Cryptography Reference
In-Depth Information
in 1989 [14,15]. Roughly speaking, two entangled photons propagate toward
two distant interferometers that both contain a long path L and a short path
S. The photons are emitted at the same time in a parametric down-conversion
source, and if they arrive at the detectors at the same time, it follows that
they both must have traveled the longer path or they both must have trav-
eled the shorter path. Quantum interference between these two probability
amplitudes gives rise to nonlocal quantum correlations that violate Bell's in-
equality.
As early as 1989, John Rarity noted that a two-photon interferometer of
this kind could be used as a method of quantum key distribution [16]. Ek-
ert, Rarity, Tapster, and Palma later [17] showed that tests of Bell's inequality
could be used to ensure that an eavesdropper cannot determine the polar-
ization states of the photons without being detected, which allows secure
communications to be performed. Systems of this kind have now been experi-
mentally demonstrated [18]. One potential advantage of an entangled-photon
approach of this kind is that no active devices are required in order to choose
a set of random bases for the measurement process. Instead, 50-50 beam split-
ters can randomly direct each photon toward one of two interferometers with
fixed phase shifts.
The interferometric approach of Figure 6.1(a) has the disadvantage of
requiring a parametric down-conversion source, which typically has a lim-
ited photon generation rate. Charles Bennett realized [19], however, that the
need for an entangled source could be eliminated by passing a single photon
through two interferometers in series, as illustrated in Figure 6.1(b). Although
nonlocal correlations cannot be obtained in such an arrangement, it does al-
low the use of weak coherent state pulses containing much less than one
photon per pulse on the average. The ease in generating weak coherent state
pulses combined with the relative lack of sensitivity to polarization changes
made this type of interferometer system relatively easy to use. As a result, a
number of groups [20-23] demonstrated quantum key distribution systems
of this kind, including work by Townsend, Rarity, Tapster, and Hughes.
One of the disadvantages of the interferometric approaches of Figures
6.1(a) and (b) is that the relative phase of the two interferometers must be
carefully stabilized. In addition, the polarization of the photons must still
be controlled to some extent in order to achieve a stable interference pat-
tern. Gisin and his colleagues [24] avoided both these difficulties by using a
very clever technique illustrated in Figure 6.1(c). Here the system is essen-
tially folded in half by placing a mirror at one end of the optical fiber and
reflecting the photons back through the same interferometer a second time.
By using a Faraday mirror, the state of polarization is changed to the orthog-
onal state during the second pass through the optical fiber, which eliminates
any polarization-changing effects in the optical fiber. Plug-and-play systems
of this kind are very stable and are now in widespread use.
The remaining problem in existing quantum key distribution systems is
the limited range that can be achieved in optical fibers due to photon loss.
As a potential solution to this problem, we performed the first demonstration
