Create a free profile to get unlimited access to exclusive videos, sweepstakes, and more!
Remember that scene in E.T.,The Extra-Terrestrial when the homesick and confused alien, wanting to phone home and trying to make sense of Earth technology, puts together a mashup of several human artifacts into what he thinks is a device for contacting his fellow extraterrestrials?
Earthlings are just as mystified as E.T. when trying to imagine what a message from (hypothetical) aliens could possibly look like. Everything from radio signals to gamma rays to interstellar objects like ‘Oumuamua has been wondered at, but nothing has been proven. There is no way to duplicate a signal beamed at us by some (hypothetical, again) advanced civilization thousands, or even millions, of light years away if we don’t even know whether or not such a civilization exists. How would it appear? There is a way to at least get some idea.
We may not know exactly what sort of message other intelligent beings in the universe would send us — or even if we’d recognize it as such. But astronomer Bryan Brzycki of UCLA has now figured out how to simulate possible ETI (extraterrestrial intelligence) signals that could show up in SETI (search for extraterrestrial intelligence) investigations. He led a team of alien seekers, from institutions including the SETI Institute and Breakthrough Listen, in developing Setigen, a machine learning tool that can be used with AI algorithms to predict what we should look out for.
“In SETI, we are looking for signals that haven’t been discovered yet, so we can’t create datasets full of technosignatures,” Brzycki told SYFY WIRE. “The next best thing is to create synthetic signals, add them to the background of real observational data, and see if our algorithms can find them.”
Setigen was originally developed for training AI to find narrowband radio signals. Because so many narrowband signals are produced on Earth, and they don’t require much energy, radio SETI continues to look for them. We are obviously going to look for what we already know. What alien signals could look like — from frequency to shape and everything in between — is unknown, which is why SETI searches need something to focus on. That something can only be guessed at. Setigen itself doesn’t search for signals, but what it can do is synthesize them.
By creating synthetic alien signals to be detected by AI algorithms, Setigen could help scientists identify a message if E.T. does phone home. These narrowband signals can be customized with the tools Setigen provides for potentially endless predictions of what a signal zapped here by extraterrestrials could look like. Features like intensity and frequency can all be adjusted before the artificial signal is plugged into the same radio data formats used by Breakthrough Listen.
New search algorithms are being developed using Setigen, and it is also used with the TurboSETI algorithm, which is a version of the incoherent tree deDoppler algorithm, that hunts for potential technosignatures coming our way. The issue with TurboSETI is that it automatically assumes a possible techno-signal is continuous. It might not recognize other types of signals, but Brzycki believes that it is still a decent takeoff point for extraterrestrial signal detection.
“The deDoppler method searches for signals by correcting the data over a set of trial frequency drift rates and seeing which drift rate maximizes the detected signal-to-noise ratio (SNR),” he said. “Only signals that exceed the threshold we set for any drift rate we try are identified.”
Frequency drift happens when there is a progressive, though unwanted, change in frequency over time. It can be caused by an unstable transmitter or environmental factors that get in the way. If intelligent aliens exist, they probably experience glitches. The deDoppler algorithm is still the primary way that SETI tries to make out what could be spectral signals, since genes are in the Doppler effect, which makes it seem that there is a difference in the frequency of light or sound waves when they leave their source and when they reach the observer. This is caused by motion that increases or decreases the distance between the observer and the source.
With this in mind, Doppler drifting occurs because the frequency of a constantly accelerating signal will often vary over time. Say that alien signals are being sent from some planet or spacecraft that is orbiting a star. Because the body they are emitting these signals from is moving, the frequency will vary, though it can look the same over very short stretches of time, about a few minutes or less. Another algorithm used in SETI (and audio processing in the human realm) is a Fast Fourier Transform. It can tell which parts of a signal show up in different spectra, which gives away that signal’s frequency, and helps figure out signal intensity, measured as voltage.
“We use Fast Fourier Transforms (FFT) of the time series voltage data to calculate the power of each frequency towards the input time series data,” said Brzycki. “We take a section of the voltage time series, get the FFT, then repeat this process over the entire observation.”
Though we still haven’t gotten a message or a response from any other intelligent species, Setigen will continue evolving. Whether the synthetic algorithms it allows scientists to create are realistic or not is still a question floating beyond Earth. There may be more noise to actual alien signals, if we ever do pick up on one, and they could be more or less complex than we imagined. Setigen can possibly expand to make it possible to synthesize both noisier signals and also more types of radio signals, such as broadband. Brzycki is optimistic.
“As more research is done to search for different signal morphologies, Setigen can also expand to synthesize these kinds of signals,” he said. “As an open-source library, it’s conducive to further improvement and collaboration as the field of radio SETI continues to develop.”