developing and testing space-based hardware for more sensitive and extensive systems, if they are needed. The four sites are:

1. On the far side of the Moon

2. In Earth orbit around the Sun and 60° from the Earth as viewed from the Sun

3. At lunar distance at L-3, on the opposite side of the Earth from the Moon

4. In synchronous orbit around the Earth and at the longitude of the data processor

Site 1 seems hardly worth consideration until well after the year 2000, for reasons of cost. International efforts are already under way to keep the electromagnetic environment essentially pure as seen from the far-side lunar surface. The Moon itself is an excellent shield against terrestrial radiations.

Site 2 employs distance rather than an RFI shield, to protect itself from Earth's radiations. It is likely to be more costly in the long run chiefly because of servicing costs, reliability requirements, and multiple, long distance relay link demands. Also, some minor but necessary limits on maximum permissible EIRP on Earth would be required in the search frequency bands. The greater distance provides only about 50 dB improvement over Site 3.

Site 3 requires either a shield against Earth and Earth satellite radiations or a nearly sole allocation of the search frequency bands to the search effort. A shield would allow search of the entire free-space window and pose no allocation requirements. However, RFI shields could be costly.

Site 4 can tolerate almost no Earth or satellite transmissions in the search bands, unless an RFI shield is used. Even then, there would likely be restrictions on satellite transmissions in the search bands. A careful engineering assessment would be required of any concrete proposal involving this site.

The RFI shields mentioned above are worth brief description. They are needed because realizable antennas are imperfect in the sense that there is no direction of signal arrival to which they are totally unresponsive. The purpose of the shield then is to attenuate signals by reflection, by absorption, and by diffractive loss, such that the signal energy reaching the antenna does not produce a detectable signal at the antenna output terminals.

Precise definition of shield requirements in any particular instance depends on the nature of the interfering fields (direction and strength) and on the directional characteristics of the antenna. The latter are variable, of course, since the antenna is required to look in many directions relative to the direction of the source of interference. Typically, in situations studied so far, shield attenuations on the order of 50 to 200 dB (i.e., power ratios of 105 to 102°) are required. To visualize the realization of such a shield, imagine a 300-m space antenna in Earth orbit. Adjacent to it and between it and the Earth is a large disk consisting of a ring supporting a thin conducting

membrane. The disk diameter would be of the order of 450 to 600 m. The conducting membrane may be only several tens of microns thick; even be a fine conducting mesh. The whole is equipped with means for orbital station keeping, engines, sensors, telemetry, etc., so that it rotates about the Earth in step with but unattached to the antenna. A smaller or larger antenna would require a proportionately smaller or larger shield.

Because such shields have not yet been developed, the need for thorough study is obvious. It is also possible that the shield may somehow be incorporated into the structure of the antenna itself.

Near-Earth orbit sites are omitted from the list because (as argued later on) the data processor must be on Earth, and because an antenna moving rapidly relative to the signal processor would seem to present very difficult telemetry problems, considering the SETI need for precise phase and frequency control, and the need for a very great degree of freedom from telemetry and atmospheric noise effects.

Control of miscellaneous interference from neighboring devices such as powerlines, switches, motors, etc., is well understood, but it is not a trivial matter and must not be ignored in the design stages of any SETI system.

In summary, the very real problem of RFI suggests a family of strategies for its alleviation, the nature of a particular strategy depending on a variety of wide-ranging factors such as possible sites, practical antenna characteristics, search frequency ranges desired, cost considerations, etc. There is no easily obtainable "quiet site" in view. On the other hand, the RFI protection problems for some combinations of search system parameters seem to be either minor in nature or at most only a moderate nuisance. This is true because searches are under way now on Earth in some clear bands and can continue for some time before the freedom to search is slowly closed down further by steadily increasing sources of RFI. But the time scale is not tomorrow; rather it is a decade or more in the future. This leaves adequate time for government departments, communication agencies, and so forth to make gradual adjustments toward cooperating with the search effort, thus avoiding exceptional costs and upset plans. Then too, advancing technology hurries obsolescence, so adjusting to the needs of a search system requires mainly thought and willingness, and comparatively small material cost or other inconveniences. It is most fortunate that the next and crucial World Administrative Radio Conference (WARC, see Section III-8) is scheduled for late 1979 and that so far, the water-hole band has not yet been occupied by extensive interferenceproducing installations such as are present in the bands on either side of the water hole.

Directional Search Modes

There are two distinct directional search modes: (1) the target search mode and (2) the area or "whole sky" search mode. Preference for one or the other, or for a mixture of the two, as major elements in an initial overall strategy for SETI, is a matter of judgment and practicality. Both modes have been and are being utilized. The point of view espoused here favors a modal mix at first, followed by increasing concentration on target search.

Target search assumes the existence of a known set of most likely transmitting sites. For example, in the Cyclops study it was proposed to observe presumed planets around Sun-like F, G, and K dwarf stars, in ascending order of distance from the Sun. There are about 150 of these within 50 light years, but the number increases to about 1.7X106 within 1000 light years. Some have argued strenuously that M0-M5 dwarf stars should be included in the target list. This would increase the size of the list by a factor of 2 to perhaps 5. Others have suggested giving first priority to a narrower selection of stars, say, only F5-K5 dwarf stars. Other discrete objects have also been proposed for special study.

The only civilization we know is on a planet orbiting a G2 dwarf star and it is appealing to expect to find other and somewhat similar civilizations on planets revolving around roughly comparable stars. There has been little change in the relevant astrophysical data since the Cyclops study in 1971. As a result, this strategy is widely supported.

Clearly, to minimize effort, target searches should be guided by the latest relevant knowledge in such areas as stellar and planetary formation and evolution, circumstance and origin of life, and cultural evolution. Because of the early stage of our knowledge in some of these fields, an overall search strategy should include a balanced substrategy for increasing relevant background knowledge; such a strategy would likely shorten the time for detection.

At present, astronomical star catalogs can identify perhaps only 0.1 percent of the stars within 1000 light years. A basic catalog listing all stars to 14th or 15th apparent magnitude just has not been constructed. Section III-4 outlines a strategy for generating an adequate Whole Sky Catalog within perhaps a decade and at moderate cost. Such a catalog would be invaluable to general astronomy as well.

In the meantime, the search can proceed by examining the truly local stars which are fairly well cataloged. Our nearest neighbors would seem to deserve rather intense study. Being near, intrinsically weaker radiations are detectable over the entire accessible spectrum. Being few in number, one can afford longer integration times. In any target search scheme, each time sensitivity is noticeably increased, one should reexamine previously observed objects.

The operation of a selected target search is clearly not a simple matter. When the committee or committees, which doubtless will control such matters, sets the following year's list of targets and areas to be searched, a fraction of the observation time should be set aside for trying inspired suggestions, as formal acknowledgment that the establishment is also working in the dark unknown. It is not beyond sensible conception that the first detection will be serendipitous.

Area search specifies nothing about the location of possible transmitting sites in the Universe. It merely characterizes the whole sky as a function of flux level, direction, and frequency.

In Section III-3 an elegant theorem is developed to this effect:

The received (signal) pulse energy is independent of the antenna area and is the energy that would be received by an isotropic antenna in the full sky search time.

This holds as long as all the received data are properly used in the data reduction process. It applies whether the sky is continuously scanned in any essentially nonoverlapping fashion or whether the antenna is used in isolated target search procedures, where a fictitious search time is easily calculated.

From this theorem one can derive an expression for the flux level attainable using an ideal receiver in the real universe and compare its performance with that of a state-of-the-art receiver. This expression is:

where the flux density, (6) varies directly with the system noise power spectral density (= kT), and with the signal-to-noise ratio, (m), required to keep the false alarm rate due to noise peaks at an acceptable level, and inversely with the wavelength, (X), squared and the full sky search time, (ts).

In perhaps a year's time, given sufficient wideband data processing equipment (see below), it should be possible to conduct a comprehensive search in the radio astronomy bands over the whole sky and with a sensitivity to narrow band signals many orders of magnitude greater than that used in existing radio astronomy observations. There would then be no need for retrospective studies of existing observations and surveys on the chance that radio astronomers have already observed an artificial signal from some fixed direction in space. Again, if we can search a 300 MHz frequency band at one time, it would take only a few years to search the whole sky over the entire microwave window. This would improve the state of our knowledge by many, many orders of magnitude.

Such whole sky area searches are quick and easy to perform with modest (~25 m) antennas, given the data processing equipment. They might well detect an ETI signal. Furthermore, an attractive dividend of a search throughout the microwave window would be the characterization of the whole sky at these frequencies to a systematic, known set of flux levels, spatial and frequency resolutions. The resulting astronomical data would be valuable, and it does not seem unduly optimistic to expect new discoveries in the spectral line domain, independent of precise frequency prediction.

Some Technological Aspects

The flux received at the Earth from a transmitter r light years away, per watt of equivalent isotropic radiated power (EIRP) is

at S/N = 1, when the system equivalent temperature lies in the range 2.7 K <T, < 10 K, and the resolution bandwidth (B,) is equal to or greater than the received signal bandwidth as observed over a unit time 7 ≈ (B,) ́1.

For purposes of this discussion, assume T, = 10 K and that we are searching for moderately stable carrier signals, so B, = 0.1 Hz. Then,

The disparity between equations (3) and (5) can only be overcome by some combination of real transmitted power (P) and transmission directivity (g) (EIRP = Ptgt) at the source, and effective antenna collecting area (A) at the receiving end.

These relationships are illustrated in figure 2 where range in light years (ly) is plotted against effective collecting area expressed in the number of 100-m radio telescopes required. Expressed another way, the range is approximately

r = 20[n100(P8/109)] 1/2 ly

(6)

where n1oo is the number of 100-m dishes required if they are 80 percent efficient (n = 0.8). The horizontal bands indicate the ranges likely needed for conditional detection probabilities in the range 0.63 Pc <0.95 and under four assumptions about the density (M) of transmitting civilizations in the Galaxy.

Antenna area and resolution bandwidth are interchangeable. The collecting area required dominates system costs to such a degree if high flux level signals are absent, that detection