Академический Документы
Профессиональный Документы
Культура Документы
Josephson junction
Chandan Kumar,∗ Jaskaran Singh,† and Arvind‡
Department of physical sciences, Indian Institute of Science Education and Research (IISER),
Mohali, Sector 81 SAS Nagar, Manauli PO 140306, Punjab, India
We present a quantum key distribution (QKD) protocol based on long lived coherent states
prepared on superconducting rings with a mesoscopic Josephson junction (dc-SQUIDs). This enables
storage of the prepared states for long durations before actually performing the key distribution. Our
on-demand QKD protocol is closely related to the coherent state based continuous variable quantum
key distribution protocol. A detailed analysis of preparation, evolution and different measurement
schemes that are required to be implemented on dc-SQUIDs to carry out the QKD is provided. We
present two variants of the protocol, one requiring time stamping of states and offering a higher
arXiv:1808.06471v1 [quant-ph] 20 Aug 2018
key rate and the other without time stamping and a lower key rate. This is a step towards having
non-photon based QKD protocols which will be eventually desirable as photon states cannot be
stored for long and therefore the key distribution has to be implemented immediately after photon
exchange has occurred. Our protocol offers an innovative scheme to perform QKD and can be
realized using current experimental techniques.
II. BACKGROUND
Consider a superconducting ring of inductance L with In the rotating wave approximation and neglecting all
a Josephson junction with capacitance C and external the terms without annihilation and creation operators,
inductive coupling through an external flux φx , as shown Eqn. (9) can be re-written as,
in Fig. 1. The quantum Hamiltonian for the junction can
be written as, H = Ωb† b − µ(b + b† ) − ν(b† b)2 , (12)
Q2 (Φ0 − φx )2 where
H= + + EJ (1 − cos θ), (5)
2C 2L 2EJ e4 φ
ν= , µ= √x , Ω = ω − ν. (13)
where we have taken h̄ = kB = c = 1. The quantity Q is 3(ωC)2 L 2ωC
the charge operator across the junction and Φ0 is the op-
erator corresponding to the total flux through the ring. The second term in the Hamiltonian (12) depends on the
The quantity EJ is the Josephson coupling constant and external driving flux φx , which can be switched on for
θ is the phase difference of the superconducting wave- a short duration τ1 when desired. The strength of the
function across the junction. The phase difference θ is driving flux is appropriately chosen such that µ Ω, ν
related to the total flux Φ0 by θ = 2eΦ0 . The quantum for the duration when it is turned on. We further impose
mechanical operators, Q and Φ0 form a canonically con- the experimentally achievable condition Ω ν and for
jugate pair of variables with the canonical commutation the remainder of the paper work in the regime µ Ω
relationship ν [23, 24].
For the duration τ1 when the driving field is turned on,
[Φ0 , Q] = i. (6) only the second term of the Hamiltonian (12) is relevant
and it effectively generates a phase space displacement of
The commutation relation can be rewritten in terms of the ground state and its corresponding unitary operator
voltage across the junction, V 0 = Q/C as, can be written as
C [Φ0 , V 0 ] = i. (7)
†
(ΔX)2
indefinitely under no-dissipation conditions. In between,
0.5
we also see superposition of two coherent states.
0.4
C. Preparation of coherent states on a dc-SQUID
(b)
Alice can prepare the chosen coherent state |φA + ivA i
(ΔΦ)2 (ΔV)2
on a dc-SQUID by first preparing it in the ground state 0.27
|0i. This can be achieved by allowing the SQUID to de-
cohere to its ground state in the low Josephson coupling 0.26
limit [30]. By appropriately choosing the magnitude of
the driving field φx , Alice applies the displacement oper-
ator for a short duration τ1 , given by Eqn. (14), on the 0.25
ground state to get:
0 0.5 1 1.5 2
φA νt/π
D |0i = |0 + iφA i. (17)
µ1
FIG. 2. (a) Solid line represents the variance of Φ quadrature
Since µ1 is a real quantity, the state is displaced only while the dashed line represents variance of V quadrature.
along the V quadrature. The driving field is then The condition for squeezing in quadrature X is h(∆X)2 i <
switched off for a time τ2 such that Ωτ2 = π/2 which 1/2. It can be seen that while one quadrature is squeezed,
results in rotating the quadratures of the coherent state its conjugate is de-squeezed.(b) For state (22) the product of
by an angle π/2: the variance is always greater than 1/4 except when the state
π revives to a coherent state. To clearly show squeezing and
desqueezing of quadratures, we have taken Ω = 5ν, φA = 0.3
R |0 + iφA i = |φA + 0i. (18)
2Ω and vA = 0.3.
V
0
Φ
0.7
C D
(ΔΦ)2
0.5
A B V 0.3 (V, Φ)
0 0.5 1 1.5 2
νt/π
(a) (b) (c)
FIG. 4. (a) Preparation of the desired coherent state, by displacing the ground state from A to B followed by a rotation to C
and a subsequent displacement to D. (b) Collapse and revival of the state in the Φ quadrature when kept in storage for long
durations. We have taken values Ω = 20ν for the purpose of clarity and φA = 0.3, vA = 0. (c) Measurement of the state by
Bob by randomly choosing a quadrature from either V or Φ.
corresponding to Bob’s measurement quadrature. rate that can be achieved when Bob measures each dc-
The correlated data is thus generated. SQUID at a random time. To that end, we define a
quantity CAB (X) which quantifies the noise observed by
S6: The correlated data is transformed into errorless Bob given the information encoded by Alice when a mea-
bit strings by using sliced reconciliation protocols surement of X quadrature is performed,
detailed in [8, 14]. Finally they perform privacy
amplification to distill a secure key. Fig. 4 illus- CAB (X) = h(X B − X A )2 i
trates the protocol schematically. (30)
= h(X B )2 i + (X A )2 − 2X A hX B i,
In the step S4 above Bob can employ two different mea-
surements schemes, one in which he does not care about where the average is taken over the measurement results
the precise time for which storage was done and begins of Bob, and X A and X B denote the information encoded
to measure the dc-SQUID quadratures at a time of his by Alice and measurement result of Bob in the X quadra-
choice and the other in which time stamping is used ture.
where when he begins measurements he uses a specific If the state prepared by Alice is |αi upon which Bob
time for each dc-SQUID to maximize his correlation with performs a measurement at a random time t, the noise
Alice. For both the cases we find out the average corre- CAB (X) in the two quadratures is found to be
lations that can be shared between Alice and Bob. As
we shall see it is possible to carry out QKD in both the 1h 2
cases while the measurement scheme with specific time CAB (Φ)= 1 + 2|α|2 + 4 (Re (α))
2
measurements has much higher key rate. h −1
i
+ α∗ e(|α| (γ −1)+2itζ )
2 2
−4Re (α) e−itζ αe|α| (γ−1)
−2
i
+α2 e|α| (γ −1)−2itξ − β 2 e(|α| (γ −1)+2itξ) , (31)
2 2 2
P(Φ)
0.2
1.5
0.1
1
0.5 - 10 -5 0 5 10
2.5 (b)
Φ
CAB { α〉} (X)
2
FIG. 6. The probability distribution of the state (25) when
Alice Gaussian distribution is centered about q0 = p0 = 4 and
1.5
Ω = 100ν.
1
Case 2: Measurements at specific times
0.5
0 0.5 1 1.5 2
In this scheme, upon the need for QKD, Bob performs
νt/π the measurements of φ or V at specific times chosen to
maximize his correlation with Alice. These times corre-
FIG. 5. (a) Represents the error in either quadrature for spond to the times when the state (22) reverts to a pure
Ω = 5ν. When Ω/ν is odd, the state (22) at time t = 0, coherent state (|αi or | − αi) or an equal superposition of
π/ν and 2π/ν is |αi and hence, noise in each case is equal to them as given by Eqn. (25). The time scale of oscillations
1/2. (b) Represents the error in either quadrature for Ω = 6ν. of states are expected to be much smaller compared to
When Ω/ν is even, the state (22) at time t = 0, and 2π/ν is the long storage time. This time keeping has to be done
|αi and noise in both the cases is equal to 1/2. However, at
for each dc-SQUID and therefore we assume that Alice
time π/ν, the state is | − αi and hence, the noise is maximum.
does precise time stamping for each dc-SQUID. There
has to be synchronization of very precise clocks between
Alice and Bob in order to decide on the measurement
Alice is then given as
times.
{|αi}
Z ∞Z ∞
1 2 2 For this scheme, Alice chooses her Gaussian distribu-
CAB (X) = dx1 dx2 e−(x1 +x2 ) CAB (X) tion with variance VA N0 = 1/2, but centered around a
−∞ −∞ π
large value, e.g. q0 = p0 = 4. Bob also utilizes a slightly
3 9 cos(t(ν − Ω)) − 6 cos(t(ν + Ω)) + cos(t(3ν + Ω)) different scheme for measurements. Apart from measur-
= − ,
2 (5 − 3 cos(2νt))2 ing at specified times, he takes the absolute value of the
(33) outcomes he observes. For both the cases when Ω/ν is an
odd or an even integer and Bob records only the absolute
where X can be either of the quadratures and the ex- values of his outcomes, the statistics as observed by him
pression is same for both the quadratures. Fig. 5 shows for states at time t = 0, t = π/2ν, t = π/ν, t = 3π/2ν
{|αi}
the plots of CAB (X) for odd and even values of Ω/ν. and t = 2π/ν will have the same mean and variance as
Finally, under the approximation Ω ν, we average Alice’s. As is shown in Fig. 6 he has a bimodal distri-
the noise over the time period t = 0 to t = 2π/ν to get bution for the times t = π/2ν and t = 3π/2ν which gets
folded onto the positive side when we take the absolute
Z 2π value. For the rest of the times he gets the same statistics
T ν ν
{|αi} 3
CAB (X) = dt CAB (X) = , (34) as Alice’s state |αi after taking the absolute value. This
2π 0 2
is the reason that Alice needs to generate states with
and the average correlation between Alice and Bob I(A : large displacement on her side for this protocol to work.
B) can be evaluated as The average noise (34) for this measurement scheme by
Bob at these specific time intervals is only the vacuum
noise N0 = 1/2. The average correlations can then be
1 V A N0
I(A : B) = log2 1 + T . (35) computed as
2 CAB (X)
1 VA N0
For VA N0 = 1/2 and CAB T
(X) as given by Eqn. (34) the I(A : B) = log2 1 + = 0.50 bits. (36)
2 N0
average correlation comes out to be I(A : B) = 0.20 bits.
The protocol thus achieves its goal of generating data As has been emphasized, in order to achieve the afore-
that can be converted into a key by reconcilliation and mentioned correlations, it is essential to perform mea-
privacy amplification. surements at the precisely specified times. Since Bob
8
T
will be storing these SQUIDs for a long time, it is also for secure key rate can be found by putting CAB (X) =
required to keep an accurate track of time for each and N0 = 1/2 in Eqn. (37). For this case we again arrive
every SQUID. From an experimental point of view we at the condition χ < 1. Therefore in both the cases,
can assume that the collapse and revival time period lies observation of external noise χ ≥ 1 implies that either
in the range of a few hundred microseconds (10−4 s) cor- Eve is present or there has been too much noise and the
responding to Ω ≈ 104 Hz and Ω/ν = 100. Commer- key cannot be distilled.
cially available atomic clocks can keep track of time upto V. CONCLUSION
an error of 0.1µs per day amounting to 1000 days un-
til the error becomes significant. Furthermore, with the In this paper we explored the possibility of carrying out
current computer processors in the range of GHz, it is QKD with a non-photonic quantum system. The proto-
possible to perform precise measurements with a resolu- col involves preparation of coherent states on a dc-SQUID
tion of nanoseconds. Therefore the correlations (36) are by Alice, their transportation to Bob’s location, subse-
achievable with the current technology for storage times quent storage under no dissipation conditions [26, 27]
of approximately 30 months. in Bob’s lab, and finally quadrature measurements and
classical information processing between Alice and Bob.
We exploited the longevity of superconducting coherent
IV. SECURITY states to perform on-demand secure quantum key dis-
tribution where Bob stores the dc-SQUIDS prepared in
coherent states and carries out measurements only when
We analyze security of the protocol for both the cases the key is required. Given the Hamiltonian of the dc-
(with and without time stamping) under an assumption SQUID, during the storage period the correlations be-
of individual attacks by an eavesdropper, Eve, where she tween Alice and Bob undergo collapses and revivals. This
ends up introducing Gaussian noise in the dc-SQUIDs. motivated us to design two variants of the protocol, one
The security proof of the protocol is quite similar to without time stamping and other with time stamping.
the one for continuous variable QKD based on coherent While the protocol with time stamping gives a higher key
states [14]. We assume that Eve has gained access to the rate of 0.5 bits, it is more difficult to carry out. What
ensemble of SQUIDs prepared by Alice while being trans- is noteworthy is that even the simpler protocol, where
ported through a channel with Gaussian noise. Accord- Alice and Bob do not perform any time stamping, has a
ing to a result in [29] based on the no-cloning principle reasonable key rate of 0.2 bits. Our protocol enjoys sev-
and incompatibility of measurements, if the noise added eral advantages over standard photon based QKD proto-
on Bob’s side is χN0 , then the minimum added noise on cols, the foremost being the storage possibility wherein
Eve’s side is χ−1 N0 . In the presence of this noise, for Alice and Bob can introduce a delay of years between
the scheme where Bob is performing measurements at the exchange of states and key generation which can be
arbitrary times, the secure key rate can be found out as done on demand. Secondly, the on-demand QKD proto-
col acts as a bridge between SQUIDs, which have found
∆I = I(A : B) − I(A : E) extensive application in quantum information process-
T ing in recent times and quantum communication [33–35].
1 (VA + χ) N0 + CAB (Xi )
= log2 T (X) This opens up several new and interesting avenues of re-
2 χN0 + CAB (37)
T
search in application of other condensed matter concepts
1 (1 + VA χ) N0 + CAE (X)χ to quantum communication. In particular it would be in-
− log2 T (X)χ + N
,
2 CAE 0 teresting to see how entangled states of SQUIDs can be
prepared for application in entanglement assisted QKD.
where CAET
(X) is the average noise in the correlations We hope that this work will generate interest in inventing
of Alice and Eve. The strategy employed by Eve should more such protocols and also in experimentally building
be such that her average noise is minimized. One of the such a QKD system.
best strategies that accomplishes this is the one employed
by Bob when measuring at specified times. The collapse
and revival phenomenon also ensures that the minimum ACKNOWLEDGMENTS
noise in Eve’s data cannot be less than the vacuum noise
N0 . In order to achieve a positive secure key rate from J.S. acknowledges UGC India for financial support. A
Eqn. (37), it is required that χ < 1. For the case when and C. K. acknowledge funding from DST India under
Bob is also measuring at specified times, the condition Grant No. EMR/2014/000297.
[1] C. H. Bennett and G. Brassard, in Proceedings of the and Signal Processing (IEEE Press, New York, 1984) pp.
IEEE International Conference on Computers, Systems 175–179.
9
[2] C. H. Bennett, Phys. Rev. Lett. 68, 3121 (1992). [19] H. Takenaka, A. Carrasco-Casado, M. Fujiwara, M. Kita-
[3] N. Gisin, G. Ribordy, W. Tittel, and H. Zbinden, Rev. mura, M. Sasaki, and M. Toyoshima, Nature Photonics
Mod. Phys. 74, 145 (2002). 11, 502 EP (2017).
[4] P. W. Shor and J. Preskill, Phys. Rev. Lett. 85, 441 [20] A. Joshi, Physics Letters A 270, 249 (2000).
(2000). [21] M. J. Everitt, P. Stiffell, T. D. Clark, A. Vourdas, J. F.
[5] C. Branciard, N. Gisin, B. Kraus, and V. Scarani, Phys. Ralph, H. Prance, and R. J. Prance, Phys. Rev. B 63,
Rev. A 72, 032301 (2005). 144530 (2001).
[6] A. K. Ekert, Phys. Rev. Lett. 67, 661 (1991). [22] R. L. Fagaly, Review of Scientific Instruments 77, 101101
[7] A. Acı́n, N. Gisin, and L. Masanes, Phys. Rev. Lett. 97, (2006).
120405 (2006). [23] J. Zou and B. Shao, International Journal of Modern
[8] N. Cerf, S. Iblisdir, and G. Van Assche, The European Physics B 13, 917 (1999).
Physical Journal D - Atomic, Molecular, Optical and [24] J. Zou and B. Shao, Phys. Rev. B 64, 024511 (2001).
Plasma Physics 18, 211 (2002). [25] J. R. Friedman, V. Patel, W. Chen, S. K. Tolpygo, and
[9] T. C. Ralph, Phys. Rev. A 61, 010303 (1999). J. E. Lukens, Nature 406, 43 EP (2000).
[10] M. Pawlowski, Phys. Rev. A 82, 032313 (2010). [26] D. Mailly, C. Chapelier, and A. Benoit, Phys. Rev. Lett.
[11] J. Singh, K. Bharti, and Arvind, Phys. Rev. A 95, 70, 2020 (1993).
062333 (2017). [27] L. P. Lévy, G. Dolan, J. Dunsmuir, and H. Bouchiat,
[12] V. Scarani, H. Bechmann-Pasquinucci, N. J. Cerf, Phys. Rev. Lett. 64, 2074 (1990).
M. Dušek, N. Lütkenhaus, and M. Peev, Rev. Mod. [28] C. E. Shannon, Bell System Technical Journal 27, 379
Phys. 81, 1301 (2009). (1948).
[13] T. C. Ralph, Phys. Rev. A 62, 062306 (2000). [29] F. Grosshans and P. Grangier, Phys. Rev. A 64, 010301
[14] F. Grosshans and P. Grangier, Phys. Rev. Lett. 88, (2001).
057902 (2002). [30] M. J. Everitt, T. D. Clark, P. B. Stiffell, A. Vourdas,
[15] F. Grosshans, G. Van Assche, J. Wenger, R. Brouri, N. J. J. F. Ralph, R. J. Prance, and H. Prance, Phys. Rev. A
Cerf, and P. Grangier, Nature 421, 238 EP (2003). 69, 043804 (2004).
[16] H. Takesue, S. W. Nam, Q. Zhang, R. H. Hadfield, [31] J. Banerji, Pramana 56, 267 (2001).
T. Honjo, K. Tamaki, and Y. Yamamoto, Nature Pho- [32] J. Hannay and M. Berry, Physica D: Nonlinear Phenom-
tonics 1, 343 EP (2007). ena 1, 267 (1980).
[17] A. Fedrizzi, R. Ursin, T. Herbst, M. Nespoli, R. Prevedel, [33] J. Clarke and F. K. Wilhelm, Nature 453, 1031 EP
T. Scheidl, F. Tiefenbacher, T. Jennewein, and (2008).
A. Zeilinger, Nature Physics 5, 389 EP (2009). [34] L. DiCarlo, J. M. Chow, J. M. Gambetta, L. S. Bishop,
[18] P. Jouguet, S. Kunz-Jacques, A. Leverrier, P. Grangier, B. R. Johnson, D. I. Schuster, J. Majer, A. Blais, L. Frun-
and E. Diamanti, Nature Photonics 7, 378 EP (2013). zio, S. M. Girvin, and R. J. Schoelkopf, Nature 460, 240
EP (2009).
[35] J. M. Gambetta, J. M. Chow, and M. Steffen, npj Quan-
tum Information 3, 2 (2017).