Information about http://w3.antd.nist.gov/pubs/2007/Active-QKD-Network-SPIE2006.pdf
Demonstration of an Active Quantum Key Distribution Network1 …
Tags: bienfang, boisvert, communication partner, cryptographic keys, expandability, fiber link, gaithersburg md, hershman, national institute of standards and technology, network architecture, network nodes, node network, node networks, optical switches, quantum communications, quantum cryptography, speed point, time pad, video signals, video surveillance system,Pages: 6
Language: english
Created: Fri Jul 14 15:13:08 2006
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
Demonstration of an Active Quantum Key Distribution Network1
Xiao Tang, Lijun Ma, Alan Mink, Anastase Nakassis, Hai Xu, Barry Hershman, Joshua Bienfang,
David Su, Ronald F. Boisvert, Charles Clark, and Carl Williams
National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, MD 20899
xiao.tang@nist.gov; alan.mink@nist.gov
ABSTRACT
We previously demonstrated a high speed, point to point, quantum key distribution (QKD) system with polarization
coding over a fiber link, in which the resulting cryptographic keys were used for one-time pad encryption of real time
video signals. In this work, we extend the technology to a three-node active QKD network - one Alice and two Bobs. A
QKD network allows multiple users to generate and share secure quantum keys. In comparison with a passive QKD
network, nodes in an active network can actively select a destination as a communication partner and therefore, its
sifted-key rate can remain at a speed almost as high as that in the point-to-point QKD. We demonstrate our three-node
QKD network in the context of a QKD secured real-time video surveillance system. In principle, the technologies for the
three-node network are extendable to multi-node networks easily. In this paper, we report our experiments, including
the techniques for timing alignment and polarization recovery during switching, and discuss the network architecture and
its expandability to multi-node networks.
Keywords: Quantum cryptography, quantum key distribution, quantum communications network, optical switches.
1. INTRODUCTION
Since Quantum Key Distribution (QKD) was first proposed in 1984 [1], many groups have implemented and
demonstrated QKD systems. A variety of these efforts were reported in the comprehensive review article [2]. In recent
years tremendous efforts have been focused on increasing both the distance between communicating parties as well as
the speed in which cryptographic keys can be established. Most QKD systems reported have been point-to-point links.
Only a few groups, such as BBN [3], have demonstrated quantum networks with three nodes or more.
Phoenix et al. [4] proposed the concept of passive quantum networks using passive optical components. Such a network
adopts passive optical couplers as network nodes to split photons sent by Alice, the transmitter, to more than one Bob,
the receiver. In this architecture, simultaneous communication, or "broadcast", from one user to all others in the network
is established and quantum key distribution between one user to any other one can be realized. Townsend et al. [5]
conceptually demonstrated that quantum key distribution is feasible between any user to any other one within a passive
quantum network. However, in the passive communication network, photons (and hence the bit that it represents) are
split by couplers according to their ratio and hence only a subset of these photons reach each terminal node. So, the
actual key rate between specific terminals is greatly reduced. In comparison, in an active QKD network a network
controller actively controls optical switches to direct the communication direction (orientation). In this case, all photons
emitted by Alice (except those lost in the link) are delivered to the selected terminal yielding the full communication
speed (therefore highest quantum key rate) between two users in the network.
In this work, we demonstrate a high-speed three-node active QKD network with one Alice and two Bobs each connected
to Alice by 1 km fiber links. Two optical switches are actively controlled to alter the connection between Alice and each
1
The identification of any commercial product or trade name does not imply endorsement or recommendation by the National
Institute of Standards and Technology.
of the Bobs. Over 1 Mbps sifted-key rate is generated in either link. Based on this architecture we discuss feasible
schemes for one-to-any and any-to-any active quantum networks as well as their potential applications.
2. SYSTEM CONFIGURATION
We previously reported [6-8] point-to-point polarization encoding QKD systems with protocols of both B92 [9] and
BB84 [1] operated over 1 km of optical fiber. These QKD systems used 850 nm photons in the quantum channels to
match the detection range of silicon avalanche photo diode (APD), and telecom wavelengths (1510 nm and 1590 nm) in
the bi-directional classical channels. The systems were synchronized at a clock rate of 1.25 Gbit/s, with a quantum
channel transmission rate (QCTR) as high as 625 MHz (2 clock periods).
Based on our previous work we have demonstrated an active three-node QKD network as shown schematically in Fig. 1.
Two MEMS optical switches are installed at the Alice side: one at 1550 nm band to redirect the classical channels; the
other at 850 nm band for the quantum channels. The optical switches can be controlled manually or by a computer.
There are two fibers connecting Alice and Bob1: one is an 850 single mode fiber (HI780, 1 km) for the quantum channel
and the other is a standard telecom fiber (SMF28, 1 km) for the classical channel. There are two standard telecom fibers
(SMF28, 1 km for each) used to connect Alice and Bob2: one for the classical channel and the other for the quantum
channel. A short piece (~ 20 cm) of HI780 fiber is fusion spliced at the end of the SMF28 fiber for the quantum channel.
This short piece of HI780 fiber functions as a spatial filter to remove 850 nm higher mode components generated in the
1550 nm fiber. Two different types of programmable polarization controllers (liquid crystal type at Bob 1 and piezo type
at Bob 2) are installed to recover and automatically compensate for polarization evolution caused during transmission in
the fiber loops.
Bob1
Transceiver
Link 1 SMF28
HI780
Alice Optical Fiber Bob1 box
Optical Switch
Transceiver (1550)
Internet
Alice box
Alice PC
Optical Switch
Bob2
(850)
SMF28 Transceiver
SMF28
Link 2
HI780 Bob2 box
Figure 1. Configuration of active three-node QKD secured network
High-speed data handling boards, designed and implemented at NIST, are installed at both Alice and Bobs' computers.
The board at Alice generates pseudo-random data streams to fire corresponding lasers in accordance with the QKD
protocol. The photons from the suitably attenuated lasers are then directed by an optical switch to one of the two optical
paths (Link1 and Link2) and the respective Bob at the end of the path. The board at each Bob collects the detection
events and communicates with Alice via a classical channel to perform the sifting algorithm. Alice and Bob's boards
then send the sifted bit values to their own computers for subsequent error reconciliation and privacy amplification
performed over the classical channel necessary to generate shared secret keys [10-11]. The computers of Alice and Bob
then use the shared secret keys for application-level data encryption necessary to implement secured communications
through an unsecured network.
The system can perform either BB84 or B92 protocol by a straightforward altering of the Alice and Bob boxes. In this
work, we use the B92 protocol for a demonstration.
3. RESULTS AND DISCUSSION
In this experiment we use two optical switches to select a communication destination from Alice to Bob1 or Bob2. As
long as Alice and one of the two Bobs are connected, the point-to-point protocol can be readily applied to the link and
quantum keys are generated and shared by the two terminals. Therefore, the sifted-key rate and quantum bit error rate
(QBER) used to characterize performance for each link are the same as for previously demonstrated point-to-point
communications. However, the time to establish the initial connection between Alice and one of the Bobs, as well as the
time needed to switch the connection from one Bob to another, are also important parameters to characterize the
performance of the three-node network. We measured these parameters in the experiment.
The sifted-key rate R in each link of the QKD network can be estimated by the following equation when the influence of
the APD's dead time is ignored. (The sifted-key rate estimation equation considering APD's dead time has been
discussed elsewhere [12])
R = Lsw L f L o Pd L p (1)
Here the mean photon number is set to 0.1. Lsw is the loss in optical switches, which is about 0.8 dB in our system. Lf
is the loss in the transmission fiber, which is measured to be 2.3 dB/km at 850 nm for the fibers (SMF 28 and HI780)
used in the two links. Lo represents other losses such as bending, coupling and connection losses in the quantum
channel, which is about 3 dB in our case. Pd is the APD's detection efficiency, about 45% (3.5 dB) at 850 nm according
to the manufacturer's specification. Lp is the protocol related loss, which is 3 dB for BB84 since half of the counts are
discarded in the protocol and 6 dB for B92 since all incompatible photons and half of the compatible photons are
blocked before entering the detector. denotes the quantum channel transmission rate (QCTR). In our experiment we
use a QCTR of 625.0 Mbit/s, which corresponds to Alice generating one optical pulse in every 2 clock periods. The
measured sifted-key rates in the two links are 1.71 Mbps and 1.55 Mbps, respectively, which is in good agreement with
the value estimated by Eq. (1). Since standard telecom fiber is used in the second link, some photons emitted by Alice
are shifted from the fundamental mode into a higher-order mode during transmission and subsequently filtered at the end
of the link. That causes a lower sifted-key rate in the second link than that of the first. In comparison with our point-to-
point QKD system, the sifted-key rate in the network is approximately 0.8 dB lower due to loss resulting from the
insertion of optical switches.
For a point-to point QKD system, the QBER is mainly due to polarization leakage or imperfect extinction ratio, timing
jitter, data-dependant jitter and dark counts as discussed in [8]. In this work, the QBER measured in the first and second
links are 2.9% and 3.3%, respectively. Although the most of the high-order mode photons are filtered by the sliced
HI780 fiber, the presence of these photons contributes to the slightly higher QBER in the second link. Nevertheless, the
values of QBER measured in both links are very close to the QBER in our point-to-point QKD system, which indicates
that the optical switches are stable and do not contribute significantly more errors to the links in the QKD network
system.
In an active QKD network, we define a connection time, which is time between when the optical switch receives the
switch signal for connecting a link to when the link can generate secret keys. The connection time is an important metric
for active networks. After switching, Alice and the Bob need to do polarization recovery and software initialization
before the two terminals can generate secret keys. The connection time Tct consists of the following three parts:
Tct = Tsw + T pr + Tsi (2)
Here Tsw is the switching time of the optical switch, which is less than 1 ms in our case. Tpr is the time for performing
polarization recovery. In our system, the time for polarization recovery is relatively long, which are 360 sec and 480 sec
for Link 1 and Link 2, respectively. In Link 1 we used a liquid crystal type of polarization controller that has a relatively
long response time. In Link 2 we used a piezo type of polarization controller. Although the latter's response time is much
faster, its interface communication speed with the PC is slow. Tsi is the software initialization time, which includes raw
key sifting, error reconciliation, privacy amplification and initial key buffering. Tsi depends on the speed of both Alice
and Bob's computers as well as the key rate. In this experiment, it is about 60-120 sec. So, polarization recovery is the
most time-consuming process in connection. We are currently undertaking research in order to reduce polarization
recovery time [13].
4. ACTIVE QKD NETWORK AND ITS POTENTIAL APPLICATIONS
The architecture of our three-node QKD network can be extended to one-to-any and any-to-any communication within
the network. The one-to-any structure (one Alice to many Bobs) is shown in Fig. 2(a). Optical switches control the
communication destinations. The one-to-any structure is natural extension of our three-node architecture obtained by
replacing the 1 x 2 optical switch with 1 x n optical switch. The system configuration and system performance, such as
sifted-key rate, QBER and connection time, would be similar to the experimental system described above. The general
structure can also implement quantum key distribution between many Alices and one Bob. For an any-to-any
architecture, optical cross connect (OXC) switches are used to implement communications between any Alice and any
Bob as shown in Fig. 2(b). It has been reported [14] that some advanced OXCs, such as MEMS devices have been
constructed to support 1024 x 1024 optical channels, which paves the way to implementing any-to-any network based on
this architecture. One new issue in any-to-any networks is that a Bob may communicate with many Alices that may be
at a different distances, requiring Bob to conduct a time alignment procedure after each switching operation, thus adding
to the connection time. We have developed automatic time-alignment software to implement a fast time alignment.
Bob 1 Alice1 Bob 1
Bob 2 Alice 2 Bob 2
...... ...... ......
Alice OXC
Bob n Alice n Bob n
(a) (b)
Figure 2. Multiple user architecture for active QKD networks: (a) one-to-any network, and (b) any-to-any network.
The active QKD network topology can be a branch configuration or a loop configuration as shown in Fig. 3. In a branch
configuration, a provider is one or many Alices, and every node is an optical switch or an OXC to direct the connection
orientation to certain users (Bobs). In a loop configuration, a provider (one or many Alices) provides an information
loop, and users (Bobs) can access the loop by optical switches or OXC. Whether in branch or loop configuration, any
user can implement communication with a certain Alice in the providers after the optical switches or OXC are correctly
set.
User 1
Provider Provider
User 2
User n User 1
User 3
......
User 4 ...... User 2
(a) (b)
Figure 3. Potential topography of active QKD networks: (a) a branch-configuration, and (b) a loop-configuration.
In any active QKD network, whether one-to-any or any-to-any, or branch configuration or loop configuration, a QKD
network management system is needed. The management system responds to requests of terminals of the network, and
sets optical switches or the OXC to establish the corresponding connection in the network.
The above mentioned one-to-any and any-to-any structures can implement quantum key distribution between one Alice
to one of many Bobs (or one of many Alices to one Bob) and one of many Alices to one of many Bobs. However, it can
not implement simultaneous quantum key distribution between two Alices or two Bobs. In order to implement quantum
key distribution between any two network nodes, a full-function QKD terminal, or duplex QKD terminal, which has
both Alice and Bob functions, must be available at each node in the network. In some specific applications, however,
some nodes in a network need not communicate between each other, and hence a one-function QKD terminal, or
simplex QKD terminal (Alice or Bob), can be used to reduce cost.
As described above, an active QKD network can generate secure keys without any significant degradation compared to a
point-to-point system. For such a network, a wide range of potential applications can be realized in Local Area
Networks (LAN). One example of an important and feasible applications is a QKD secured video surveillance network,
in which a one-to-any network structure is used as shown in Fig. 4. Each Bob at different locations is equipped with a
monitoring video camera, while Alice is installed at surveillance station. Alice can control the optical switches to
connect to any Bob terminal and initiate quantum key distribution. Once the secret quantum keys are generated and
shared between Alice and Bob, the video content taken by the monitoring camera at the Bob can be encrypted using the
secret key bits and sent to Alice over an unsecured public network, such as the Internet. Alice can then decrypt the
transmitted data to view the video. The speed of our system enables the real-time key exchange sufficient to support one-
time pad encryption/decryption of streaming video. We are developing such a two-camera surveillance system based on
the three-node QKD network.
Figure 4. QKD secured network application: secured surveillance system.
5. CONCLUSION
We have demonstrated an active three-node QKD network, which generates sifted-key at a rate over 1 Mbps between
terminals. Two optical switches actively control its communication orientation. This network architecture can be
extended to allow one-to-any and any-to-any quantum key distribution in optical networks. As an example of its
potential applications, a QKD secured surveillance network is also discussed.
ACKNOWLEDGEMENT
This work was supported in part by the Defense Advanced Research Projects Agency under the DARPA QuIST
program.
REFERENCES
1. C. H. Bennett and G. Brassard, "Quantum cryptography: Public key distribution and coin tossing" in Proc. of the
IEEE Int. Conf. on Computers, Systems & Signal Processing, pp. 175-179, Bangalore, India, December 10-12,
(1984).
2. N. Gisin G. Ribordy, W. Tittel, and H. Zbinden, "Quantum cryptography," Rev. Mod. Phys. 74, 145-195 (2002).
3. Chip Elliott, "Building the quantum network", New Journal of Physics, Vol. 4, page 46.1-46.12 (2002).
4. S. Phoenix, S. Barnett, P. Townsend, and K. Blow, "Multi-user quantum cryptography on optical networks", Journal
of Modern Optics, 72 (6): 1155-1163 (1995).
5. P. Townsend, S. Phonenix, K. Blow, and S. Barnett, "Design of quantum cryptography systems for passive optical
networks", Electronics Letters, 30(22), 1875-1877 (1994).
6. Xiao Tang, Lijun Ma, Alan Mink, Anastase Nakassis, Barry Hershman, Joshua Bienfang, Ronald F. Boisvert,
Charles Clark, and Carl Williams, "High Speed Fiber-Based Quantum Key Distribution using Polarization
Encoding," in Optics and Photonics 2005: Quantum Communications and Quantum Imaging III, Proc. SPIE 5893,
1A-1-1A-9 (2005).
7. Xiao Tang, Lijun Ma, Alan Mink, Anastase Nakassis, Hai Xu, Barry Hershman, Joshua Bienfang, David Su, Ronald
F. Boisvert, Charles Clark, and Carl Williams, "Quantum Key Distribution system operating at sifted key-rate over 4
Mbit/s", Defense and Security 06, Proc. SPIE 6244, 62440P-1~ 62440P-8 (2006).
8. Xiao Tang, Lijun Ma, Alan Mink, Anastase Nakassis, Hai Xu, Barry Hershman, Joshua Bienfang, David Su, Ronald
F. Boisvert, Charles Clark, and Carl Williams, "Experimental study of high speed polarization-coding quantum key
distribution with sifted-key rates over Mbit/s", Optics Express, Vol. 14 (6): 2062-2070 (2006).
9. C. H. Bennett, "Quantum cryptography using any two nonorthogonal states," Phys. Rev. Lett., 68, 3121-3124 (1992)
10. A. Nakassis, J. Bienfang, and C. Williams, "Expeditious reconciliation for practical quantum key distribution," in
Defense and Security Symposium: Quantum Information and Computation II, Proc. SPIE 5436, 28-35 (2004)
11. Alan Mink, Xiao Tang, Lijun Ma, Tassos Nakassis, Barry Hershman, Joshua C. Bienfang, David Su, Ron Boisvert,
Charles W. Clark, and Carl J. Williams, "High Speed Quantum Key Distribution System Supports One-Time Pad
Encryption of Real-Time Video", Defense and Security 06, Proc. SPIE 6244, 62440M-1~62440M-7 (2006)
12. Hai Xu, Lijun Ma, Joshua Bienfang, and Xiao Tang, "Influence of the dead time of avalanche photodiode on high-
speed quantum-key distribution system", CLEO/QELS 2006, CLEO digest JTuH3, May 2006.
13. Lijun Ma, Hai Xu and Xiao Tang, "Polarization recovery and auto-compensation," in Optics and Photonics 2006,
San Diego, Aug.13~17.
14. T. Bifano, " Optical Cross Connect Switches for Communication Network",
http://mle2.bu.edu/mn500/pdf/Class7&8.pdf