Analysis of the Breakthrough Listen signal of interest blc1 with a technosignature verification framework

Brief

On 2019 April 29, the Parkes 'Murriyang' telescope observed Proxima Centauri using the Ultra-Wideband Low receiver and the turboSETI algorithm (v1.2.2), producing a narrowband event at 982.0024 MHz with average S/N 17.956, a drift rate of 0.0326 Hz/s, and 5.03 hours of persistence that was absent in all nine off-source calibrator panels. Sheikh et al. searched ~7,000 archival Murriyang UWL observations and found 36 morphological 'lookalikes' and 27 'mirrored lookalikes' at other frequencies, all confirmed as RFI by their appearance in off-source pointings; the signal's frequencies cluster around integer multiples of standard digital clock oscillator values. The team conducted three re-observation sessions in November 2020 and a two-year-anniversary session in April–May 2021, detecting no signal at ν_blc1 ± 2 kHz in any of them, and formally classified blc1 as an electronically-drifting intermodulation product of local, time-varying interferers aligned with the observing cadence.

Metadata

Category

Search

Venue

Nature Astronomy

Type

Peer-reviewed

Year

2021

Authors

Sofia Z. Sheikh, Shane Smith, Daniel C. Price, David DeBoer, Brian C. Lacki, Daniel J. Czech, Steve Croft, Vishal Gajjar, Howard Isaacson, Matt Lebofsky, David H. E. MacMahon, Cherry Ng, Imke de Pater, Andrew P. V. Siemion, Claire Isabel Webb, Jamie Drew, S. Pete Worden, Peter A. Gilman, Jane Greaves

DOI

10.1038/s41550-021-01508-8

arXiv

2111.06350

Access

Open access

Length

24.6 M

Programs

Breakthrough Listen, Breakthrough Starshot

Instruments

Parkes 'Murriyang' radio telescope, Ultra-Wideband Low (UWL) receiver, BL Parkes Data Recorder backend, turboSETI v1.2.2

Data sources

Breakthrough Listen archival UWL observations (7,000 files, 2019–2020), NASA Horizons ephemeris, space-track.org orbital elements, ATNF calibrator database, ACMA (Australian Communications and Media Authority) frequency registry

Tags

SETI, technosignature, RFI-mitigation, narrowband-search, radio-astronomy

Key points

• blc1 appeared at 982.0024 MHz with average S/N 17.956, bandwidth < 3.81 Hz, drift rate 0.0326 Hz/s, and persisted 5.03 hours across five 30-minute ProxCen on-source panels while absent in all nine off-source calibrator observations.p.2
• The non-standard observing cadence (30-min on-source vs. 5-min off-source) created an inherent S/N asymmetry factor of √(30/5) × (48/38) ≈ 3, which partially explains why blc1 was not detected in shorter off-source integrations even if a real signal had been present.p.4
• Ground-based vehicles, aircraft, LEO satellites, MEO/GEO satellites, and six tracked deep-space probes (Parker Solar Probe, Gaia, Juno, New Horizons, Ryugu, Voyager 2) were all individually modeled and ruled out as sources; none produced drift characteristics matching blc1's magnitude and temporal morphology.p.5
• The drift rate of blc1 is not consistent with the barycentric motion expected from the direction of Proxima Centauri but is consistent with the order-of-magnitude drift producible by orbital and rotational motions of planets in the ProxCen system.p.7
• Searching 7,000 archival Murriyang UWL observations (2019–2020), the team found 15 similar features in non-ProxCen data; one, appearing 4 days before blc1, was clearly RFI due to persistence across on- and off-source panels.p.7
• Of 111 turboSETI hits drifting proportionally to blc1 at other frequencies, 36 (32%) were visual morphological 'lookalikes'; an additional 27 'mirrored lookalikes' with negative drift rates were identified, all confirmed RFI by appearance in off-source observations.p.9
• The lookalike clusters (Triple Feature and Single Feature sets) have inter-signal spacings consistent with common digital electronics clock oscillator frequencies, 133.33 MHz and 15/128 MHz, matching expected values within 1–1000 Hz.p.9
• No signal was detected at ν_blc1 ± 2 kHz across re-observations on 2020 November 19, 26, and 30 or during a five-day anniversary campaign (2021 April 29 – May 3) that replicated the original cadence, local time, receiver, and backend.p.4

Verbatim

• “blc1 is not an extraterrestrial technosignature, but rather an electronically-drifting intermodulation product of local, time-varying interferers aligned with the observing cadence.”

  p.1

• “We find dozens of instances of radio interference with similar morphologies to blc1 at frequencies harmonically related to common clock oscillators.”

  p.1

• “We find that the drift rate of blc1 is not consistent with the barycentric motion expected from the direction of ProxCen, but is consistent with the order-of-magnitude drift that could be produced in the system, based on the orbital and rotational motions of its planets”

  p.7

• “We can conclusively state that all lookalike and mirrored lookalike signals are RFI due to their appearance in off-source observations.”

  p.9

Most interesting

• Australian Communications and Media Authority records showed no catalogued RFI at Parkes Observatory at 982.002 MHz and no registered transmitters at that frequency in Australia; the band near 982 MHz is formally reserved for aircraft.
• The off-source calibrators (quasars PKS 1421-490 and PKS 1934-638) were 12.57° and 45.82° from Proxima Centauri respectively, far enough that the telescope slew times visible in Figure 1 allowed the local RFI environment to change between pointings, a key confound in the initial sky-localization argument.
• A frequency comb with ~80.1 Hz spacing was present across a 128 MHz sub-band (960–1087 MHz) throughout the entire observation, appearing in both on- and off-sources and in unrelated archival campaigns throughout 2019, yet never correlated with blc1's appearance, determined to be independent RFI.
• The ProxCen campaign used an asymmetric 30-min on / 5-min off cadence chosen for stellar-flare science, not SETI, a design choice that inadvertently reduced the sensitivity for detecting blc1 in off-source panels by a factor of roughly 3 and contributed to the initial appearance of sky localization.
• turboSETI had already rejected 111 of 112 proportionally-drifting hits as RFI at an early pipeline stage; the 32% that were visual lookalikes to blc1 were identified only through manual inspection of the full cadence waterfall plots.
• blc1 is described throughout as a 'signal of interest' rather than a 'candidate', the authors explicitly note the terminological distinction, arguing 'signal-of-interest' is more appropriate given the low prior probability of an ETI detection, even though 'blc1' as a label was already in common usage before the paper was published.