SETI imaging pipeline keeps pace with the telescope

Karen Perez and colleagues built a GPU-accelerated pipeline that turns Allen Telescope Array interferometric visibility data into images, suppresses contaminated frequency channels, and searches many image pixels for dispersed radio pulses. Imaging adds location information that a conventional beamformed time-frequency search does not automatically provide, helping distinguish a celestial candidate from local interference.

In a controlled ten-minute recording, the system recovered a periodic injected pulse at the correct phase-center position and dispersion measure. Its core science processing took about 63 milliseconds per integration against a native cadence of about 98 milliseconds, leaving enough time for the main operations. The test replayed recorded data and relied on statistics calculated over the full observation; live operation will need causal estimates based only on information available at that moment. No extraterrestrial signal was detected.

Why it matters

Fast, spatially resolved searches can make follow-up more immediate and rejection of terrestrial interference more defensible—two essentials if a genuinely interesting transient appears only once.

What the source establishes

Fresh methods preprint submitted October 5. The recovered event was deliberately injected to validate the pipeline; real-time live-telescope operation remains a next step.