I. How Aliens Think
If you were a member of an alien civilization trying to communicate across the immeasurable distances of space, how would you go about it? Not being an alien yourself it is hard to answer this question, but you can do your best: You would, most likely, send out radio signals, since they are extremely fast – traveling at the speed of light – have a very long range, and are relatively easy to transmit. Your radio transmission, furthermore, would probably be a narrow-band signal, to distinguish it from other transmissions in neighboring bands as well as naturally occurring noise. In other words, if you wanted to communicate across interstellar space you would almost certainly use the exact same technology that has worked so well for us in the much shorter distances of Earth: continuous narrow-band radio transmissions. Wouldn't aliens do the same?
Perhaps they would. That, at least, is what most SETI researchers have assumed since the earliest days of the field half a century ago. As a result, the majority of SETI searches over the years have concentrated on finding precisely this kind of signal – a clear and crisp needle of a narrow-band transmission buried in the haystack of broadband cosmic noise. [email protected] itself is typical in this regard, devoting most of its enormous processing power to carving out the raw noise from Arecibo into thin bands of data where a narrow-band signal could be hiding.
But then again, maybe they wouldn't. Perhaps the aliens, for their own reasons, would choose to communicate using a very different type of signal. For example, instead of sending a continuous narrow-band transmission they might choose to send distinct broad-band pulses. These would stand out against the background noise not because they are precisely centered on a particular wavelength, but because they are very short and punctuated bursts of energy. Why would the aliens choose this method over our own? Who knows, after all we are not aliens and cannot begin to imagine the technological choices they face. The main thing is to acknowledge that such a form of communication is possible, and just as practicable as our familiar narrow-band radio transmissions. And if aliens could be sending out this type of signals, then it follows that SETI researchers should be on the lookout for them.
With this thought in mind, [email protected] Chief Scientist Dan Wethimer and his team have worked hard for several years to develop a new type of [email protected] with new capabilities. Like traditional [email protected] the new program uses the raw data collected during sky-surveys at Arecibo. As before, the data is carved up into work units and sent to users for processing, and users' computers then send their results back to [email protected] headquarters in Berkeley. The difference is that this time, instead of looking for clear narrow-band transmissions, the software will search for extremely short broad-band bursts, or "pulses," coming from the stars. To distinguish it from traditional [email protected], the team also gave the new project a distinct name: Astropulse.
II. Reconstructing an Alien Signal
"Searching for a short broad-band signal is a completely different process than searching for a traditional narrow-band signal" explained Josh Von Korff, the [email protected] team member who was responsible for programming Astropulse. Traditional [email protected] looks at a radio band around the hydrogen line, between 1418.75 MHz and 1421.25 MHz, but the program does not examine the entire 2.5 MHz wide band all at once. Instead it slices the raw data into band segments as thin as 0.07 Hertz apiece in search of a narrow-band signal. The challenge is then to reconstruct the original signal by compensating for the Doppler drift caused by the relative motion of Earth and the originating planet. Since that motion is not known, the program runs through a gamut of different possibilities, trying out a wide range of different drift rates in search of an actual signal.
The Astropulse program also looks at the same 2.5 MHz band around the hydrogen line, but it spends no time trying to compensate for Doppler drift. This is because Astropulse is looking for signals that would cover the entire bandwidth of 2.5 MHz – that is two and a half million Hertz – more than thirty million times broader than the finest traditional [email protected] band. Any Doppler drift in the signal would fall within this wide band anyway and will form part of the total signal. As a result, there is no need to compensate for the drift as is the case with a narrow-band signal.
But although Astropulse does not need to concern itself with Doppler drift, it does have to worry about a different problem that does not arise in traditional [email protected] This is the inconvenient fact that electromagnetic waves, including radio signals, travel at slightly different speeds through space, depending on their frequency. As we learned in school, radio signals all travel at the speed of light, but this is literally true only in an absolute vacuum. When traveling through a medium higher frequency waves travel ever so slightly faster than lower frequency ones. In light waves we know this effect well as refraction, the familiar effect where a beam of white light is divided into its component colors when passing through water or a prism. This is caused by the fact that the different colors, representing different wavelengths, pass through the medium at slightly different speeds.
At first glance it would seem that this phenomenon would hardly affect alien transmissions through space. Light might be affected by water or prisms, but isn't interstellar space emptiness itself? As it happens, the answer is no. When compared with our dense Earth environment, interstellar space certainly appears empty, but is in fact far from a true vacuum. It is mostly filled with varying concentrations of free-floating hydrogen atoms, composed of a single proton and a single electron. In many of these the proton and electron have become separated, resulting in free-floating charged particles called ions. All together, the atoms, ions, and free electrons form the "interstellar medium" through which radio signals must pass.
Now as long as traditional [email protected] searches for a narrow band signal, this is not a problem. Since the entire transmission is concentrated at one narrow frequency, all of it travels at the same speed and arrives on Earth at the same time as a single coherent signal. But Astropulse searches for broad-band transmissions that are spread across a 2.5 MHz band of the spectrum. We can think of such a transmission as a combination of many narrow band signals at adjacent frequencies, all broadcast simultaneously as a single broad-band signal. Because of the differing speeds at which the different frequencies travel, however, the high frequency portions of the signal will arrive on Earth before the lower frequency portions. This means that a broad band pulse that was strong and coherent when it set on its way will be smeared across a time-span of several milliseconds when it is received on Earth. No clear pulse will be evident, and the transmission would be easily lost within the background noise.
The first task for Astropulse then is to reverse the smearing effect and reconstruct the original strong signal. To do this Astropulse the Fast Fourier Transform (FFT) algorithm, the same one used by the traditional [email protected] program. FFT divides the raw data into thin narrow-band slices, which it then recombines with other portions along a timeline. A slice containing the longest wavelength is combined with a slice of slightly shorter wavelength that was received just before it, and so on, step by step, until the shortest wavelength signal, which arrived earliest, is added in. If a strong pulse had been sent in the first place, then the combination of all these slices will reconstruct it and the signal will appear loud and clear.
There is however a serious flaw in this method. In order to properly reconstruct a signal in this manner, we would have to know the exact time-lag between the highest frequency portion and the lowest frequency portion of the signal. If for example the actual time-lag on a signal is 4 milliseconds, but Astropulse combined a highest and lowest frequency bandwidth slices that were received only 1 millisecond apart, then no pulse will be registered.
The only way to reconstruct a broad-band signal is to add its narrow-band components together by taking into account the correct time-lag between them. This time-lag depends on the distance the signal has traveled through the interstellar medium: the longer the distance, the greater the time lag. Unfortunately we have no clue where an alien civilization might be located, and what distances its transmissions must cover before being received on Earth. Not knowing the distance, we don't know the time-lag in the signals, and cannot reconstruct the aliens' transmission.
Astropulse's solution to this problem is to try out a whole range of different possible time-lags, one after the other. In each case Astropulse processes the entire work unit looking for a broad-band signal by combining narrow band signals at a particular time interval. The shortest time-lag that the program tries between the highest and lowest frequency slice is 0.4 milliseconds, and longest is ten times greater – 4 milliseconds. Between these two extremes, Astropulse processes each work unit nearly 15,000 times!
III. How Long is a Short Signal?
Processing each chunk of data from beginning to end that many times requires a gargantuan amount of computing power, which would be unthinkable for most scientific projects. Only [email protected], with its millions of volunteers around the world running the program on their home computers can conceive of analyzing each chunk of data with such depth and precision. But even this is not enough: after all this work it is still possible that we would miss the broad-band pulse the aliens sent our way if we did not know how long the original signal lasted.
For example, suppose the aliens sent a signal 10 microseconds long, but we were checking for signals only 1 microsecond long. In that case we would never add up all the parts of the signal at the same time, and would never see the clear spike that tells us that a pulse from outer space has been received. The reverse is also true: if we were looking for a relatively long signal, while the actual signal lasted only a fraction of that time, it is likely that the pulse would disappear into the background noise and never be detected. All of which is to say that in order to find a signal in the data that lasts a certain amount of time, we have to be looking for a signal that lasts that amount of time – or close to it.
Unfortunately, just as we don't know where the aliens are and how far their signal must travel, so we have no way of knowing how long their signal would last. And so, once again, Astropulse tries a whole range of possibilities one after the other: beginning with the shortest pulse of 0.4 microseconds, it tests for 9 additional lengths of time, each one double the previous length (that is 0.4 microseconds, 0.8 microseconds, 1.6 microseconds, 3.2 microseconds, etc.). Astropulse tests all ten of these possibilities each and every time it processes the entire set of data to account for a different possible time-lag.
To recap: Astropulse processes the entire set of data nearly 15,000 times, each time assuming a different time-lag between the highest and lowest frequency portion of the signal. Each and every time the program completes one of these 15,000 cycles it goes over the processed data ten times looking for signals of different lengths. The amount of computer time involved would indeed be unimaginable for any project other than [email protected]
IV. On Aliens and Black Holes
As an integral part of [email protected], Astropulse is first and foremost a search for an intelligent transmission from outer space. Nevertheless, as the [email protected] researchers are quick to admit, there is really no telling what Astropulse will actually find. After all, nothing resembling such a systematic all-sky search for a broad-band signal from space has ever been attempted before, so scientists really don't know what's out there. Will Astropulse finally detect an elusive signal from an alien civilization? Or will it, perhaps, discover a natural source of broad-band radio pulses?
Dan Werthimer and his group have thought carefully about this issue, and came up with several possible natural sources for Astropulse signals. One possibility is pulsars – rotating neutron stars that emit strong radio transmissions. Known pulsars rarely produce signals shorter than 100 microseconds, but it is possible that Astropulse will discover a new class of pulsars with much shorter transmission times.
A more exotic possibility is that Astropulse would register the "dying gasps" of exploding black holes. Astrophysicist Martin Rees has theorized that black holes that explode through Hawking radiation would produce a strong but brief burst in radio frequencies, and this could potentially be detected by Astropulse. And then of course there is the possibility that Astropulse will discover something new entirely, that we cannot imagine beforehand. This, after all, might be the likeliest outcome.
Like all [email protected] data, Astropulse data is collected at Arecibo during sky surveys conducted by the ALFA consortium (Arecibo L-band Feed Array), using the radio telescope's multi-beam receiver. The data is recorded and then packaged into work units of 8 MegaBytes each, which are sent out to users all over the world for processing. Since the the Astropulse software downloads automatically onto volunteers' computers, users don't have to take any action in order to join the broad-band search.
The first Astropulse work units went out in early August, and overall users will not see a significant change in the way [email protected] operates on their computers. At 8 MegaBytes Astropulse work units are larger than traditional [email protected] units, and as we have seen they undergo particularly intensive analysis. As a result users will notice that they take longer to process on their computers. Meanwhile traditional [email protected] work units will continue to go out alongside Astropulse units, and they will continue to be processed on users' computers just as they had before.
Astropulse is now off and running in search of brief broad-band radio signals coming from space. What will it find? Will it be the long sought signal from an alien civilization? Will it detect new pulsars, black holes, or perhaps some novel natural phenomenon of which we have no inkling? We don't yet know. But like Galileo who four centuries ago turned a telescope upon the night sky, Astropulse is looking at the heavens in a new and unprecedented way. Who knows what wonders it will reveal.