In the quiet 2025–2110 MHz uplink band a range where satellites are supposed to listen, not speak more than 170 classified Starshield spacecraft were detected transmitting powerful downlinks toward Earth. The finding, accidentally made by Canadian amateur astronomer and engineering technologist Scott Tilley, exposed a deliberate reversal of standard satellite communications practice and raised urgent questions about compliance with International Telecommunication Union regulations.
Starshield is a government-operated offshoot of the Starlink program at SpaceX, built for the US National Reconnaissance Office and US Space Force. Where Starlink’s commercial network operates at much higher frequencies to deliver broadband internet, the signals from Starshield occupy a low S-band segment, generally reserved for Earth-to-space uplinks. That band is used for spacecraft command and control by agencies like NASA and NOAA, as well as terrestrial services such as broadcast auxiliary links, according to ITU frequency allocation tables. Using it for continuous space-to-Earth transmissions without international coordination risks causing interference to legitimate operations.
Tilley’s measurements revealed emissions up to 9 MHz wide, with a signal-to-noise ratio of around 10 to 20 decibels-meaning the power in the received signal was between ten and one hundred times that of the noise in the same bandwidth. He verified the sources by matching the Doppler shifts in the signals to orbital data that global amateur satellite trackers have been collecting. And these emissions have been detected over Canada, the United States, and Mexico; given Starshield’s global coverage, they’re likely over other nations, too.
The ITU’s Radio Regulations say that uplink bands, such as 2025–2110 MHz, are reserved for transmissions from Earth to spacecraft, with only limited use for space-to-space transmissions. Article 4.4 says that any other use must operate on a “non-interference basis,” which cannot claim protection against interference. Yet Tilley could find no evidence of formal ITU coordination for Starshield’s downlinks. Canadian regulators conceded that no coordination process had been started, and the ITU said it could only act if a national administration filed a formal interference complaint.
From an engineering perspective, this anomaly is quite significant. Downlinking in an uplink band risks radio-frequency interference in nearby spacecraft receivers, which may result in missed or ignored commands. Though no such incidents have been publicly reported, the technical risk is real: uplink receivers are tuned for weak, targeted signals from Earth, not strong, wideband transmissions from neighboring satellites. This may propagate throughout several systems during dense orbital shells.
The Starshield case also connects to a larger issue: UEMR originating from large satellite constellations. Initial studies with the LOFAR radio telescope revealed that SpaceX’s second-generation Starlink satellites emit UEMR up to 32 times stronger than first-generation models in some protected radio astronomy bands. UEMR, however, unlike out-of-band emissions from intentional signals, is not regulated by ITU interference limits, and thus observatories are left unprotected. With the number of satellites reaching up to an estimated 100,000 by 2050, aggregate UEMR would make certain ground-based radio astronomy impossible.
Starshield’s use of the uplink band may be unrelated to UEMR, but both issues illustrate deficiencies in spectrum governance. Existing EMC standards for satellites, such as MIL-STD-461 or ESA ECSS-E-ST-20-7C, are designed to prevent self-interference within the spacecraft rather than limit emissions that could impact external systems. In the absence of binding international requirements concerning UEMR or other unconventional frequency use, operators can deploy systems that pass domestic reviews while bypassing global coordination.
Experts say the NRO would have almost certainly obtained permission for Starshield’s downlinks from the NTIA, which regulates US civilian use of the radio spectrum, but such permission does not apply outside the United States. Without transparency, other nations can neither estimate nor take steps to mitigate such interference, and spacecraft operators uninformed of the emissions may not monitor that band at all. The combination of classified missions, atypical frequency use, and rising background radiation due to mega-constellations underlines a critical engineering and policy challenge: how to preserve an orderly, predictable radio environment in low Earth orbit. As Tilley put it, “This deployment is not one or two satellites, it is a distributed constellation of hundreds of objects with potential global implications.”
