The link budget
Aperture product over wavelength times range.
For a diffraction-limited optical link, received power equals transmitted power times the square of the product of transmit and receive aperture diameters, divided by the square of the product of wavelength and range.
Take a generous gram-class craft: one watt transmitted, the 4-meter sail reused as the transmit aperture, 1.06 micron wavelength, and 4.02 × 10¹⁶ meters of range.
| Receive aperture | Received power | Photon rate | Practical data rate |
|---|---|---|---|
| 30 m | 7.9 × 10⁻¹⁸ W | 42 photons/s | about 4 bit/s, 17 MB per year |
| 100 m | 8.8 × 10⁻¹⁷ W | 472 photons/s | about 47 bit/s, 186 MB per year |
| 300 m | 7.9 × 10⁻¹⁶ W | 4,246 photons/s | about 425 bit/s, 1.7 GB per year |
Modeled, assuming ideal photon counting, a background-free channel, and roughly 10 received photons per delivered bit as a coding and detection margin. Received power scales with the square of receive aperture, so this table is the single strongest lever in the entire mission design.
Read the first row carefully
This is the honest Starshot baseline.
42 photons per second. Four bits per second. 17 megabytes over a year of continuous transmission — which requires the craft to survive a year past its encounter, remain powered by a milligram-scale radioisotope or betavoltaic source, and hold its aperture pointed at a 30-meter target 4.25 light-years away with no station-keeping, after 21 years of proton bombardment and dust erosion.
That is enough for spectra and a handful of images. It is nothing like enough for a survey.
And it is, by a wide margin, the least-solved element of the beamed-sail architecture. The published literature devotes perhaps a tenth of the attention to the downlink that it devotes to the launch, and the ratio should be closer to reversed.
The fix is on this end
Infrastructure built for other reasons.
Received power scales with the square of the receive aperture. Tripling the receiver from 30 to 100 meters multiplies the data rate by 11. Going to 300 meters multiplies it by 100.
The interstellar communications problem is solved at the home end, not the far end, and the solution is infrastructure the discovery economy builds anyway. Sparse phased optical arrays with effective apertures in the hundreds of meters are already placed in the outer system for interferometric astronomy and exoplanet imaging. Pointing one of them at a returning probe costs observing time, not capital.
This is the cleanest example anywhere on this site of the general pattern: the interstellar program does not require its own infrastructure, it requires priority access to infrastructure built for other reasons.
The other fix
Eight orders of magnitude, from a one-meter telescope.
There is a second lever on the receive aperture, and it is large enough that it cannot be treated as a subsection.
Light from a distant source passing near the Sun is bent by solar gravity and comes to a focus along a line beginning at 547.6 AU, with an optical amplification on the order of 10¹¹. A meter-class telescope placed on that line collects like an aperture hundreds of kilometers across.
| Receiver | Photon rate | Practical data rate | Data per year |
|---|---|---|---|
| 30 m conventional telescope | 42 per second | 4 bit/s | 17 MB |
| 300 m sparse array | 4,246 per second | 425 bit/s | 1.7 GB |
| 1 m telescope on the SGL focal line | 5.5 × 10⁹ per second | 550 Mbit/s | 2.2 PB |
There is no other single decision available anywhere in the interstellar case that changes a mission parameter by eight orders of magnitude. If a probe program is going to happen at all, the receiving node is not an accessory to it. It is the larger half of it.
The full treatment is on the solar gravitational lens page, including the three caveats that keep it honest.
One more rung on the autonomy ladder
Seven thousand times the last step.
This site’s light-lag thresholds run: below Ceres, human supervision is degraded but feasible; at Ceres, about 20 minutes round trip, it reaches the edge of viability; beyond, autonomous decision authority is architecturally required rather than merely convenient.
Pluto is about 11 hours round trip. Proxima Centauri is 8.5 years.
The step from Pluto to Proxima is roughly 7,000 times larger than the step from Ceres to Pluto. Every decision made about the craft after launch, if made from Earth, is made on information 4.25 years old and takes effect 4.25 years later. There is no supervision, no correction, no anomaly response, and no possibility of any. Full autonomy is not a design feature here. It is the only operating mode that exists, and it must hold for 21 years through a radiation environment that destroys unshielded silicon, with no maintenance and no reboot from the ground.
This is the requirement about which there is the least doubt. Autonomy at this level is not a stretch from the 2075 baseline. It is a straightforward extension of systems that will by then have operated unsupervised at Callisto for decades. Of everything in the interstellar case, the thinking part is the part we are most confident about.