Page 122 - Kaleidoscope Academic Conference Proceedings 2021

2021 ITU Kaleidoscope Academic Conference

2. RELATED WORK DIQRNG requires a more complex and larger system
to implement loophole-free Bell tests, thereby providing
A quantum random number generator generates a sequence of higher security. The DIQRNG has been implemented in
random numbers using the inherent randomness in quantum the laboratory for any quantum or classical physics-based
mechanics. Therefore, the generated quantum random strategy, random numbers are unpredictable [7, 8, 9].
numbers are intrinsically random and can be called true However Bell’s inequality itself requires randomness input,
random numbers. A quantum process to generate quantum a random seed is necessary in such a random number
noise can be decomposed in two steps: the quantum state generator. Therefore, such a process should be more of a
preparation and the quantum state measurement. Therefore, random expansion process. Randomness expansion generates
the generic functional architecture of a quantum entropy a longer sequence of random numbers from a short random
source is defined in Recommendation ITU-T X.1702 [5]. It number, which is feasible in quantum mechanics, but is not
also introduces a common method to estimate and validate allowed in classical mechanics due to the deterministic nature
the entropy of a noise source under evaluation, and a common of classical algorithms [3, 10, 11]. With the development
method to specify randomness extractors when they are part of cutting-edge single-photon detection techniques, and
of the implemented system. theoretical protocols [12, 13, 14, 15, 16, 17], loophole-free
However, the quantum random number generator faces Bell test experiments based on entangled photons have
many practical issues in the applications, such as device the opportunity to implement the randomness expansion
imperfections, component deviation, classic noise, side [18, 19, 20].
channels, adversary attacks, etc. Without the assumption
about the inner workings of the devices, the outputs may not 3. DEVICE-INDEPENDENT QUANTUM RANDOM
be genuinely random and unpredictable. Typical QRNGs NUMBER GENERATOR
require a detailed characterization of their operation to
ensure the quality of their output, and thus could be called The DIQRNG is based on a loophole-free Bell test
device-dependent QRNGs. Ensuring the conditions are met experiment. A typical Bell test involves multiple separated
places a significant burden on the user. So far, various measurement devices. Each of them has more than one
QRNG schemes have been proposed and demonstrated, and measurement setup and they are selected randomly. If some
a variety of commercial products have been on shell. The correlation of the measurements is broken, which is called a
randomness from a device-dependent QRNG relies on the violation of Bell’s inequality, quantum entanglement between
accurate characterization of its devices. However, device these devices can be demonstrated, and the measurements can
imperfections and inaccuracy may lead to wrong entropy be used to generate theoretically secure informational random
estimation and bias in practice, which highly affects the numbers.
genuine randomness generation and may even induce the In proving that Bell’s inequality is violated, three loopholes
disappearance of quantum randomness in extreme cases. need to be closed: the locality loophole, the detection
In order to effectively solve the problems of device loophole, and the freedom-of-choice loophole. Physicists
imperfections and inaccuracy, different QRNG protocols can interpret the experimental results according to classical
have been proposed recently to obtain certified genuine physics with these loopholes open, so it is impossible to
randomness even when devices are untrusted and prove that the generated randomness comes from quantum
uncharacterized, e.g., DIQRNG and semi-DIQRNG. effects. To close the locality loopholes, we need the
The DIQRNG protocol produces certified randomness measurements of different devices to satisfy the non-signaling
based on the violation of Bell’s inequality without conditions. The measurement devices should not have the
trusting the quantum devices. However, the DIQRNG opportunity to communicate with each other. One way is
requires efficiency-loophole-free Bell tests, which makes to precisely control the measurement process and time to
the experimental implementation rather challenging and ensure that the spatial-like relations between measurement
inefficient. In practice, there is a trade-off between events are satisfied, thus ensuring that the measurement
system security and performance. By adding a few events are independent of each other. The other way is
reasonable assumptions to the devices, the DIQRNG becomes to block the communication between measurement devices
much more practical, which is called semi-DIQRNG. by electromagnetic shielding. Obviously, the former is
The device of QRNG is mainly divided into two more thorough in closing the loophole, while the latter is
parts: source and measurement. Therefore, there easier to implement. Closing the detection loophole requires
are two typical semi-DIQRNGs: Source-independent analysis of all measurement results without any post-selection
Quantum Random Number Generator (SI QRNG) processing. As shown in [21], the detection efficiency must
and Measurement-Device-Independent Quantum Random exceed the threshold to ensure a violation. The free choice
Number Generator (MDI QRNG), depending on which part of the vulnerability requires that the measurement setup
is untrusted. To date, the fastest semi-DIQRNG has been options are chosen individually and randomly. This loophole
proposed is the SI QRNG at 17 Gbps [6], and the rate cannot be closed perfectly, and most people use a space-like
can satisfy the requirements of most applications. This independent quantum random number generator to close it.
suggests that semi-DIQRNG has the potential for large-scale The recent progress of loophole-free tests of Bell inequality
applications. provides a way to implement DIQRNG, the implementation

– 60 –