FSO and Non-Terrestrial Networks: The Physical Foundation for a Quantum Internet
Building a quantum internet that spans the globe does not start with choosing an algorithm. It starts with a more basic question: how do you send photons over long distances? Right now, the answer is Free-Space Optics, combined with platforms flying at several different altitudes.
What Free-Space Optics actually is
FSO sends an optical signal, usually a laser, directly through air or vacuum, with no physical waveguide involved. It requires line-of-sight: transmitter and receiver have to see each other, with nothing blocking the path. Compared to fiber, FSO has two properties that matter a great deal for quantum applications. First, the path carries almost no significant thermal noise, unlike fiber, where spontaneous Raman scattering inside the core adds background noise of its own. Second, FSO handles single-photon transmission and detection well, which is a hard requirement for encoding a qubit in a photon’s polarization or phase.
Two characteristic sources of error
The laser beam has to be aimed at the receiver with extreme precision. A misalignment of just a few microradians can send the beam missing the receive aperture entirely, especially over distances of hundreds or thousands of kilometers. With a satellite constantly in motion, precise tracking becomes an engineering problem in its own right, requiring a high-precision Attitude Determination and Control System. This is called pointing error.
The second source comes from the atmosphere itself. Local temperature and pressure fluctuations change the air’s refractive index, producing two effects: beam wandering, where the beam drifts off-target unpredictably over time, and scintillation, where the received signal intensity flickers randomly. Both are commonly modeled with the Gamma-Gamma statistical distribution, the standard used across FSO channel research.
Three non-terrestrial platform layers
A quantum internet in the sky does not rely on a single type of satellite. It combines three altitude layers, each playing a different role.
LEO, at 500 to 600 kilometers, gives wide coverage through Starlink-style constellations, but its fast motion means each overpass offers only a short communication window. HAPS, at 19 to 22 kilometers, flies above the cloud layer, avoiding cloud blockage entirely, experiencing less turbulence, covering a stable footprint of roughly 500 kilometers, and often running on solar power. LAPS, below 5 kilometers, is deployed on demand using drones, suited to last-mile key distribution or temporary operation such as disaster response.
The key insight sits with HAPS: flying above the clouds is a strategic advantage, because it eliminates entirely the largest source of atmospheric loss that a ground-based link has to endure, leaving only the much thinner high-altitude atmosphere to contend with.
The shared constraint: SWaP-C
Whatever the layer, every quantum device deployed on a satellite, a HAPS, or a drone runs into SWaP-C: size, weight, power, and cost. This is not a minor engineering footnote, it is a central design constraint. A single-photon source or detector that performs perfectly well on an optical bench is often far too large, too heavy, or too power-hungry to fit on a small satellite or strap onto a drone.
Two ways of extending the network
When two nodes have no direct physical link, they can still be connected through an intermediate node, by performing a Bell-state measurement there on two separately entangled photon pairs. The result is that the two end nodes become entangled with each other, even though they never interacted directly. This is entanglement swapping, and it extends a quantum network’s reach without classical signal amplification, which is fundamentally incompatible with quantum states under the no-cloning theorem.
Another real-world limit is finite-key effects. A LEO overpass lasts only a few minutes, so the number of detected photons is always finite, not infinite as asymptotic theory assumes. With a finite sample, statistical error in estimating QBER and SKR grows significantly, meaning the achievable key rate is always lower than the theoretical limit. Any serious satellite QKD performance analysis has to account for this rather than relying on idealized formulas.
These ideas form the physical foundation for the problem I am actually working on: modeling the tropical satellite FSO channel for a QKD system, covered in more depth in Quantum Communication for Satellite Technology.