CONCEPT Cited by 1 source
Free-space optical communication¶
Free-space optical communication (FSO) is data transmission through free space — vacuum or atmosphere — using modulated laser beams, rather than through confined media (fibre, coaxial, waveguide) or via lower-frequency RF.
Definition¶
The transmitter encodes data onto a collimated laser beam; the receiver is a photodetector pointed at the transmitter; line-of-sight between the two endpoints is required. The link is characterised by its optical wavelength (typically near-IR), beam divergence (narrow for long links to concentrate power on the receiver), pointing-and-tracking tolerance (arcsecond-class for long-range), and link budget (transmit power vs. beam spread vs. receiver sensitivity vs. atmospheric / vacuum losses).
Key differentiators vs. RF:
- Bandwidth density. Optical carriers (~10^14 Hz) give multiple orders of magnitude more bandwidth per link than microwave RF (~10^9–10^10 Hz), enabling tens-of-Gbps to Tbps per beam.
- Narrow beam. Laser beams are spatially confined in a way RF is not; this gives low interference + inherent transmission security, but also requires precise pointing — endpoints must track each other.
- No spectrum allocation. Optical is not subject to the same regulatory spectrum-assignment regime as RF.
- Sensitivity to weather / atmosphere. Terrestrial FSO is degraded by clouds, fog, rain, turbulence; space-to-space FSO in vacuum sidesteps atmosphere — the most attractive deployment regime.
Why it matters for space-based compute¶
Project Suncatcher's modular disaggregated constellation depends on inter-satellite FSO links to behave as a single logical compute substrate. AI workloads on the constellation — training (gradient / activation exchange) and serving (model-parallel intermediate tensors, KV-cache traffic) — require per-link bandwidth comparable to datacenter fabrics. RF cannot deliver the bandwidth density FSO can at satellite-class power budgets.
Google Research's 2025-11-04 announcement (sources/2025-11-04-google-exploring-space-based-scalable-ai-infrastructure) names free-space optical links as the explicit Suncatcher choice:
"Compact constellations of solar-powered satellites, carrying Google TPUs and connected by free-space optical links."
and names "high-bandwidth communication between satellites" as one of three named foundational-research challenges for the programme. The link-budget, wavelength, modulation, and per-link bandwidth figures live in the preprint paper, not in the raw announcement.
Key challenges¶
- Pointing and tracking. Satellites in close formation still move relative to each other; maintaining FSO line-of-sight at arcsecond precision under vibration and orbital motion is non-trivial.
- Link acquisition. Establishing a link initially requires finding the beam in a narrow angular window; acquisition time matters for constellation reconfiguration and link recovery.
- Redundancy. Unlike broadcast RF, FSO is point-to-point — a constellation topology must plan for link failures with alternate paths rather than opportunistic re-routing.
- Terrestrial downlink. Space-to-ground FSO does cross atmosphere and inherits weather / turbulence penalties that space-to-space FSO avoids. The downlink problem is qualitatively different from the intra- constellation problem. (Raw does not address Suncatcher's downlink design.)
Seen in¶
- sources/2025-11-04-google-exploring-space-based-scalable-ai-infrastructure — announced as the inter-satellite fabric for Project Suncatcher; named without link-budget detail (that lives in the preprint).
Related¶
- systems/project-suncatcher — the Google moonshot where FSO is the constellation's inter-satellite network substrate.
- concepts/space-based-compute — the compute-architecture class whose inter-node fabric FSO underwrites.
- patterns/modular-disaggregated-constellation — the architectural-shape pattern for which FSO is the load-bearing connective tissue.