engineers, vol. 45, pp. 109-115, 1926. W. K. Wootters, and W. H. Zurek, A single quantum cannot be cloned, Nature, vol. 299, pp. 802-803,1982. C. Santori, M. Pelton, G. Solomon, Y. Dale, and Y. Yamamoto, Triggered single photons from a quantum dot, Physical Review Letters, vol. 86, pp. 1502-1505, 2001. J. A. Timpson et al. Single photon sources based upon single quantum dots in semiconductor microcavity pillars, Journal of Modern Optics, vol. 54, n.:2-3, pp. 453 -465, 2007. P. D. Townsend, Experimental investigation of the performance limits for first telecommunications-window quantum cryptography systems, IEEE Photonics Technology Letters, vol. 10, pp. 1048-1050, 1998. P. D. Townsend, C. Marand, S. J. D. Phoenix, K. J. Blow, and S. M. Barnett, Secure optical communication systems using quantum cryptography, Phil. Trans. R. Soc. Lon. A, vol. 354, pp. 805817, 1996. P. A. Hiskett, G. Bonfrate, G. S. Buller, and P. D. Townsend, Eighty kilometer transmission experiment using an InGaAs/InP SPAD-based quantum cryptography receiver operating at 1.55m, Journal of Modern Optics, vol. 48, pp. 1957-1966, 2001. P. C. M Owens, J. G. Rarity, P. R. Tapster, D. Knight, and P. D. Townsend, Photon-counting with passively quenched germanium avalanche, Applied Optics, vol. 33, pp. 6895-6904 , 1994.

[15] K.J. Gordon, V. Fernandez, G.S. Buller, I. Rech, S.D. Cova, and P.D. Townsend, Optics Express, vol. 13, pp. 30153020, 2005. [16] C. Gobby et al., Quantum key distribution over 122 km of standard telecom fiber, Applied Physics Letters, vol. 84, n 19, pp. 3762-3764, 2004. [17] V. Fernandez, R.J. Collins, K.J. Gordon, P.D. Townsend, and G.S. Buller: Passive optical network approach to gigahertz-clocked multiser quantum key distribution", IEEE Journal of Quantum Electronics, 43, n 2, pp.130-138, 2007. [18] J.G. Rarity, P.R. Tapster, P.M. Gorman, Secure free-space key exchange to 1.9km and beyond Journal of Modern Optics, vol. 48, 1887, 2001. [19] R. J. Hughes, J. E. Nordholt, D. Derkacs and C. G. Peterson, Practical free-space quantum key distribution in daylight and at night, New Journal of Physics, vol. 4, pp. 43.1-43.14, 2002. Veronica Fernandez received the B.Sc. degree (hons.) in physics with electronics from the University of Seville, Spain, in 2002 and the PhD degree in physics from Heriot-Watt University, Edinburgh, U.K., in 2006. She joined the Scientific Research Council (CSIC), Spain, in 2007. Her areas of research include quantum key distribution and chaotic cryptography. David Arroyo received the M.S. degree in telecommunications engineering from the University of Seville, Spain, in 2002. He is working toward the PhD degree of in physics of complex systems at the Polytechnic University of Madrid, Spain, since 2005. He joined the Scientific Research Council (CSIC), Spain, in 2005. His research interests are in the area of cryptography and chaotic systems. Maria-Jose Garcia-Martinez received the Telecommunications Engineering degree from the Universidad Politcnica de Cartagena (UPCT), Spain, in 2006, and did her graduate thesis while at their Department of Information Technologies and Communications. She is currently working as a grant holder in collaboration with the Scientific Research Council (CSIC). Philip Hiskett received a B.Sc. degree (Hons) in Physics from the University of Leeds, UK, in 1994; an M.Sc. degree in Optoelectronic and Laser Devices from Heriot-Watt University, Edinburgh UK in 1996 and a Ph.D. degree in Physics from Heriot-Watt University in 2000. He has held post-doctoral positions at Heriot-Watt University, working on time-of-flight ranging, single photon-counting technologies and quantum key distribution, and at Los Alamos National Laboratory, USA, working on fibre-optic quantum key distribution systems. He is currently employed as a Senior Systems Engineer at Selex Galileo, Edinburgh, UK. Robert J. Collins received an MPhys. with honors degree (4 years) in physics from the Heriot-Watt University, Edinburgh, UK, in 2003. In 2004 he started a PhD in physics at Heriot-Watt University. His areas of research include quantum key distribution, single-photon sources, random number generation, equipment interfacing, time-of-flight laser ranging, data analysis and software development.Robert is a member of the Institute of Physics (UK) and the Institute of Electrical and Electronics Engineers. Gerald S. Buller (M06) received the BSc degree (with honors) in natural philosophy from the University of Glasgow, Glasgow, UK, in 1986 and the PhD degree in physics from Heriot-Watt University, Edinburgh, UK, in 1989. He was made a professor of physics at Heriot-Watt University in October 2006. Since 1990, he has led the Photon-Counting Group in researching timeresolved photoluminescence, time-of-flight ranging and quantum key distribution. Prof. Buller has coauthored more than 200 journal articles, conference papers, and patent applications in the areas of photon-counting, semiconductor optoelectronics, optical coatings, and optical interconnects. In 2002 Prof. Buller founded Helia Photonics Ltd. of Livingston, Scotland. He is a Fellow of the Institute of Physics (UK) and a member of the Optical Society of America. Amalia Beatriz Orue received the Telecommunications Engineering degree from the Universidad de Oriente, Santiago de Cuba, Cuba (1984) and the DEA from the UPM en 2007. She joined the rea de Cultura Cientfica of the CSIC in 2005. She has been Professor of Circuit Theory at the Universidad de Oriente (1984/2002) and visiting scientist at the CSIC (1999/2000). Her areas

[2]

[3]

[4]

[5]

[6]

[7]

[8] [9]

[10]

[11]

[12]

[13]

[14]

6 of research include cryptography, chaotic systems, circuit theory, knowledge structure and learning theory.