Starlink Signal Leakage Threatens Radio Astronomy's Most Critical Frequencies

Astronomers spent decades designing the Square Kilometre Array Low (SKA-Low) to perceive the faintest susurration successful the observable beingness — neutral hydrogen signals from the cosmic dawn, erstwhile the first stars flickered connected astir 13 cardinal years ago. What they’re picking up alternatively is electronic leakage from SpaceX satellites. Not the broadband beams Starlink intentionally transmits. The accidental fixed bleeding from onboard hardware, coupling done outer structures and radiating crossed frequencies cipher authorized.

A Curtin University squad utilizing the Engineering Development Array 2 — a prototype SKA-Low position successful Australia — analyzed astir 76 cardinal power images complete 29 days. The results, published successful Astronomy & Astrophysics, are uncomfortable reading. Researchers catalogued 112,534 individual power emissions from 1,806 unsocial Starlink satellites crossed 73–235 MHz, the nonstop wave scope SKA-Low needs. Some satellites emit periodic 13-kHz tones astatine astir 137 MHz each 100 seconds. These aren’t scheduled downlinks that operators tin scheme around. They’re unpredictable hardware leakage — and researchers cannot subtract a awesome they cannot model.

The numbers show the communicative bluntly:

• 112,534 emissions detected from 1,806 unsocial Starlink satellites
• Leakage reaching up to 10⁶ Jy/beam; early-universe hydrogen signals require sensitivity adjacent 10⁻⁵ Jy
• Starlink signals reportedly astir 10,000 times stronger than the cosmic signals SKA-Low is built to detect
• Up to 30% of images astatine immoderate frequencies contained Starlink interference
• Emissions recovered wrong 2 ITU-protected bands — 73–74.6 MHz and 150.05–153 MHz — wherever signals aren’t expected to beryllium astatine all

“Comparable to the brightest earthy power sources successful the sky.” — Steven Tingay, Curtin University, connected Starlink’s leaked emissions

What SKA-Low is hunting demands almost incomprehensible sensitivity. Trying to observe those ancient hydrogen signals pinch Starlink overhead is for illustration attempting to perceive a speech from 13 cardinal years agone while your neighbour runs a subwoofer astatine afloat volume. The contamination doesn’t conscionable degrade information value — it risks rendering full wave bands scientifically unusable.

The Rules Don’t Cover This

Current world frameworks time off unintended outer emissions successful a regulatory grey zone, moreover erstwhile those emissions onshore wrong protected astronomy bands.

Here’s the structural problem: cipher is technically breaking immoderate rules. The International Telecommunication Union (ITU) protects definite power astronomy bands from intentional transmissions. Unintended electromagnetic radiation — hardware leakage — sits mostly extracurricular that framework, legally invisible moreover erstwhile it bleeds into protected spectrum. Think of it arsenic the power balanced of sound ordinances that screen deliberate sound but disregard a neighbor’s HVAC system, nary matter really large it runs astatine 3 a.m.

SpaceX has engaged constructively earlier — outer visors reduced optical brightness, and an NSF coordination statement addressed higher-frequency power bands. But beam guidance doesn’t hole low-frequency hardware leakage. The ITU is reportedly discussing the issue, though chat isn’t regulation, and the constellation already exceeds 6,000 satellites. Researchers shared their findings pinch SpaceX, who are reportedly unfastened to speech connected early hardware changes. Algorithmic mitigation is being explored, but scientists picture it arsenic “embryonic” — perchance demanding computing powerfulness that rivals the subject processing itself.

Engineering fixes are the existent answer, the aforesaid measurement creation changes addressed optical brightness. The mobility isn’t conscionable what’s happening to astronomers — it’s who decides what the power entity is worthy protecting, and whether that determination gets made earlier the model closes.