Working alongside Nvidia and using technology deployed on the SETI Institute’s Allen Telescope Array (ATA) in California, the team has achieved a 600-fold performance improvement in detecting fast radio bursts (FRB), which are mysterious, high-energy signals from deep space.
The breakthrough, detailed in the journal Astronomy & Astrophysics, uses Nvidia’s Holoscan platform, which is purpose-built to process massive real-time data streams. Traditionally, FRB searches have been slowed by a process known as “dedispersion”, which involves scanning through thousands of possible signal parameters to correct for frequency-dependent time delays.
The new AI-driven architecture removes that bottleneck entirely, allowing signals to be analysed instantly, transforming how astronomers search for transient and potentially artificial signals from space.
At the ATA, the existing state-of-the-art pipeline currently takes around 59 seconds to process 16.3 seconds of data, running almost four times slower than real time. The new system completes the same task 600 times faster, operating more than 160 times faster than real time.
“This represents a paradigm shift in how we search for fast transient phenomena across the cosmos,” said Peter Ma, who led the research while an undergraduate at the University of Toronto and is now a graduate student at the University of California, Berkeley. “We’ve created a system that can outpace enormous data streams while retaining the sensitivity needed to detect the unexpected.”
The new system is not only faster but also 7 per cent more accurate, reducing false positives by nearly tenfold compared to current pipelines – a crucial improvement for large-scale surveys that must sift through millions of candidate signals. This precision allows astronomers to conduct rapid follow-up observations, vital for identifying FRB sources and for detecting technosignatures – possible evidence of intelligent life.
“This technology doesn’t just make us faster at finding known types of signals, it opens the door to discovering entirely new signal morphologies,” said Dr Andrew Siemion, the Bernard M Oliver chair for SETI at the SETI Institute and principal investigator for Breakthrough Listen. “An advanced civilisation might use burst-like or modulated transmissions we haven’t even imagined. This AI system can learn to recognise patterns that human analysts might overlook completely.”
During testing, the system successfully detected giant pulses from the Crab Pulsar, managing an 86 gigabit-per-second data stream with ease. Its scalability means it could be deployed across observatories worldwide, forming a global, real-time detection network for both natural cosmic events and potential extraterrestrial transmissions.
The project represents years of collaborative effort between Breakthrough Listen, the SETI Institute, and partners from academia and industry, including Nvidia. Together, they’re pushing the limits of what real-time AI can achieve in astronomy and in humanity’s ongoing search for life beyond Earth.