Cryptography Reference
In-Depth Information
-0.168
Single event:
-0.170
-0.172
36 42
48
45 46
Event Number
Figure 5.14
Typical oscilloscope trace of the detector signal with constant basis. The
insert shows the whole trace, with visible jumps due to basis switching by the EOM.
dark mode
a
y
, thus giving a considerable overlap according to Equation (5.2).
A typical readout of the oscilloscope is shown in Figure 5.14. The main graph
shows eleven events with constant measurement basis. In the inset one can
see the whole measurement trace, with a basis change visible as a large jump
of the signal curve. After the measurement, a bit value was assigned to each of
Bob's measurements. In the case of low loss (21%), a postselection threshold
of
1 reduced the number of bits from 1001 to 222. The information ad-
vantage
I
AB
−
|
α
|
>
I
AE
was larger than 0.7 bits per event for this threshold. For
high losses (64%), the threshold was set to
5, giving an information
advantage of 0.4 bits per event. 139 out of 956 bits survived the postselec-
tion. The error rates for both low and high losses are shown in Table 5.1.
Postselection reduces the initial error rate, as expected. As the coherent am-
plitude
|
α
|
>
1
.
decreases with the losses, the overlap of the states increases (cf.
Equation (5.2)). Thus the rising error probability has to be compensated for
by error correction, giving also a hint that the channel quality is low. Even
though the ratio of secret bits to transmitted bits drops for high losses, it is
α
Table 5.1
Measurement Results for Low and High Loss
Initial
Residual
Information
Loss (%)
Errors (%)
Threshold
Errors (%)
Advantage
21
15
1
4
0.7
64
22
1.5
10
0.4
Note:
Initial errors are before postselection with a threshold as in the
third column. After postselection, the fourth column shows the
remaining error rate. The information advantage in bits per event
is shown in the fifth column.
