| Optical SETI Map | Conferences Map | Illustrations Map | Photo Galleries Map | Observations Map | Constructing Map |
|
EJASA - Part 5
Table 2, Line 26 -
This is the bottom line, showing the SNR (CNR) normalized to a
1 Hz bandwidth. The 34 dB CNR for the 656 nm system corresponds to a
photon detection rate of 2,640 per second (Equ. 36). For practical
Professional Optical SETI searches, we should be looking for signals
with minimum bandwidths of about 100 kHz. As long as the Signal-To-
Planck and Signal-To-Daylight ratios are larger than the quantum SNR,
the former do not reduce the system performance. It should be noted
that at a frequency of 1.5 GHz (wavelength = 20 cm), the full
6.4-kilometer diameter microwave Cyclops Project [5], which in 1971
would have cost about ten billion dollars, only achieves an SNR of
60 dB (see Table 1, Page 19). This is about 26 dB greater than for a
10-meter diameter symmetrical visible system.
Other than the fact that interstellar absorption at microwave
frequencies for distances in excess of a few thousand light years is
significantly less than in the visible spectrum, the Microwave Cyclops
system has little to commend it for communications within the solar
Page 27
neighborhood, particularly as the cost of the receiver is about one
hundred times that of a single-aperture ground-based optical counter-
part. This is good grounds for thinking "small is beautiful". For
some strange reason, while free-space laser communications appears to
be fine for future terrene GEO (Geosynchronous Earth Orbit) to LEO
(Low Earth Orbit) and deep-space communications (much of this work is
being coordinated by NASA [63-66]), the SETI community appears to be
convinced that ETIs would not use such technology for interstellar
communications! This is illogical. A presently favored operating
wavelength for terrene free-space communications systems is 530 nm
(green), obtained by frequency-doubling the 1,060 nm wavelength
produced by a laser-diode pumped Nd:YAG laser.
As previously mentioned, terrene SETI programs appear to have been
distorted by poor assumptions in the Cyclops study (see Table 1,
Page 19). [5] As we showed earlier, the efficacy of the optical
approach was severely hampered by constraining the near-infrared
transmitting telescope size to 22.5 cm. It boggles the mind to think
that ETIs would be trying to contact us with their equivalent of a
Celestron or Meade telescope. This would put the onus on us to build
very large and expensive multi-aperture receiving telescopes to pick up
their weak signals; surely the very opposite would be the case! The
Cyclops study was unable even to predict the rise in ascendancy of the
ubiquitous semiconductor chip over the following five years, and the
effect it would have on SETI signal processing, even though integrated
circuits were being developed in the editors' backyard!
Present-day experimental ground-based free-space communications
links are already using receiving telescope apertures as large as
1.5 meters. [66] Since the overall performance of symmetrical systems
is proportional to the telescope diameter raised to the sixth to eighth
power (allowing for power density limitations due to heating effects
at the transmitter mirror), poor estimations about transmitting and
receiving telescope apertures can drastically skew a comparative
systems analysis. In practice, transmitting and receiving telescopes
are likely to be extremely asymmetric. If we do discover an optical
ETI signal in the next few decades, it will probably be found to have
been transmitted by a huge optical array, while our receiving antenna
will be a relatively puny telescope.
Figure 4 shows a graph of received signal spectral density,
superimposed on the Planckian spectral density curve for a (solar-type)
black body radiator at a temperature of 5,778 K. The microwave system
performance shown in this graph is based on the 300-meter diameter
Arecibo telescope; producing a CNR some 19 dB greater than for the
100-meter radio telescope system modelled in Table 2 (Page 22).
The reader is encouraged to compare this graph to that given in
FIRST CONTACT [26] (Chapter 4, Page 151, by Dr. Michael Klein). The
first impressions from that graph (Figure 1 of Chapter 4) is again
that optical communications are useless. This is far from the truth.
Indeed, the graph is very misleading. One might be forgiven for
thinking that in this model the ETIs are using Compact Disc-type
laser-diodes and/or hobby model-type telescopes! The assumed
Page 28
Spectral Density, W/m^2.Hz
|
10^-15 |
| EIRP = 2.3 X 10^18 W 23rd Mag.
| *
| EIRP = 8.7 X 10^15 W * CNR = 34 dB
| * * .
10^-20 | CNR = 22 dB * *. Quantum
| *. * Noise
| . * *
| * #656 nm Beacon
| EIRP = 2.2 X 10^10 W * # | #(10 m Dia.)
10^-25 | * 10,600 nm Beacon # | |
| * CNR = 20 dB (10 m Dia.) # |V| #
| . .*. . . . . . . # |I| 2nd Mag.
| * Temp. = 10 K # ^ |S| #Starlight
| * # | |I|
10^-30 | * # Planckian |B| #
| * # Black Body |L|
| # 1.5 GHz Beacon Curve |E| #
| # (300 m Dia.) | |
|# |L| #
10^-35 | |I|
| |G| #
| |H|
| Microwave Millimeter Infrared |T| Ultra #
| | | Violet
10^-40 ----------------------------------------------------------------
10^8 10^9 10^10 10^11 10^12 10^13 10^14 10^15 10^16
Frequency, Hz
Figure 4 -
Spectral density and interstellar CNR for 1 kW (SETI) signals at
ten light years. Quantum Efficiency at Visible and Infrared = 0.5.
Microwave system is based on 300-meter diameter Arecibo-type
telescopes. Optical systems are based on perfect 10-meter diameter
telescopes as modelled in Table 2. The Carrier-To-Noise Ratios (CNRs)
are normalized to a 1 Hz bandwidth. The EIRP of a solar-type star =
3.9 X 10^26 W, and has an apparent magnitude equal to 2.2.
optical EIRPs are much too low. Also, the graph is plotted in terms
of EIRP, and therefore exaggerates the efficacy of the microwave
approach for an electronic receiver (instead of an observer), because
it does not show the typical 10 K noise floor of a high-quality
microwave receiver, only the radio brightness of a quiet G-type star.
The latter is about 54 dB beneath the 10 K systems noise floor, as
shown in Figure 4, and could only be detected after considerable signal
integration. At 1.5 GHz, it is generally the Cosmic Background, i.e.,
the 2.73 K aftermath of the theoretical Big Bang, and the electronic
noise in the microwave front-end that limits signal detectability, not
Planckian radio noise from the star.
Page 29
LASERS
Table 3 gives a list of many of the more important laser types
presently known. [79] As previously mentioned, the CO2 wavelength of
10,600 nm has been identified as an "optical magic wavelength".
[46-47,51-53,57] However, there are many laser wavelengths in the
visible and infrared spectrums that might be suitable for ETI trans-
mitters and local-oscillators. We should not discount the possibility
that ETIs may use efficient frequency-doubled lasers, so we might
consider exploring the visible spectrum for near-infrared lasers at
half the wavelengths quoted below. For example, the 532 nm wavelength
corresponding to the frequency-doubled Nd:YAG 1,064 nm transition may
be a suitable wavelength; one that is presently favored for terrene
optical communications.
=====================================================================
| Table 3 Important laser types and wavelengths |
|=====================================================================|
| Type | Wavelength (nm) |
|----------------------------------|----------------------------------|
| Free-Electron | Ultra-violet to far-infrared* |
| Krypton-Fluoride Excimer | 249 |
| Xenon-Chloride Excimer | 308 |
| Nitrogen Gas (N2) | 337 |
| Organic Dye (in solution) | 300-1,000 (tunable)** |
| Krypton Ion | 335-800 |
| Helium-Cadmium | 422.0 |
| Argon Ion | 450-530 (main lines 488 & 514.5) |
| Helium Neon | 543, 632.8, 1,150 |
| Semiconductor (GaInP) | 670-680 |
| Ruby | 694 |
| Semiconductor (GaAlAs) | 750-900 |
| Neodymium YAG | 1,064 |
| Semiconductor (InGaAsP) | 1,300-1,600 |
| Hydrogen-Fluoride Chemical | 2,600-3,000 |
| Semiconductor (Pb-salt) | 3,300-27,000 (tunable)** |
| Deuterium Fluoride | 3,600-4,000 |
| Carbon Monoxide | 5,000-6,500 |
| Carbon Dioxide (CO2) | 9,000-11,400 (main line 10,600) |
=====================================================================
* Extremely high peak powers available within the decade (> 100 GW).
** Suitable for wide-tunability receiver local-oscillators.
Carbon Dioxide and Semiconductor lasers are very efficient. In
addition to the types listed above, there are a variety of chemical
lasers, including: Iodine, Hydrogen Bromide, Xenon Hexafluoride,
Uranium Hexafluoride, and Sulphur Hexafluoride. These chemical lasers
are efficient and very powerful.
Lasers like the Helium-Cadmium and Helium-Neon can be discounted
because of their very poor efficiency and low power, even though their
temporal coherence is excellent. Similarly, the original Ruby laser is
Page 30
=====================================================================
| Table 4 The most intense Fraunhofer lines from the Sun{1} |
|=====================================================================|
| Wavelength, nm Bandwidth, nm Bandwidth, GHz Element |
|---------------------------------------------------------------------|
| 410.1748 0.3133 558.7 H_delta |
| 413.2067 0.0400 71.0 Fe I{2} |
| 414.3878 0.0466 81.4 Fe I |
| 416.7277 0.0200 34.5 Mg I |
| 420.2040 0.0326 55.4 Fe I |
| 422.6740 0.1476 247.9 Ca I |
| 423.5949 0.0385 64.4 Fe I{2} |
| 425.0130 0.0342 56.8 Fe I{2} |
| 425.0797 0.0400 66.4 Fe I{2} |
| 425.4346 0.0393 65.1 Cr I{2} |
| 426.0486 0.0595 98.3 Fe I |
| 427.1774 0.0756 124.3 Fe I |
| 432.5775 0.0793 127.1 Fe I{2} |
| 434.0475 0.2855 454.6 H_gamma |
| 438.3557 0.1008 157.4 Fe I |
| 440.4761 0.0898 138.9 Fe I |
| 441.5135 0.0417 64.2 Fe I{2} |
| 452.8627 0.0275 40.2 Fe I{2} |
| 455.4036 0.0159 23.0 Ba II |
| 470.3003 0.0326 44.2 Mg I |
| 486.1342 0.3680 467.2 H_beta |
| 489.1502 0.0312 39.1 Fe I |
| 492.0514 0.0471 58.4 Fe I{2} |
| 495.7613 0.0696 85.0 Fe I{2} |
| 516.7327 0.0935 105.1 Mg I{2} |
| 517.2698 0.1259 141.2 Mg I |
| 518.3619 0.1584 176.9 Mg I |
| 525.0216 0.0062 6.7 Fe I{3} |
| 526.9550 0.0478 51.6 Fe I{2} |
| 532.8051 0.0375 39.6 Fe I |
| 552.8418 0.0293 28.8 Mg I |
| 588.9973 0.0752 65.0 Na I(D2){2} |
| 589.5940 0.0564 48.7 Na I(D1) |
| 610.2727 0.0135 10.9 Ca I |
| 612.2226 0.0222 17.8 Ca I |
| 616.2180 0.0222 17.5 Ca I |
| 630.2499 0.0083 6.3 Fe I{3} |
| 656.2808 _____________ 0.4020 ________ 280.0 ________ H_alpha |
| 849.8062 0.1470 61.1 Ca II |
| 854.2144 0.3670 150.9 Ca II |
| 866.2170 0.2600 104.0 Ca II |
=====================================================================
Table reproduced from "Astrophysical Formulae", edited by K.R. Lang,
Springer-Verlag, 1978, p. 175. [90]
{1} After MOORE, MINNAERT, and HOUTGAST.
{2} Blended line.
{3} Magnetic sensitive line.
Page 31
inefficient and low power. Probably, one of the more important
considerations for an ETI transmitting laser is that it should be
capable of being deployed in space, be able to produce extremely high
C.W. or pulse powers, and be nuclear or stellar (solar) pumped.
Organic dye lasers are suitable for local-oscillators, with their
wide tunability and narrow linewidth (< 5 kHz). Lead-salt semi-
conductor lasers are suitable for infrared local-oscillators.
FRAUNHOFER LINES
Table 4 is a list of the most intense Fraunhofer lines from the Sun
and their effective bandwidths. The H_alpha Hydrogen line upon which
the visible Optical SETI model is based, has a wavelength of
656.2808 nm (frequency = 4.57 X 10^14 Hz), and an effective linewidth
or bandwidth of 0.402 nm (280 GHz). [88-90] The actual FWHM linewidth
is somewhat less that 280 GHz.
THE OPTICAL SEARCH
An "All Sky Survey" of the type planned for the Microwave Observing
Project (MOP), which pixelizes the entire celestial sphere, does not
make sense in the optical regime. [40-45] The 10^16 beams (Equ. 20)
for a diffraction limited 10-meter diameter visible-wavelength
telescope are mainly wasted looking out into empty (local) space. For
a celestial sphere one thousand light years in radius, containing one
million solar-type stars, the average angular separation between stars
is 0.23 degrees (see Figure 10). A 34-meter diameter radio telescope at
1.5 GHz has a typical field-of-view (FOV) of 0.41 X 0.41 degrees, and
thus, on average, its FOV encompasses several stars. It is efficient
when conducting a radio "All Sky Survey" to continuously scan the
celestial sphere in consecutive or adjacent strips or sectors.
The 10-meter diameter Professional 656 nm Optical SETI Telescope
would have a typical FOV = 0.33 X 0.33 degrees and a 128 X 128
photodetector array FOV = 2.1" X 2.1". Since the average separation
between stars is 0.23 degrees, the average number of stars in the
optical array FOV is 6.4 X 10^-6. Thus, the narrow diffraction-limited
field-of-view means that for most of the time the optical detector(s)
would be viewing empty space. A similar situation prevails for the
smaller, single detector amateur optical telescopes to be discussed
later. The argument has been advanced by Dr. Bernard Oliver, in
correspondence with the author and at the author's SETI Institute talk,
that because an "All Sky Survey" would be out of the question at
optical frequencies, this implies that ETIs would not use these
frequencies.
The author's response to this is that there is nothing "holy" about
the "All Sky Survey" approach. What we may wish to do is to have a
Targeted Search of tens of thousands of stars, instead of a mere eight
hundred as presently planned for MOP (see Page 11). However, each time
we wish to scan another star in the frequency domain, we will move the
Page 32
telescope to an adjacent sector of the sky that contains the desired
object.
While there is the possibility that ETI transmitters exist in the
interstellar voids, far from their home stars, the author thinks that
this scenario is unlikely (except perhaps within our own solar system,
i.e., von Neumann-type probes), if for no other reason than it would
place the energy-intensive transmitters far from a "cheap" and
plentiful energy source.
One of the many objections made to the optical approach to SETI is
that there are just too many frequencies to search. As Figure 5
illustrates, under the author's rationale, this is more a perception
than a reality because of the wider signal bandwidths assumed.
21-cm Water-Hole Channel or Bin
| |
------------------------------------------------------------------
| * # |
| * MICROWAVE HAYSTACK # |
| * # |
------------------------------------------------------------------
| | --> <-- |
1 GHz 10 GHz 1 Hz 100 GHz
Number of 1 Hz frequency channels or bins between
1 GHz and 10 GHz = 9 Billion.
10,600 nm 656 nm
| |
------------------------------------------------------------------
| * # * |
| * OPTICAL HAYSTACK # * |
| * # * |
------------------------------------------------------------------
| | --> <-- |
10 THz 100 THz 100 kHz 1,000 THz
Number of 100 kHz frequency channels or bins between
20 THz and 920 THz = 9 Billion.
Figure 5 -
The Microwave and Optical Cosmic Haystack frequency domains. This
demonstrates that the number of frequencies to search in the microwave
and optical haystacks are of similar magnitude.
Wide bandwidth means that laser linewidths, Doppler shifts, and
chirps (drifts) are less significant, and the number of frequencies to
search in the optical spectrum is more manageable. Just because
visible frequencies are over five orders of magnitude higher than
Page 33
microwave frequencies does not mean that there are over 10^5 more
frequencies to search in the optical frequency domain. The modulation
bandwidth of proposed optical ETI signals as a percentage of the
carrier frequency may be as large or larger than the percentage
modulation bandwidth of proposed microwave ETI signals. In fact,
assuming minimum bin bandwidths of 100 kHz, the number of frequencies
to search in the entire optical spectrum may not be much greater than
the number of 1 Hz frequencies between 1 and 10 GHz, i.e., nine
billion! This is illustrated diagrammatically in Figure 5. This
clearly has important ramifications in terms of the search time.
The reader should note that for a drifting carrier signal, i.e.,
one subjected to Doppler Chirp, the optimum detection bandwidth is
equal to the square root of the frequency drift rate. [5,8] This
assumes that the local-oscillator laser is not de-chirped. Thus, the
optimum bandwidth for a monochromatic 1.5 GHz signal drifting at a
local Doppler Chirp rate of 0.17 Hz/s (see Table 2, Line 30, Page 22)
is about 0.4 Hz, while for a monochromatic 656 nm signal drifting at
51 kHz/s, the optimum bandwidth is 226 Hz. If the bin bandwidth is
excessive, too much system noise is detected, and the CNR is degraded.
On the other hand, if the bin bandwidth is too small, the response time
of the filter (approximately 1/Bif) is insufficient to respond to all
the energy in the signal as it sweeps by, again leading to a reduction
in CNR and detectability.
It is an interesting exercise to estimate the time that would be
required at visible wavelengths for both an All Sky Survey and a
Targeted Search. We will assume the use of a 10-meter diameter
receiving telescope, a 128 X 128 photodetector array (16,384 pixels),
and initially, a single 10 GHz bandwidth Multi-Channel Spectrum
Analyzer (MCSA) that sequentially samples all 16,384 photodetectors.
These MCSAs could have final bin bandwidths of about 100 kHz. At this
time, 10 GHz MCSAs do not exist, and the state-of-the-art for single-
chip devices employed in Microwave SETI is about 10 MHz. However, it
is only a question of time before these more powerful 10 GHz devices
are developed.
For the purposes of this brief analysis we shall not concern
ourselves with the huge amount of data storage that must be provided,
or the data reduction time overhead required. Equ. 20 (Page 81) shows
that the number of received beams for such a telescope is about 10^16.
Since the minimum sampling time per pixel for a 10 GHz bandwidth is
100 ps, the time to sample the entire array of 16,384 instantaneous
beams is 1.64 microseconds. The number of array sets of beams in the
celestial sphere consisting of 10^16 beams is 6.1 X 10^11. Thus, the
time just to "look" at one 10 GHz wide band of the visible spectrum,
assuming that a continuous scan of the sky could be made with no dead
time or overlap, is 10^6 s, i.e., 11.6 days! This is a substantial
amount of time for a single band just 10 GHz wide.
Since there are 42,857 bands of 10 GHz bandwidth between 350 nm and
700 nm, the time required to search the entire sky and all visible
frequencies, is at a minimum, 1,360 years! Even if we had 128
parallel MCSAs (don't even consider having 16,384 - 10 GHz MCSAs!), the
Page 34
time to search even a 10 GHz band is long, notwithstanding the "slight"
data storage problem. Clearly, we can forget about this form of
optical All Sky Survey, since it is a grossly inefficient way of
scanning or pixelizing the sky. Almost all the data bins will be empty
bins, having been derived from beams pointing to empty (near) space.
The situation for an Optical All Sky Survey is actually much worse than
just implied, due to the additional time that each pixel must be
sampled to ensure a high probability of detecting the fewer, but more
energetic optical photons - more about this in a moment.
On the other hand, if we only consider a Targeted Search, the time
required is much shorter and allows for the search to be done across
the entire optical spectrum, not just at selected laser frequencies or
Fraunhofer lines. As we have just seen, if the photon arrival rate is
sufficiently high, the time with a single 10 GHz MCSA for a single scan
of the entire array is 1.64 microseconds. To scan for one star over
the entire 350 nm to 700 nm band would take 0.070 seconds (assuming
suitable L.O. lasers). This is a trivial amount of time, and the amount
of data that has to be collected and stored is relatively insignificant.
Indeed, it is the time to do the FFTs and move the telescope to a new
position that will be the most significant overheads here.
The above times are highly optimistic because the basic flux
sensitivity of any kind of receiver, be it microwave or optical, depends
on the sampling or integration time. Hence, before we can estimate the
realistic length of time for a given search, we must decide what are the
minimum detectable flux levels that we wish to detect. This, in turn,
will determine the minimum sampling time for each pixel. Usually, SETI
minimum detectable flux estimates are based on integrating a very weak
signal for a period of time, and not for providing sufficient SNR to
allow actual demodulation. We must also decide if we want to model a
system based on short pulses or on continuous wave (C.W.) signals.
Of course, it is extremely unlikely that the signal flux would be
sufficiently high to allow for a high probability of detecting the
photons in a sampling bandwidth of 10 GHz. In reality, our minimum
MCSA bin bandwidths would be about 100 kHz, and the sampling
(integration) time is at least a factor of 10^5 longer. For the
purposes of this further analysis, we shall assume a C.W. signal and a
100 kHz minimum bin bandwidth, so that the pixel sampling time is now
10 us. For our 10-meter diameter 656 nm symmetrical heterodyning
telescope system, we can estimate the minimum detectable signal flux
density by calculating the flux required to reduce the CNR to 0 dB.
We have already shown (Table 2, Line 12, Page 22), that a flux
intensity of 2.04 X 10-^17 W/m^2 will produce a CNR = 34 dB re 1 Hz.
Therefore, in a 100 kHz bandwidth, the CNR will be -16 dB. To increase
the CNR to 0 dB means that the intensity must be increased by 16 dB to
8.12 X 10^-16 W/m^2. Thus, the minimum detectable signal flux for
this bandwidth and sampling rate is 8.1 X 10^-16 W/m^2. This is
equivalent to saying that during the 10 microsecond sampling time, if
an ETI signal is present on one pixel, we would have a reasonable
probability of detecting one photon (Equ. 36). This signal flux would
be produced by a ten meter diameter transmitter at a range of ten light
Page 35
years, with a power of 16 dB re 1 kW, i.e., 40 kW. This is a trivial
amount of power for an ETI.
Continued
| ||||||