A secure approach for cloud-based quantum computing utilizes the data encryption capability of quantum mechanics.
Although quantum technology has advanced rapidly, the day
when every household or business will have a quantum computer is still far off.
The initial quantum computing will likely rely on a quantum "cloud"
where consumers send their computing tasks and data to a state-of-the-art
quantum computer hosted by companies like Google,
IBM, or others.
 |
| Security Advantage of Quantum-Based Computers |
But is this strategy secure? It is possible because
quantum-based protocols provide unbreakable
privacy. In a recent study, trapped ions were used to demonstrate a form of
"blind quantum computing". Since the protocol is scalable, it can be
integrated into larger quantum computing platforms.
Quantum computers could be revolutionary in computationally
challenging areas such as drug discovery
and material design. The use of a cloud-based quantum computer will raise
questions in these fiercely competitive industries. According to Peter Drmota
from the University of Oxford, "a company looking for a new miracle drug
or a high-performance battery material wouldn't want to expose their trade
secrets." However, theoretically, computations can be performed on a
remote quantum computer while keeping the data and operations performed on it
hidden. Drmota suggests that "a customer might feel safer using a quantum
computer that performs blind quantum computation."
Previous research on blind quantum computation using
photonic methods has been conducted by various groups. The fundamental drawback
of these configurations is their probabilistic nature, requiring repeated attempts
due to the success or failure of quantum entanglement operations and the
selection of the desired output afterwards.
Joe Fitzsimons from Horizon Quantum Computing, a company
producing integrated software for quantum computers, says, "Performing blind
quantum computation solely with photons is challenging due to the lack of
deterministic entanglement operations." Fitzsimons, not involved in this
study, expects the community to anticipate a demonstration of blind quantum
computation using matter-based qubits rather than photon-based qubits.
Drmota and his colleagues successfully demonstrated this
using a basic blind quantum computation setup consisting of two trapped ions,
one made of strontium and the other calcium. With extended coherence times, the
calcium ion serves as a memory qubit, while the strontium ion acts as a network
qubit transmitting photons to a "client". Together, the two ions form
the "server" of the quantum cloud system.
The network qubit sends a photon to the client via an
optical fiber as the first step of the team's blind computation protocol. Since
the polarization of the photon is entangled with the electronic state of the
network ion, the two objects are quantum entangled. The client uses this
entanglement to "steer" the ion's state by measuring the state of the
photon.
Crucially, the client secretly selects the direction of the
polarization measurement while measuring the photon's polarization. Using this
measurement, the client prepares the state of the network qubit. According to
Dominik Leichtle from Sorbonne University, "The entire system 'collapses'
to a specific state known only to the client." "The server doesn't
know which state the network qubit collapsed to because it's unaware of the
measurement."
However, the information of the network qubit can be
processed by the server through a laser-based procedure intertwining the
network and memory qubits. Furthermore, to better guarantee the security of the
protocol, the data is encrypted using a technique known as "one-time-pad
encryption". With this method, a list of random integers is generated by
adding additional rotations to the instructions given by the client to the
server. According to Drmota, "everything leaving the client and everything
returned to the client is meaningless."
Additionally, the client has a tool to verify that the
computation is performed correctly. According to Leichtle, such verification is
crucial for trusting an untrusted or error-prone quantum computer. Previous
verification techniques had been developed, but they often required a
significant amount of computing power. Leichtle and his colleagues devised a
more effective methodology that involves testing fake inputs and replacing real
data with them.
The researchers implemented this protocol using a two-ion
setup and demonstrated how a client can confirm the accuracy of their quantum
computations.
The group's initial presentation revealed that a client
could instruct the server to perform a simple quantum operation known as qubit
rotation. When the data is analyzed and the encryptions are decrypted, the
client discovered the expected fringe pattern. By adding more memory qubits,
the trapped ion system can be made more powerful and capable of performing more
complex computations. Connecting all these qubits will not be easy, but quantum
information experts have shown that they can connect tens of trapped ions and
propose systems with 1000 ions.
According to Drmota and Leichtle, the blind quantum
computation method they developed is "scalable" as this term is
defined as the client device and interface remain unchanged regardless of the
server's size.
According to Broadbent from the University of Ottawa, "the final demonstration of blind quantum computation using trapped ions and photonic detection represents a significant milestone towards scalable and secure quantum communication." "These developments open the door to a quantum internet that guarantees privacy and verifiability as practical use approaches." Fitzsimons agrees, noting that researchers have had to overcome significant technological hurdles to connect matter qubits to a photon-based communication network. "However, more work will be needed to enable blind quantum computation on quantum processors with higher qubit counts because the current demonstration is still limited by a small number of qubits," he adds.