Make your paper interactiveCreate

Text from the interactive explanation

Satellite-Based Continuous-Variable Quantum Communications: State-of-the-Art and a Predictive Outlook

Nedasadat Hosseinidehaj, Zunaira Babar, Robert Malaney, Soon Xin Ng, Lajos Hanzo

Original paper

Could the laser light that already carries the internet also carry keys no computer could ever crack, between any two places on Earth? The authors picture perfectly secure communication “for anyone, anywhere, anytime.”

In 2017 the Micius satellite sent entangled single photons to two ground stations 1,203 km apart. This survey looks at the next step: continuous-variable quantum light, meaning ordinary laser pulses read by the same kind of receivers telecom uses. Here is what stands in the way, and how to get past it.

Simulation built on the beam-wandering model surveyed in the paper (its Eqs. 36–39): amber pulses fly along the beam; the receiver’s ring flashes with the share it catches, and missed light scatters away. Slowed down and not to scale. The survey anticipates losses of about 5–10 dB for a downlink from a low-orbit satellite and 25–30 dB or more for an uplink.

Why go to space at all?

Glass fibre already links the planet. Why launch a satellite?

Quantum key distribution (QKD) lets two people, traditionally called Alice and Bob, build a shared secret key by sending light. Its security rests on physics, not on a maths puzzle that a future quantum computer might solve. But the light has to arrive.

In optical fibre, light fades exponentially: even the best fibre, at a wavelength of 1550 nm, loses about 0.2 dB per kilometre . Every 15 km halves what remains. The survey notes this limits fibre links to a few hundred kilometres unless “quantum repeaters” mature.

A satellite beam spends most of its journey in empty space. Only the last few kilometres cross dense air, so its losses barely depend on how far apart the two ground stations are.

Predict first: beyond roughly what distance does a satellite beat fibre?

Light is a fuzzy blob

What exactly gets written into a laser pulse?

A light wave has two independent “handles”, called quadratures , labelled q and p. Roughly, they describe how strong the wave is and where it sits in its cycle (its amplitude and phase). Continuous-variable (CV) quantum communication writes information onto these two numbers.

Plot q across and p up and a perfectly steady classical wave would be a sharp dot. Quantum light can’t be: the uncertainty principle says the spreads of q and p can’t both be small. A laser pulse is a round blob whose width is fixed by nature: the vacuum noise, also called shot noise. The plot is a Wigner function, the paper’s Fig. 8.

Try it: drag the laser blob around. That is how a sender encodes numbers. Then push “Squeeze”. One direction tightens below the vacuum level while the other balloons, and the product of the two spreads never drops below 1. Add heat and the blob only grows.

To read a pulse, a receiver mixes it with a bright reference laser, the local oscillator . This is homodyne detection. Its phase picks which handle you measure, q or p, never both perfectly. Each reading is one random draw from the blob’s shadow on that axis. Press “Measure 200” and watch the histogram rebuild the blob.

Units follow the paper: vacuum noise variance = 1. A Gaussian state with mean (q₀, p₀) and variances V(q), V(p) has the Wigner function

Uncertainty principle: V(q)·V(p) ≥ 1. Squeezing with parameter r gives V(q) = e^(−2r), V(p) = e^(2r). A thermal state with mean photon number n̄ has V = 2n̄ + 1 in both. Fig. 8 shows vacuum, a coherent state centred at (3, 5), a thermal state with V = 5 and a squeezed state with r = 0.5. The survey notes 15 dB is the strongest vacuum squeezing achieved so far.

Writing a key in blobs

How do two strangers end up holding the same random numbers?

The simplest CV protocol, laid out step by step in the paper’s Table V, works like this. Alice draws two random numbers from a bell curve and shifts a laser pulse to that point, so its blob sits at (q, p). Bob flips a coin and measures either q or p. Afterwards Bob announces publicly which handle he measured (never the value), and Alice keeps the matching number. This step is called sifting .

Every dot is one kept pulse: Alice’s number across, Bob’s reading up. Even over a perfect channel the cloud has thickness, because each of Bob’s readings carries the blob’s unavoidable vacuum noise. The two lists are strongly correlated but not identical. Error correction (“reconciliation”) and “privacy amplification” later distil them into an identical, secret string of bits.

And the eavesdropper? A quantum state cannot be copied (the no-cloning theorem ). Anything Eve learns, she must take from the light itself, and that leaves Bob with less. To see how much, we have to look at what happens when light goes missing.

Every lost photon becomes noise

If only part of the light arrives, why not just turn up the gain?

The paper models a lossy channel as a beam splitter : a fraction τ of the light continues to Bob, and the rest is swapped for empty-space vacuum noise (Fig. 10). For a laser pulse, the blob’s centre shrinks toward zero by √τ while its fuzz stays the same size. Alice’s neatly spread constellation crowds together and blurs.

This is the big difference from single-photon (“discrete-variable”, DV) systems such as Micius. There, a lost photon is simply a missing click with no error. In CV, lost light turns into noise.

Security proofs make the most pessimistic assumption: Eve holds all the light that went missing (the “entangling-cloner” attack of Fig. 14). The secret key per pulse is what Alice and Bob share, trimmed by error-correction efficiency ξ, minus the most Eve could know:

K = ξ·IAB − IE

Try it: drag the loss up through the downlink band, then the uplink band. The key does not vanish outright. It shrinks roughly in step with the light. Then lower ξ: at high loss Bob’s signal is so faint that imperfect error correction alone can wipe out the key. The survey names efficient error correction at low signal-to-noise as a key open problem for satellites.

The sky wobbles

Why is sending up so much harder than sending down?

Reproduce the paper’s Figs. 18–19 (aperture radius = 1):

Air is full of pockets of slightly different temperature, called turbulent eddies . Each bends light a little. The survey singles out three effects: beam wandering (the whole beam drifts), beam spreading, and scintillation (twinkle). In clear weather, beam wandering and spreading dominate.

Here you are the receiver, looking along the beam. The ring is your telescope (radius 1). The glow is the arriving beam, of radius W. Its centre drifts randomly with typical size σb. The share you catch follows the paper’s Eq. 38, and collecting many moments traces out the log-negative Weibull distribution (Eq. 40) in the histogram.

So the channel is not reciprocal: “there will be no directional reciprocity”. Downlinks lose orders of magnitude less light. Micius measured at least 41 dB of loss on its uplink. Try it: switch between the presets, then press the paper’s figure buttons. The simulator reproduces the mean losses printed in the paper.

Measure the sky, keep the bright moments

Can throwing data away create a secret key?

The atmosphere flickers a few thousand times a second, at least a thousand times slower than pulses are sent. So the receiver can measure the channel as it goes , for example from the bright local-oscillator beam sent alongside the signal (Section V-E).

If Alice and Bob ignore the wobble, good and bad moments are blended. The blend looks exactly like extra noise, and a cautious security analysis must blame it on Eve. Under uplink conditions the key vanishes. The survey’s verdict: at such losses, satellite QKD “will remain a fruitless endeavor without” post-selection (or entanglement distillation). Post-selection means keeping only the windows where the beam was bright.

Key versus threshold

Secret bits per sent pulse after keeping only moments above the threshold. The star marks the best threshold.

Try it: on the uplink, drag the dashed line (or the slider) upward until the key turns amber. You will be discarding most of the data, and gaining a key. Knowing which moments were good is worth more than the extra pulses.

“Track every moment” sorts each window by its measured brightness and treats it separately (the paper’s Eq. 44 approach), which does better still. The survey also covers entanglement distillation, which purifies the received states without measuring the channel.

Transmission samples come from the beam-wandering model (Eqs. 36–39) with the uplink (W = 8, σb = 8) or downlink (W = 2.5, σb = 0.5) presets. Ignoring the fading, the averaged state carries correlations ∝ ⟨η⟩ but noise ∝ ⟨η²⟩. The mismatch acts as excess noise εfade = Var(η)(V−1)/⟨η⟩². Post-selection applies the same Gaussian treatment to the kept subset, times the fraction kept. Key rates use the same formulas as Chapter 4 (ξ = 0.95). These are simulations, not measurements.

Five ways to wire the sky

Where should the delicate quantum hardware live: in orbit or on the ground?

Architectures from the paper’s Fig. 21. Loss figures are the survey’s typical values for a low-orbit satellite. All five can distribute entanglement or run CV-QKD; they differ in what has to fly and whom you must trust.

Single photons or laser waves?

Micius used single photons. Why bother with continuous variables?

Discrete variables: single photons + photon counters

Continuous variables: laser blobs + homodyne

Illustrative animation of the qualitative comparison in the paper’s Section VIII and Table VI. No measured rates are shown.

The road ahead

What still has to happen before quantum keys rain from the sky?

When the survey was written, quantum communication via satellite was “in its infancy”. In the CV domain no space deployment had yet been achieved, despite many theoretical studies. One encouraging sign: homodyne detection on the ground of signals from a geostationary satellite had already been demonstrated. The authors highlight five research fronts:

Channel statistics

Micius gave the first real loss data: at least 41 dB on its uplink. Validating the beam-wandering model (uplinks) and the beam-broadening models (downlinks) against such data matters, because CV performance is so sensitive to the transmission statistics you explored in Chapters 5–6.

Error correction at low signal

Turning correlated numbers into identical bits needs codes that work at very low signal-to-noise (multi-edge LDPC codes are the favoured route). The signal changes as a satellite rises and sets, but predictably, so rate-adaptive codes could track it.

Finite-key effects

Security proofs usually assume endless data. A low-orbit satellite is overhead for only minutes, so analyses must account for short, finite keys.

CV quantum error correction

Purely Gaussian operations cannot undo Gaussian errors. Non-Gaussian tricks such as photon subtraction and addition (Fig. 23) can distil entanglement and are becoming mainstream in labs worldwide.

Microwave ↔ optical

Superconducting quantum processors work with microwaves. Converters could lift their quantum information onto light, beam it via satellite and convert it back. This “could well represent how future quantum computers will share information globally through the quantum Internet.”

Anyone, anywhere, anytime

You have now seen the whole chain the survey lays out. Light carries information as fuzzy blobs. Loss turns into noise that an eavesdropper could hide behind. The sky’s wobble makes uplinks brutal and downlinks gentle. And measuring the sky lets you keep only its best moments, rescuing keys that would otherwise vanish.

The authors conclude that, after Micius, CV quantum communication over large distances “has become entirely plausible”. Its big advantage is that it runs on off-the-shelf lasers and homodyne detectors, the same toolkit as classical optical communication. Whether single photons, laser waves or a hybrid of both will carry the future quantum internet remained an open question. What they set out is the engineering map for getting there.

Source & credits

Nedasadat Hosseinidehaj, Zunaira Babar, Robert Malaney, Soon Xin Ng and Lajos Hanzo, “Satellite-Based Continuous-Variable Quantum Communications: State-of-the-Art and a Predictive Outlook,” IEEE Communications Surveys & Tutorials, vol. 21, no. 1, pp. 881–919, 2019. doi:10.1109/COMST.2018.2864557

The article is published under a Creative Commons Attribution 3.0 license. Every visual on this page is an original drawing or simulation based on the paper’s equations and descriptions; Figs. 8, 10, 16–21 and Tables V–VI are reinterpreted, not reproduced.

Beam wandering (hero, Chapters 5–6): transmission from the paper’s Eqs. 36–40 with receiver aperture radius 1; downlink preset W = 2.5, σb = 0.5 (mean loss 6.2 dB), uplink preset W = 8, σb = 8 (22 dB). The hero’s motion is slowed and its geometry is not to scale.

Key rates (Chapters 4 and 6): standard asymptotic formulas for Gaussian-modulated coherent states with homodyne detection, reverse reconciliation and collective Gaussian attacks, ideal detectors, VA = 4, excess noise 0.01 shot-noise units. Fading and post-selection use the averaged covariance matrix of the (kept) ensemble. These are simulations for illustration, not results reported in the paper.

Fibre arithmetic (Chapter 1): 0.2 dB/km from the paper; the “billion pulses per second” timing is illustrative.

Single photons versus waves (Chapter 8): qualitative illustration of Section VIII / Table VI.

Topics in this iPaper

Physics
A single mode of a CV system can be described as a single quantum harmonic oscillator of a specific frequency

Section III, page 887

Engineering
The main sources of loss in FSO communication are diffraction, absorption, scattering and atmospheric turbulence

Section V, page 900

Computer science
multi-dimensional reconciliation via multi-edge LDPC codes is considered by many as the most appropriate path

Section IX-B, page 912

quantum key distribution
CV-QKD protocols using Gaussian quantum states have been richly analysed in theory

Section IV, page 894

continuous variable quantum information
we survey the imminent extension of space-based quantum communication to the continuous-variable regime

Abstract, page 881

satellite communication
In the context of satellite-based quantum communication we are faced with two different channels, namely, the uplink

Section VI-A, page 904

free space optical communication
FSO quantum communication via satellites [39]–[69] has the additional advantage that communications can still take place

Section I, page 883

atmospheric turbulence
This makes beam-wandering the dominant effect in the uplink

Section V-B, page 901

quantum entanglement
As a result of the entanglement distribution in each quantum communication scheme of Fig. 21

Section VI-A, page 905

Satellite-Based Continuous-Variable Quantum Communications: State-of-the-Art and a Predictive Outlook — Interactive explanation | iPaper.ai