IEEE - Aerospace and Electronic Systems - March 2021 - 41

Amer et al.
that they can be bounded leading to more optimistic keyrate computations (and, thus, greater allowed distances
between users). Indeed, experimentally, the record distance for a QKD protocol over fiber is 421 km [92] using
a decoy state protocol introduced recently in [93] (which
is a more advanced variant of the one we discussed here).

DEVICE INDEPENDENCE
So far we have briefly mentioned some attacks on QKD
systems and in a later section we will look at these attacks
in more detail. Attacks on QKD systems side-step the theoretical unconditional security guarantees by exploiting
imperfections in the devices used to implement the protocols. In an effort to mitigate against the existence of these
imperfections, device independent (DI)-QKD has been
proposed and is an active area of research [94]-[98].
These protocols are devised in such a way as to be secure
even in the case where the adversary has full information
over the devices in Alice and Bob's nodes. Such protocols
generally involve the creation of entangled photons and
verifying they are maximally entangled through Bell
inequality violations [99]. Using the fact that maximally
entangled systems cannot be entangled with any other systems (e.g., cannot be correlated with any adversary), Alice
and Bob are then able to be certain that Eve's system is
independent of the key.
An alternative to fully DI-QKD is MDI QKD [51],
[52], [100]. Unlike DI-QKD, in which it is assumed that
all of Alice and Bob's apparatus are untrustworthy, in
MDI-QKD, we must assume that at least their state preparation devices can be trusted but not the measurement
devices. With this assumption, MDI-QKD can guarantee
security against any and all known and yet unknown sidechannels in measurement devices (as these are assumed
under complete control of the adversary). This guarantee
is achieved by having Alice and Bob communicate with
an untrusted third party, Charlie, rather than each other.
This third party (who may be the adversary) is responsible
for performing measurements on the qubits received from
A and B, reporting the outcomes. Even if this third party
is adversarial, security can be guaranteed. Since A and B
do not need any measurement devices (only qubit sources), side-channel attacks against measurement devices
are eliminated. The current record distance for MDI-QKD
is 404 km [101]. More recently, a new protocol called
Twin-Field-QKD [102], [103], which in some cases can
be considered a variant of the MDI-QKD protocol, has
been shown to be capable of breaking the 421 km record
held by the decoy state protocol in [92]. A TF-QKD protocol generally operates with both users A and B having
their own sources and preparing a phase randomized pulse
to a centralized measurement device (which may even be
adversarial; which is why some TF-QKD protocols may
be considered a form of MDI protocol as proven in [104]).
MARCH 2021

FREESPACE QUANTUM COMMUNICATION
Complimentary to fiber optic channels is the use of free
space QKD. While fiber optic channels exhibit an exponential increase in loss over distance generally due to Rayleigh scattering loss, freespace communication's loss rate,
generally due to diffraction, is only quadratic giving it a
significant advantage; see [10] for information on the loss
rates of freespace quantum communication versus fiber.
Free space QKD may allow us to conduct QKD over vast
distances. As would be expected, however, communicating quantum states over free space is a more complex task
than simply using fiber optic channels, and so the hardware needed to do so is more complex and has its own
associated disadvantages. Here, we discuss the implications of the various possible configurations of free-space
communication.
A free space communication link is generally composed of at least two entities: namely ground station A
and a satellite SAT (though it may also be some other
remote entity, such as a plane in flight [8]). We can add a
third entity, ground station B, and use SAT as a kind of
trusted node (described later in this section) to establish
communication between ground stations A and B; this
kind of network was used in the ground-breaking Micius
satellite demonstration where QKD secured communication was conducted over 7500 km between Beijing and
Vienna [105]. The two main decisions that must be made
when configuring a free space network are the location of
the satellite and the division of labor [106]-[109].
First we will discuss the location of the satellite, for
which we consider two main options. The satellites can be
placed in low earth orbit (LEO), or geostationary orbit
(GEO) [108]. In LEO, the satellite is located at an altitude
of 160 km to 3000 km, orbiting the earth at very high
speeds. The low altitude of LEO satellites comes with the
advantage of causing less loss than is found with GEO satellites, but LEO satellites must move at relatively high
velocities to maintain their orbits, thus they have a much
tighter temporal window in which a link can be established with a given ground station. The regularity and frequency of these opportunities can also vary greatly
depending on the exact orbit of the satellite. For example,
the Micius satellite, which is in a 500 km LEO sun-synchronous orbit, is able to link with a ground station in
Xinglong for 5 min every night. A satellite in the International Space Station orbit would be limited to only 150
links over the course of a year. A GEO satellite, on the
other hand, is relatively static, and so can continuously
service the same stations that wish to use it to establish
QKD generated keys with one another. Of course, GEO
satellites must be located at an altitude of 35 786 km, and
so experience significantly more loss than LEO satellites.
In practice, LEO satellites have thus far been the more
commonly used satellites for QKD, though there have

IEEE A&E SYSTEMS MAGAZINE

IEEE - Aerospace and Electronic Systems - March 2021

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - March 2021