117
M. S. JIANG ET AL.
polarization states to represent his bit value. The encoded
single-photon pulse entering a Michelson-interferometer
is split into two pulses by a beam splitter (BS) and trav-
els through two modes a and b. In mode a, the pulse is
reflected back by a Faraday mirror (FM), and always
confined within Alice’s secure zone. In mode b, the pulse
travels from Alice’s site to Bob’s site. Bob also ran-
domly chooses one of the two orthogonal polarizations
representing his bit value by blocking the corresponding
polarization state, named as polarization-selection. If an
optical pulse incident to Bob’s site is horizontally polar-
ized, it passes thro ugh the polarizin g beam splitter (PBS)
and goes directly to the high-speed optical switch (SW).
However, if the pulse is vertically polarized, it is first
reflected by the PBS, passes through the optical loop
(OL), and then goes to the SW. Therefore, through accu-
rate control of the switch timing, Bob can effectively
switch the selected polarization state to the detector D3,
while the other was reflected back to Alice’s BS. Thus, if
the bit values chosen by Alice and Bob are different, the
split pulse going through mode b is not blocked by Bob,
and the two split pulses are recombined in the BS, and
the single photon is detected at detector D2 certainly as a
result of quantum interference. On the other hand, if the
two bit values are equal, the split pulse going through
mode b is blocked by detector D3. Then, the photon can
be detected at detector D1 with a finite probability,
which is caused by the breakage of the interference. In
this case, the photon has been completely confined
within Alice’s secure zone, and Eve can never access the
photon, as it has only traveled through mode a. Alice and
Bob can then establish a sifted key by selecting only the
events for which D1 clicks alone. In summary, the proc-
ess can be described as follows: 1) when the bit values of
Alice and Bob are different, D2 clicks with probability
1/2; 2) when the bit values of Alice and Bob are equal,
D1 clicks with probability RT2, D2 clicks with prob-
ability 2
R2, and D3 clicks with probability T2. The
events for which D1 clicks alone are used to extract a
secret key, and the other events are used to detect the
latent eavesdropper (Eve). Here and T1 are
the reflectivity and transmissivity of the BS, respectively.
RR
Now, we analyze the PCQC protocol in the aspects of
QBER and stabilization. It is mentioned in [10] that it
may be hard to stabilize a long-armed interferometer,
which is related to QBER and stabilization. For a long-
armed interferometer, the symmetry of the interferometer
relies sensitively on th e environmental disturbances such
as temperature fluctuations. The breakage of this sym-
metry will cause a variation of the interference, for ex-
ample, phase drift may even completely destroy the in-
terference. Ideally, it is easily seen that th e single photon
is detected at detector D2 with certainty when the bit
values of Alice and Bob are different, as a result of
quantum interference. But in fact, we can never keep the
interference perfect for a long-armed interferometer un-
der environmental disturbances. The extreme result is
that the optical path difference of the interferometer is
larger than the coherence length of the light source be-
cause of fiber length drift. Consequently, the interference
is completely destroyed, that is, mode a and mode b of a
single photon are not coherent any more. In this case, a
single photon can be detected at detector D1 with prob-
ability 1/2, which is an error ev ent and add s an add ition al
QBER in the raw keys. Generally speaking, the inter-
ferometer can be stabilized using feedback control. Here
we assume that, once the bit va lues of Alice and Bob are
different, mode a and mode b of a single photon are al-
ways coherent under feedback control. However, error
events may still happen with some probability as a resu lt
of phase drift. Here we note this event as phase-crosstalk,
and the corresponding probability as
hase . Note that in
PCQC protocol, the events for extracting a secret key
have a probability of RT/2, and the probability of error
events caused by phase-crosstalk is
C
phase
C2. Thus, the
additional QBER caused by phase-crosstalk can be writ-
ten as
phase
phase phase
C2
QBER C2RT
2
(1)
Furthermore, since Bob must perform polarization-
selection through the PBS to represent his bit value, in-
stability due to the polarization mode dispersion effects
in long-distance single-mode fiber should also be con-
sidered. In fact, a long-distance single-mode fiber should
be considered as a birefringent medium as a result of its
intrinsic imperfection and environmental disturbances.
When a single photon passes through such a birefringent
medium, the polarization mode dispersion effect is visi-
ble, which will result in the instability of polarization.
Therefore, well performed polarization compensation is
needed to compensate for the instability of polarization;
otherwise the polarization-selection will not perform
well, resulting in another additional QBER in the raw
keys. But in fact, the polarization of a single photon will
of course drift away from its original state more or less
after traveling from Alice’s site to Bob’s site, regardless
of how well the polarization compensation is performed.
In the process of polarization-selection, if a single photon
enters Bob’s PBS with horizontal polarization, it passes
through the PBS and goes directly to the SW; and if the
single photon is vertically polarized, it is firstly reflected
by the PBS, passes through the OL, and then goes to the
SW. However, polarization dr ift may sometimes bring in
unexpected result. For example, when the bit values of
Copyright © 2011 SciRes. JQIS