Washington Apple Pi

A Community of Apple iPad, iPhone and Mac Users

electric pi


by David L. Harris

Washington Apple Pi Journal, July/August 1999, p. 41-43, reprint information


SETI = the Search for ExtraTerrestrial Intelligence. So far this has been mostly carried out by looking for unusual radio signals coming from far outside the earth. SETI@home is a project out of the University of California at Berkeley to engage hundreds of thousands of personal computers worldwide to help analyze the data collected by the 1000 foot wide reflector dish of the radio telescope near Arecibo, Puerto Rico.

Radio is the main means of long-range communication that we know about. SETI assumes that any other civilizations will use it too. Looking at the universe in radio waves we discover that it is very noisy over most of the radio spectrum, both due to our own atmosphere and to natural processes in outer space. The amount of noise that might be expected to mask any intentional signal varies by frequency, however; see Figure 1. The SETI radio search looks for signals in a narrow frequency band between the strong natural signals from molecular hydrogen in space at 1.42 GHz and from the OH ion at 1.64 GHz ("the water hole"), frequencies that are assumed to be familiar to any advanced civilization. These have the advantages that they occur near the natural noise low; H and OH produce water, assumed to be needed for any life, and therefore appealing to ETs too; and the frequencies from 1.420 to 1.427 are protected by international agreement from man-made interference due to their importance to radio astronomy. Therefore, although artificial radio signals may not be located there, it has seemed a good place to start.

Figure 1: The radio spectrum and favorable frequencies.

Any radio signal intended to convey information is expected to be confined to a narrow frequency range for the sake of efficiency. That will make detecting it in the presence of broadband noise much easier, both for the intended recipients and for us. The reflector of the radio telescope at Arecibo is a giant dish placed in a natural depression in the terrain, and cannot be pointed. It focuses radio waves on receivers placed 450 feet above the dish, which are mounted on a bow-shaped track which does allow some movement, so the telescope can be "pointed" some 20° from the zenith. For the SETI search, however, the telescope is not operated so as to compensate for the rotation of the earth; any signal from distant space would rise to a peak of intensity and fade in a period of about 12 seconds as its source moves across the telescope's field of view. A narrow-band signal might appear as in Figure 2.

Figure 2: Observing a narrow-band signal.

An additional factor influencing the appearance of a signal is the fact that the moving earth and any home source might be expected to be changing their motion with respect to each other, resulting in a slight change in received frequency (Doppler shift) over the 12 seconds. This would produce a "chirping" sound if converted to an audible signal. A signal might be modulated too. Figure 3 illustrates how one might look.

Figure 3: A pulsed chirped narrow-band signal.


SETI@home looks at about 2.5 MHz of radio frequencies around the 1.42 GHz point. The data is recorded on magnetic tape at Arecibo and sent by mail to Berkeley--the Arecibo facility does not have a high-speed Internet connection. The data is broken up into 256 pieces, each almost 10 KHz wide, to send to individual subscribers. About 107 seconds of this data are sent to each person at a time, to cover several 12-second periods. Along with housekeeping information this makes each "work unit" of data sent at one time about 340K in size, which is small enough to send in a few minutes over an ordinary modem connection. SETI@home applications are available for Windows, Macintosh, Linux, and maybe other operating systems. In use, they connect to the SETI@home Web site, download a work unit, disconnect, and begin processing the data. Each unit requires about 175 billion computer operations. Those start by de-chirping the data in steps of .002 Hz/sec from -5 Hz/sec to +5 Hz/sec, dividing the data into 13.375-second blocks, and examining the data with bandwidths of .007 Hz, for signals above background. Different chirp rates and bandwidths are then examined also. Signals showing a strong power at some particular combination of chirp, frequency, and bandwidth are subjected to tests for terrestrial interference (rising and falling over a 12-second period).

How long does this take on your home computer? SETI@home estimates that each work unit should take about 25 hours of computation time with the average modern home computer with a CPU running at 233 MHz. I found that my 300 MHz G3 runs one in about 18 hours. That's with no other applications working.

Macintosh SETI@home

I first downloaded a beta version (0.28) of the Macintosh SETI@home. Later I got the 1.0 release version. It is a 332 KB Control Panel and can be used as a screensaver or standalone application. It requires a monitor with at least 800 by 600 pixels with 8-bit color. Since I have a 14" monitor and normally run at 640 x 480 pixels, I run it as a separate application, switching monitor resolutions only to run SETI. (I wish it did not require the higher screen resolution.) As a screensaver it has the usual properties of a "hot corner" to enable it, selectable times before it will cut in when you are doing nothing else on your computer, and the ability to blank the screen entirely. Normally it shows a picture (see Figure 4) of its progress in creating a fast Fourier transform of the data for each Doppler shift, bandwidth, etc. SETI can be set to connect to the Internet whenever it is finished with one work unit, in order to send its results and get a new unit, or it can do so only when you tell it to. When it is finished with a unit it will inform you that it needs to connect.

Figure 4: SETI@home output.

In practice the application has performed flawlessly. With the beta version my first unit finished when the progress bar indicated it was only about 50% complete, and SETI@home requested to be connected to the Berkeley site. I sent the data and got a new unit. That one took the full 18 hours, but by the time I had it done, the full release version was out and I was informed that it would not accept my work product! I got a new unit and at this writing am about half-finished with it--I usually run SETI@home when I am finished with my regular computer work. The application operates by itself and does not require understanding of what it is doing. If you are running it and want to quit, simply do so. It stores information in several state files, and upon launching, reads that data and resumes calculations.

I have my Energy Saver Control Panel set to make the monitor go to sleep after a time. I don't know if that speeds SETI operations or not. Posts on Usenet newsgroups indicate that using the screen-blanking option of the screensaver may speed completion of a work unit considerably. Quitting the Finder is also said to speed operations.

At one time, while SETI@home was running and my monitor was asleep, the computer itself shut down. I thought I had set the Energy Saver incorrectly, and restarted the computer. It took a couple of tries to do so, but after I did, I launched SETI@home again and it just started off doing its thing. I soon discovered, however, that we were having a major electrical brownout (I measured the AC voltage at 67 volts at one point), and I quickly quit SETI@home and shut down my computer. The point is that even under these conditions the application behaved without noticeable error. In addition, SETI@home has an elegant look (at least on the Mac). Even the progress bar is detailed; see Figure 5. It builds itself bit by bit! The display is interesting at first also, but it repeats itself and after a while you no longer want to look at it. Of course it is better in color.

Figure 5: Details of the progress bar.

What if you discover ET?

Yes, you will get credit if a signal is found on your watch. Each work unit is individualized and kept track of. There are, however, a number of steps that the data takes after it leaves you and goes to Berkeley. It is analyzed for terrestrial interference (99.9999% of signals, according to the SETI documentation found on their Web site). Test signals injected into the system are removed. Any remaining signals are tested against another Arecibo observation of the same part of the sky, which can take up to six months, since the SETI people do not control all observing time at the telescope. If a signal survives that test another independent group will be asked to observe it. If they too see it, an announcement will be sent worldwide as an International Astronomical Union telegram. It will then be up to others to try to confirm the observations.

Are you needed?

The Arecibo telescope is said to collect more than 200,000 "work units" of data every day. This is too much information for the limited resources of the SETI group on its own. They need all the help they can get. In a few years all the sky that can be observed from Arecibo will be scanned several times. After that, new technologies will probably be available. You can start again.

Note: much of the information and pictures included here were derived from documents at the SETI@home Web site at http://setiathome.ssl.berkeley.edu/.

Return to electric pi

Revised September 6, 1999 Lawrence I. Charters
Washington Apple Pi
URL: http://www.wap.org/journal/