Short answer: Continuous-variable QKD (CV-QKD) encodes key information in the quadratures of light and recovers it with homodyne or heterodyne detection, instead of counting single photons. Reviews say this makes it easier to build with standard telecommunication components. The trade-off is that complete security proofs are harder, because continuous values do not map neatly onto digital processing. Papers from 2015 to 2022 report progress on those proofs. They are research results, not product evaluations.
How it differs from single-photon QKD
A 2015 review in the journal Entropy says quantum key distribution was first designed for qubits, the individual quanta of information, but using quantum continuous variables presents important advantages, particularly from a practical point of view, since it allows simple implementations that require only standard telecommunication technology. The review focuses on protocols based on coherent states. It explains that all CV-QKD protocols encode information in the quadratures of the quantized electromagnetic field and recover it with coherent detection, in particular homodyne or heterodyne detection. The main distinction from discrete-variable protocols is the detection technique: single-photon detection for those. The review discusses security, state-of-the-art experimental implementations and side-channel attacks.
A 2020 Nature Communications paper puts it similarly. It says CV-QKD with homodyne or heterodyne measurements has distinct advantages of lower-cost implementation and affinity to wavelength division multiplexing compared with discrete-variable QKD, which uses photon detectors and includes BB84 and B92.
The security-proof gap
The 2020 paper explains the difficulty: the continuous nature of CV-QKD makes it harder to accommodate practical signal processing, which is always discretized, and this has led to a lack of complete security proofs. The authors propose a tight and robust method of estimating the fidelity of an optical pulse to a coherent state through heterodyne measurements. They then construct a binary phase modulated CV-QKD protocol and prove its security in the finite-key regime against general coherent attacks, using proof techniques from discrete-variable QKD. They state that such a complete proof is indispensable for exploiting the benefits of CV-QKD.
A 2022 arXiv paper (2205.11820) analyses security against collective attacks for a homodyne-based protocol using binary coherent states and postselection. The authors derive a lower bound on the secret key rate in an asymptotic scenario and optimise it numerically for a symmetric Gaussian channel. The paper notes that experimental studies suggest CV-QKD devices are relatively easy to incorporate into an optical fibre network and can operate at a high repetition rate, and that earlier work gave a sequence of asymptotic security proofs for discrete modulation and a finite-size proof for a binary coherent state protocol.
What this does not tell you
These papers do not compare specific commercial systems or give field performance. The 2015 review mentions that QKD systems operate in real-world conditions and are commercially available, but that is its general statement. Key-rate bounds depend on the channel and attack model, and the 2022 paper's bound is for collective attacks in an asymptotic setting. Side-channel attacks on implementations are a separate issue the review also discusses.
FAQ
What is continuous-variable QKD?
QKD that encodes information in light's quadratures and uses homodyne or heterodyne detection instead of single-photon detection.
Why is it attractive?
Reviews cite simpler implementation with standard telecom components, lower cost and compatibility with wavelength division multiplexing.
What is the main difficulty?
Complete security proofs, since continuous signals must be digitised in practice.
Has finite-key security been proven?
A 2020 paper proves it for a binary phase modulated protocol against general coherent attacks, as the authors report.
Is a security proof the same as a secure product?
No. Proofs cover a protocol and its assumptions, not implementation flaws.
All field notes