A star is born

The brainchild of David Gedye in 1995, SETI by distributed computing began its life. Gedye wanted to examine the radio data in detail that had never been attempted before, but the power and cost of supercomputers was prohibitive. He figured that instead of employing dedicated supercomputers at the radio telescope sites, there were thousands upon thousands of idle computers available through the Internet whose central processing unit (CPU) could be harnessed and put to work. He envisioned and created the SETI@home project over the course of the next four years. In May 1999 “volunteer computing” became reality.

How does it work?

SETI looks at a slice of the radio band that lies between 1418.75 MHz and 1421.25 MHz, or about 2½ MHz of bandwidth, all told. This information is in a continuous stream from the Arecibo radio telescope in Puerto Rico, and amounts to 35 GB of data every day. The telescope lacks a high speed data connection and so instead the data is stored on a Compaq Tape or “DLT” (Digital Linear Tape) and sent by regular mail to Berkeley, in California.

The data is then subdivided into small portions named “work-units” comprised of 107 seconds of observation along with parallel data about the work-unit so that the total comes to 340 kB of data per work unit. Each one of these units is sent out several times to different people; this assures that the results are consistent, and protects the data against loss with a crashed computer, or someone not returning a data chunk for too long. The beginning and end of each unit is overlapped with others to make sure nothing is missed at the transition point.

There’s always one in the crowd

Some jokers have attempted to pollute the data by editing the information to create an artificial “alien” signal, but ridiculous results that are inconsistent with all the other samples are quickly weeded out, so there is no point. Even if you had thousands of friends, it is mathematically unlikely that any of them would receive the same work-unit to analyze.

The system allows plenty of time to complete the computations, since if a computer were to work exclusively on one work unit it would take between 10 and 50 hours to complete. The system however is designed to run when you’re not using your computer, when it is idle with the screensaver on. As soon as you need your computer, the program backs out gracefully, without interfering with anything, as if it weren’t even there.

There is also an option to dedicate one or more processors (in our multiprocessor world) to the task and let it run all the time, even when you’re using your computer. If the demands of your computing require more of your CPU’s functionality, the program backs out gracefully, surrendering full control of the processors to your computer.

Why do they need so much computing power?

Consider this: Arecibo is fixed to the ground. Unlike the radio telescopes that we’re used to seeing that are steerable, and mounted on towers, and in some cases even on railway cars so the size, shape, and direction of the telescope can be altered, Arecibo’s dish itself doesn’t move.

There is a smaller dish antenna at the focal point of Arecibo’s dish which is suspended by cables between three towers surrounding the giant telescope. By adjusting the cables its aim can be altered, therefore changing its target. Arecibo, however, generally points in one direction. This means that a single point in the sky passes in front of the telescope and will take 12 minutes from the time it enters the view until it passes the center and exits its view.

This is convenient if we happen to be receiving a radio signal from an intelligent species. The power of the signal will increase over the course of the first 6 minutes, peak, and then decrease over the latter 6 minutes. The pattern should stand out fairly clearly among the background radiation. If it were receiving a series of modulated signals it would look like this image.

We’re moving Fast

Our planet clips along at 30 kilometres per second in its travels around the Sun. Our whole solar system clips along a 250 kilometres per second in its orbit around the galaxy. An ET radio source would obviously be moving at a quite different speed. This would result in a Doppler shift of the radio signal.

Over the course of our observation the frequency would either rise or fall depending on our relative motion to the source. This is called a “chirped signal” and it would look like the image below. Unfortunately the data doesn’t actually look as clear as this.

This portion of the spectrum around 1420 MHz was chosen for its clarity, but there is still plenty of background noise, so each signal must be de-chirped. This takes an immense amount of mathematical calculation to achieve—on the order of 200 billion calculations just to make the data useable. It starts at .002 Hz, then repeats the work at .07 Hz, then .15 Hz and keeps doubling it, up to 1200 Hz. This pushes the grand total of operations up to 275 billion on that 107 seconds worth of data! And that only finds Continuous signals. Pulsed signals take a whole new operation that is more complex than there is room to explain in this space.

Suffice to say: ultimately a single work-unit will require between 2.4 trillion and 3.8 trillion mathematical operations to complete. And the person whose computer detects the signal will be named as co-discoverer of the phenomena. All this can happen while you are away from your computer, as it relentlessly grinds the data and searches for evidence that we are not alone in the our local area of the Milky Way.

Do you want to be the person that helps find E.T. and go down in history as one of its discoverers? All this can be yours by visiting the screensaver download site and signing up to help.

But wait there’s more

Maybe SETI@home is so popular that there are no work-units to download… What will you do? Just go to this site and look at all the different projects that are available to run on the SETI@home Berkeley Open Infrastructure for Network Computing (BOINC) software. There are distributed computing projects available for just about every scientific interest.

You can help search for Earth-threatening asteroids, or help with the results for the CERN particle accelerator. You can contribute work in physics, biochemistry, molecular biology, computer science, climate study, working on intractable mathematical problems, aerospace, or other astronomical problems. You can help with questions about physiology, astrophysics, cryptography, chemistry, epidemiology, artificial intelligence, environmental research, seismology, and much more.

If your computer is running and you’re not using it, why not lend it to some needy scientists that could really use some of those wasted clock-cycles? You can do the same thing they do every day… try to take over the world …er… help improve the world and make it a better place for everyone!