Let’s imagine for a moment that John Mathews (Pennsylvania State University) is right in theorizing that space-faring civilizations will use self-reproducing probes to expand into the galaxy. We’ve been kicking the issues around most of this week, but the SETI question continues to hang in the background. For if there really are extraterrestrial civilizations in the nearby galaxy, how would we track down their signals if they used the kind of communications network Mathews envisions, one in which individual probes talked to each other through tight-beam laser communications designed only for reception by the network itself?
One problem is that the evidence we’re looking for would most likely come in the form of spread-spectrum signals, a fact Jim Benford pointed out in a comment to my original post on Mathews, and one that also pointed to recent work by David Messerschmitt (UC-Berkeley). The latter makes a compelling case for spread-spectrum methods as the basis for interstellar communication because such signals are more robust in handling radio-frequency interference (RFI) of technological origin. In SETI terms, RFI is a major issue because it mimics the interstellar signal we are hoping to find, and Messerschmitt assumes an advanced civilization, having experienced RFI issues in its own past, will use the best tools to minimize them.
Image: 3D map of all known stellar systems in the solar neighbourhood within a radius of 12.5 light-years. Can we build self-reproducing probes that could explore these systems over the course of millennia? If other civilizations did the same, could we detect them? Credit: ESO/R.-D.Scholz et al. (AIP).
Spread-spectrum techniques spread what would have been narrowband information signals over a wide band of frequencies. Think of the kind of ‘frequency hopping’ deployed in World War II, where a transmitter would work at multiple frequencies and the receiver would need to tune in to each of the transmitted frequencies. In addition to being resistant to interference, the method allows you to resist enemy jamming of your communications or to conceal communications in what would otherwise seem to be white noise. Actress Hedy Lamarr and composer George Antheil developed a frequency hopping technique that made radio-guided munitions much harder for enemy forces to jam, a spread-spectrum story entertainingly told in a recent book (Richard Rhodes’ Hedy’s Folly: The Life and Breakthrough Inventions of Hedy Lamarr, the Most Beautiful Woman in the World, 2011). Lamarr and Antheil’s system used 88 different carrier frequencies.
Messerschmitt isn’t talking about probes but about one civilization trying to reach another — he works from the perspective of the transmitter designer trying to reach a receiver about which little can be known. From the paper:
The transmitter can…explicitly design a transmit signal that minimizes the effect of RFI on the receiver’s discovery and detection probabilities in a robust way; that is, in a way that provides a constant immunity regardless of the nature of the RFI. It is shown that the resulting immunity increases with the product of time duration and bandwidth, and that the signal should resemble statistically a burst of white noise. Intuitively this is advantageous because RFI resembles such a signal with a likelihood that decreases exponentially with time-bandwidth product. Both a transmitter and receiver designer using this optimization criterion and employing the tools of elementary probability theory will arrive at this same conclusion. Although the context is different, variations on this principle inform the design of many modern widely deployed terrestrial digital wireless communication systems, so this has been extensively tested in practice and is likely to have a prominent place in the technology portfolio of an extraterrestrial civilization as well.
We’ve learned a great deal about dealing with RFI, especially given the rapid growth of wireless communications here on Earth, and we’ve also learned more about how radio signals propagate in the interstellar medium, thanks in large part, Messerschmitt notes, to advances in pulsar astronomy. Couple this with the ever-quickening pace of electronics and computer development and the search technologies in play are expanded so that we can accommodate the problem of natural noise sources as well as our own RFI. And we would have to assume that any extraterrestrial civilization would employ RFI mitigation techniques in its own communications.
In the case of accidental interception of a signal beamed between two intelligent probes, we can also look at the issue in terms of our detection algorithms. Early SETI work involved the so-called Fourier Transform to search for comparatively narrowband signals, and moved in the 1960s to the Fast Fourier Transform as the tool of choice. But as François Biraud noted as early as 1982, our terrestrial move from narrow-band to broader bandwidth communications presents a new challenge, breaking information into chunks carried by frequency-shifting carrier waves. Claudio Maccone has long argued that FFT methods are inappropriate for this kind of signal.
Enter the Karhunen-Loève Theorem (KLT) that Maccone continues to champion, a way of improving our sensitivity to an artificial signal that can dig tricky spread spectrum signals out of the background noise. Whether or not KLT algorithms are put to work with new installations like the Square Kilometer Array remains to be seen, but arguments like Messerschmitt’s point to the viability of spread-spectrum methods as a prime choice for interstellar communications. The point, then, is that spread-spectrum modulation is a factor we can deal with, allowing us to incorporate Messerschmitt’s ideas into our SETI toolkit even as we ponder the circumstances that might lead an extraterrestrial civilization to deploy a network of self-reproducing probes.
The Messerschmitt paper is “Interstellar Communication: The Case for Spread Spectrum” (preprint), while the Mathews paper is “From Here to ET,” Journal of the British Interplanetary Society 64 (2011), pp. 234-241. I have more to say about all this, and particularly about the ethical issues raised by self-reproducing technologies, but I’m running out of time this morning. The discussion continues tomorrow, when I’ll ponder how a civilization like ours might accidentally run into a network of extraterrestrial probes, and what that encounter might look like.
Assuming that probes or civilizations are tight beaming each other in a network, what are the chances that we would be in the path of one of those transmissions and therefore able to intercept it?
Over time, each iteration of SETI following a detection failure, follows our technology and shapes ideas about ET’s elusiveness. We’ve had high strength omnidirectional beacons at the “obvious waterhole”, omnidirectional beacons across any frequency, encoded beacons, tight beams, transient tight beams adjusted to our orbital period (Benfords?), tight beams between probes and now spread spectrum tight beams. Encoded spread spectrum tight beams might be next. Should quantum entanglement or some other technique work for information transmission, that will become the preferred medium for “whispering Gods”.
And why the assumption that probes will be something like our space probes – clunky, easily recognizable machines? What if they mimic humans (or animals) and collect the data in plain sight? What if they are really clever and control/direct corporate entities to do their work? What if the scientific enterprise is a cleverly implanted gene/meme set that directs us to do all the work of analyzing the local environment for the probe? There is no end to the speculative possibilities/paranoia.
While I think it is worth spending some resources on SETI ($20 bills can very occasionally be found in the gutter), more direct observations of planets by telescopic techniques and eventually our own interstellar probes are going to be the best hope of getting answers to what is “out there”.
Paul, your range of interest and understanding is amazing!
In this article there are two topics, one is about two civilizations (A and B) that want to communicate and the other is a third civilization (C) that wants to detect communication.
If we don’t consider the obvious problem of signal’s transit time, it remains the two civilizations (A and B) are interested to use the best as possible the spectrum used, so there will be as few as possible redundancies in the symbols sent on one radio frequency, compression is an example of technology that reduce the symbol redundancies. A signal where there is few redundancies looks like noise, and it’s a problem for the civilization (C).
Spread spectrum is a form of noise management because it’s very unlikely there will be natural noise at each couple (time, frequency) where a signal is sent. So the signal/noise looks better it would be when using only one carrier. This further complicates the task for civilization (C) but as you said (and this is well above my head) there may exist better mathematical tools.
However I think there is a difference in one hand between recognizing there is a signal and in the other hand being able to access the information this signal bears. To transmit information one has to send a signal above the noise level, there is no way to escape to that. The information sent or the signal itself can be so scrambled it may have a pattern that looks like noise but that doesn’t mean IMO that someone listening wouldn’t recognize someone is sending information.
Spread spectrum increases the perceived signal/noise ratio but IMO the instantaneous signal must still be more powerful than the noise on one carrier. Wither it will be easily detected or not depends on the filter curve (narrow or broad) at civilization (C), wide filters will not detect spread spectrum transmission but anyway that kind of filter is not very useful even if only one carrier is used as it will catch all the noise over a large spectrum and signal sent on only one carrier must compete with all that noise. And as you said spying an infinite range of narrow band carriers is impossible.
So IMO there is not much to say about SETI search because it would work only if the other civilization uses 20th century telecommunication technologies with 23th century power levels, which is quite unlikely.
“In SETI terms, RFI is a major issue because it mimics the interstellar signal we are hoping to find, and Messerschmitt assumes an advanced civilization, having experienced RFI issues in its own past, will use the best tools to minimize them.”
In this really true? And what if the RFI is itself spread spectrum? More to the point, I really think that are far greater issues to SETI than RFI.
I’ve only read the first few pages of the paper (if time permits I’ll keep reading). What I’ve seen so far doesn’t convince me that spread spectrum is such a great idea for SETI, or even for interstellar point to point communications. In the latter case you still have the difficult problem of locking the receiver to the transmitter’s pattern since, unlike terrestrial spread spectrum such as CDMA, there is no common clock and there is scintillation to deal with.
Radio SETI has yet to succeed. There are 3 main reasons that come to mind:
1. There is no signal.
2. We’re looking for the wrong signal characteristics.
3. The signals are too weak to be recovered.
Spread spectrum primarily falls under points 2 and 3. With regard to point 2, what this proposal does is to suggest we add to the library of signatures that are currently in the search programs or that we eliminate other signatures in favor of some subset of spread spectrum patterns (out of an almost infinite set).
Worse, searching for a spread spectrum signal with an unknown pattern is far more time consuming than searching for single-frequency carriers, with or without modulation. How will this aid the search? One could of course search for some sort of “trigger” signal on which to sync or to search around, but those would fall into the category of non-spread spectrum signals.
Coming to point 3, spread spectrum, if you don’t yet know the pattern, makes the weak signal problem even worse. For example, if an unmodulated carrier hops among 10 discrete frequencies, if you listen to any one of those frequencies the SNR is -10 db versus a solid carrier. Since, per the persistent lack of signals that are noticeably above the noise floor, this makes a terribly challenging search even worse since it requires far greater integration times, during which the signal will drift due to Doppler. This is not good.
Maybe Messerschmitt addresses these difficulties later in the paper, past where I’ve read up to.
Jean-Pierre: “To transmit information one has to send a signal above the noise level”
This is not strictly true. Noise (thermal) is not a solid wall but a statistical variation around an average value. Long integration times can raise the effective SNR. However, the information (modulation) rate that can be recovered falls as the integration time increases. Keeping the receiver locked on the signal during this process can be very difficult as well.
It takes patience when you can only communicate one bit per minute or hour.
Alex Tolley – To continue your thought(or paranoia):
Alien probes whose goal is to find a promising indigenous life form and coax them along from ape to human to spacefaring race and eventually to building new probes and continuing the expansion. Might we be a cog in the wheel of a cosmic empire? We have made amazing progress in a short period of time. Perhaps the aliens ARE among us.
Paul- I am really enjoying this whole theme (KBO’s, Oort Cloud, and slow migration outward) Please continue it.
The above discussion makes some asumptions about the ET civilisations and their use of radiofrequenzies which may not be clear.
Is the ONLY use they make of detectable elektromagnetic waves , of the kind mentioned above ? Do they have a single comandcenter capable of enforcing certain codes of radio-behavior over lightyears of distance ? No amateurs , disenters , rebels , far away settlements with limited or damaged equipment ?
I once read somwhere that some of the more powerful existing longwave radiostations on earth could be heard easily inside 50 LY . If thats true ,we should be hearing something , even if it would represent only a very smal fraction of their comunication .
Looking at this discussion over the past few weeks, I’m struck that we need more input from evolutionary biologists and theorists.
Assuming self-replicating probes, we should probably assume some learning and adaption in the probes, which will mean exploration of different ecological niches over long periods of time. The problem of making self-replicating probes that do not replicate uncontrollably (or evolve in other undesired ways) is akin to making AI’s that do what we want / have our values (see: http://en.wikipedia.org/wiki/Friendly_artificial_intelligence). It seems like something very hard once you allow something to evolve / learn / adapt independently, since some of those adaptions may go in directions you don’t want and didn’t anticipate.
So what are the equivalents of predators, plants, herbivores, parasites, commensals in interstellar probe populations? I suspect different adaptive strategies would lead to different kinds of communication behaviors (mostly cost-optimized, but then again, there maybe adaptive strategies that lead to costly-signaling, see:
http://en.wikipedia.org/wiki/Signalling_theory
). I suspect some “probes” will have no interest in communication, others may be chatty, some may lie, some may “eat” chatty neighbors, etc. etc.
Again, I think we should do more to consider ET’s as an evolutionary / ecology problem. How this work out in ecological terms, played out over very large distance and time-scales?
This is Dave Messerschmitt, author of the paper in question.
Just to make it clear, in order to detect a signal with reasonable sensitivity (that is not require the transmitter to use a lot of excess power above fundamental limits), requires that you know the signal you are looking for. Just searching for any spread spectrum signal without knowing the exact nature of that signal is hopeless. Thus, detectability depends on the receiver being able to guess the exact nature of the signal.
There is hope. The theory says that the signal should have the statistical properties of white Gaussian noise. As I argue in the paper, to be “guessable” that WGN should be generated by a pseudo-random *algorithm*, and all the receiver has to guess is the *algorithm*. There are some other obvious properties that limit the nature of the algorithm. I argue in the paper that there are relatively few “obvious” choices, and specifically I argue in favor of basing the algorithm on the simplest geometrical constructs — a circle or a square, either of which gives directly perfectly suitable choices.
As far as criticism of the nascent state of 20th century communications technology: While I appreciate the sentiment, I reject that notion. Not only is telecom fairly mature as a technology, but much of what we do (including spread spectrum) is based on optimization — by mathematical proof, the best that can be done, by any civilization at any stage of development. The argument in the paper in favor of spread spectrum is based on optimization using simple probability theory.
Of course optimization only works as a method of coordination if the transmitter and receiver agree on the optimization *criterion*. I argue that robust immunity to RFI is a good choice because the transmitter obviously has no specific info about the nature of RFI in the receiver’s vicinity.
Any other questions, I would be happy to respond.
Very close to zero, I would expect.
Let us assume that the stellar gravitational focus is as useful for communications as some of us think. Then, the galactic Internet will have an interstellar backbone made up of beams going from star to star, focused at each end onto the receiver or transmitter by gravitation. That means:
a) Signals will be relatively weak, because the focus is so good.
b) The beam, while very well collimated, will be broad (the size of the Einstein rings) and exactly in line with the two stars
These two, in combination, make it nearly impossible to intercept or even detect such a beam anywhere other than at the foci, where the receiver and transmitter are located. Even right between the stars the beam will be spread out over such a large diameter, that it will be negligible compared with background noise. To be in line with the beam, you have to look directly towards one or both of the stars, dramatically increasing the background noise.
In other words, before we even begin to worry about spread spectrum, we are already thwarted by the spatial spreading inherent in the gravitational focus technique.
The only thing we might hope to be able to pick up from other stars is the non-gravitational, local communication in the Oort cloud, which may produce stray beams that could conceivably be detected a thousand times the distance they were meant to reach, but it is quite a stretch.
For SETI, there is one interesting lead, though: If there was such a galactic network, our sun would most likely be part of it, and we could look for unusual sources of radiation 700-1000 AU out. These would have to have active station-keeping, which would be one way to identify them reliably as artificial.
I think robots are much better suited to space travel than any biological entities will ever be so it makes a lot of sense to assume that the further we travel from Earth, the more automated and mechanical the affair will be. I suppose “we” at this point would include probes or networks of probes that could years to communicate an findings with us or may never interact with humans. Even though robots do not seem to be alive according to our current understanding, if they were to be self replicating as is proposed, then what is the difference? In face it would seem that evolution (as mentioned above) would take play, perhaps resulting in a “species” far more capable and worthy of interstellar travel than us. After all, a new type of robot born and bred in the Oort cloud has a lot more going for it in terms of reaching a nearby star than any human ever will. And time will definitely work in favor of this type of evolution since the generations may well be quicker and the resources may over time come to include other solar systems.
In the interest of making a tool generally available to the space enthusiast community, I’ve recently made public a simple, interactive 3D equivalent to the graphic included in the above article, which I had recently been developing as a side project. The tool currently holds data for stars within 16 light-years, using info current to late 2011 from public domain sources. Suggestions for updates and enhancements are welcome via the hosting site: http://proximacentaurigames.com Note: the widget requires the Unity3D web plugin, which is freely available.
It’s a very interesting concept I think well worth consideration.
But I keep coming back to the same question. If an intelligent civilizatoin is thousands or even millions of years advanced, then they should be able to do things like transmit their own selves by signal at nearly the speed of light throughout their network and so be able to physically reproduce themselves everywhere within their spreading network. So again…Fermi’s Paradox rises yet again.
Obviously, they might be abiding by a Prime Directive so there could be a simple explanation. But it basically means that they would have to be choosing to not contact us rather than that they aren’t able to contact us. So if their network is aware of our situation including all of the suffering in the human race and our wish to not suffer, then they must be pretty cold-hearted to let us suffer while standing by when they could end all suffering. Does that make sense?
Eric: “The problem of making self-replicating probes that do not replicate uncontrollably.. ”
An important consequence of replication that does not remain under perpetual control is that such entities will, like life on earth, follow the Malthusian imperative. Meaning they will convert as much mass to biomass as they are capable of (and being technological, their capabilities will far exceed those of life as we know it). As a result, the vast majority of photons coming to us from any part of the sky populated by such beings will have been emitted or reflected from artificial surfaces or illumination devices — and thus quite easily detected even if we for some reason can’t pick up their communications. Much as plant life, artificial illumination, and artificial surfaces such as asphalt, shingles, paint, gold-tinted windows, etc. can be readily detected from space via spectroscopy, and sharply distinguish earth’s detailed spectroscopic signature from those of other planets with wholly natural surfaces. And even more to the point, much as an artificial satellite has a very blatantly different spectroscopic signature from any natural space object. Both highly engineered and highly evolved surfaces and illumination methods blatantly stick out from the natural universe.
So, given the premise of such easy self-replication, the rest of the premise of the cited papers falls apart unless that replication can be strictly controlled for billions of years. Of course the premise that the primary motivation for building self-replicating robots is exploration, rather than colonization, war, or many other possible motivations, is already quite shaky. All the authors can really conclude from their speculations about spread spectra is what we already know, that looking for ETI communications in our galactic neighborhoods is an incredibly poor way of searching for ETI, compared to looking at a volume of space trillions of times larger for the surfaces, illuminations, and other artifacts of advanced technologies following the Malthusian imperative.
David raises some further points about the content of his paper. I did actually read through the rest of the paper, however I am still not convinced that the proposed strategy is going to effective. I will restrict my discussion to SETI.
Discovery is key. Even fiction writers (most notably, Carl Sagan) realize that the initial discovery requires a transmission that the simplest reasonable receiver strategy will detect. It’s a bit like an “X” on an old pirate’s map, telling you to dig here, not elsewhere, to find treasure, recognizing that digging is an expensive activity that must be placed on a limited budget.
Spread spectrum does not fit the bill. Searching for white Gaussian noise rather than a monotonic carrier is problematic. It would take some serious signals analysis to distinguish true artificial from natural sources, since so much of the natural (cosmic) noise does ultimately look statistically random.
Even when you take a guess at B and T (the coarse signal parameters of the paper), you must still select an algorithm. I disagree with David that this is quite so straightforward. For example, even if you decide that an expansion of pi is a good candidate, there is still the matter of choosing the base (4, 10, 16, 27?). The algorithmic search space is vast, and even so you had better have those B and T parameters nailed, first.
Spread spectrum, as David says, is optimum. However it is optimum with respect to channel utilization. That is why commercial wireless operators use it to pack the maximum traffic into scarce and expensive spectrum. Military applications (as should be clear from Paul’s history lesson) is to be covert. That is, spread spectrum is a good method for making it difficult to both detect and decode secure wireless communications channels. This advantage works against the objectives of SETI.
While I gave short shrift to RFI in my earlier comment, I still believe it is not a serious problem. It is true that RFI does create a headache for present SETI operations due to the false positives from terrestrial and space-based commercial and (often covert) military sources. The problem with technical solutions to the elimination of false positives is that it tends to increase the occurrence of false negatives. For SETI, where candidate signals are almost certainly weak and of uncertain parameters, I have to question whether this is a good trade-off: get rid of those annoying false alarms but (continue) to detect nothing of interest.
In this regard, spread spectrum eliminates a large number of narrow-band carriers and modulated signals, but also a directed or incidental signal of similar nature from ET. Choosing to search for spread spectrum — and simultaneously eliminating conventional signals, by declaring such to be RFI — does not strike me as an effective strategy.
In our quest for life beyond earth, the search in the RF spectrum is a reasonable approach based on 20th century engineering but falls a bit short on 20th century physics from which 21st engineering will emerge. One could continue to confine the efforts simply to RF – because we can.
However, an alien civilization that is traveling through their star system of planets and even between stars would have many forms of communication beyond RF. Encountering a new life form has unexpected and unintended consequences so an advanced – and closed – communications system would be necessary. The following approach takes not just an alien view but a top down view using 2oth century physics to enable 21st century engineering. Beyond RF and EM, for an alien communications system such technologies may be common.
If I was an alien…
If I was an alien…I would need a communication system for me… and for my ship. The ship system would create a signal that could be aimed at it’s destination and survive space over great distances, the time that it takes and be self sustaining, preferably use no energy to generate. Aiming is sort of understood as a geodesic spacetime transport of a signal. One could also aim at the nearest network entry point. After all, aliens must have networks too.
If I was an alien…my preference would be to use the lightest particle able to penetrate just about anything. Photons of energy are nice but a neutrino goes thru stars with minimal gravity frequency interference. RF modulation has nothing on Neutrino Frequency modulation. After all, the real work is at and in the receiver.
Even in the simplest RF antenna, one uses array techniques. So perhaps the first step is to detect reflections or EM scattering off the array from other sources. Perhaps the alien array is simply defocused at our distance.
As an alien I would not send a beam but a planar wave in the form of a soliton. Such a wave would be self-sustaining and avoid the remanufacturing issues of successive generations. The soliton is basically regenerating continuously. In the more sophisticated forms, an array of pencil beams would appear as 2D and 3D waves. The correct antenna, processing and display would be needed.
Most likely, as an alien of an advanced civilization I would connect into a network where multipath, spread spectrum, burst and wavelet techniques were common.
If I was limited to RF or even just EM in flat spacetime, there would be major concerns about energy just to communicate. Efficiency eventually leads to a pencil beam if only to align a particle beam or focus an array effectively. However, alien engineers are not limited by an earth view of the universe that RF dominates communication within the EM spectrum. A sufficiently advanced civilization might easily use photons or other gauge bosons: graviton, and the bosons – W and Z.
In the Letpon family of electron, muon and tau paricles and each of their neutrinos, a neutrino communication system might be interesting. As an alien, engineering communications systems incorproating the weak and strong forces would be rather easy, and would necessitate pencil beam structuring to limit the explosive properties.
If I were an alien…using quarks as the basis for communications would be another approach, one quite natural for the RF folks confined to using only electrons.
If I were an alien, entanglement communication would be desirable. This form of synchronous spacetime communication at the electron level might be applied to other particles with quarks being the best candidate.
If as an alien, I might prefer to avoid normal spacetime , and create my own little bubble of spacetime where within the bubble extremely large messages could be stored, the ultimate store and forward system. If the bubble is small enough, I might even be able to cross boundaries of time. How do you detect a spacetime bubble? I can see it now…Spacetime Bubble SETI.
Finally, the zero point field communicator is an interesting alien communication device. The theory was that paired particles, in particular electron and positron pairs, were created and destroyed. Paired particle theory evolved slowly with muon and tau particles followed along with their neutrinos. Next, quark communication systems were the direct result of this ZPF research. Once ZPF theory matured to the full particle range of E8, then multiparticle communication was easy to integrate into a system.
Radio SETI…is there an Optical SETI…or an EM SETI? Perhaps there will be a Quark, Neutrino and Boson SETI as well.
Ron S, those are legitimate concerns. You however misunderstand one point. In a search we are not looking for a signal that is white noise like, but rather a very specific signal. Let me illustrate this with a different (but actually almost identical) example. I am given a sequence of a million bits, and I am trying to distinguish whether it is a million digits of pi or the result of million coin tosses. Obviously if I just look at the sequence statistically then pi and coin tosses will both look completely random. But that is not what I do. I compare the sequence to the specific sequence generated by pi. The probability that random coin tosses come out with the exact pi sequence is 2^(-1000000) — infinitesimally small. Thus, an observed sequence that matches pi is incredibly strong evidence that the sequence was generated by a technology rather than some natural process that is the equivalent of random coin tosses.
The argument for base 2 is simplicity — Occam’s razor. Sure, you could use a higher base, but there is no identifiable advantage (in terms of detectability, evidence of technology, etc) for doing so. So why use something more complicated than needed? In any case, in a search process trying different bases and/or different sequences is very cheap, although as you intimate it does increase the false alarm probability. Fortunately that increase is logarithmic due to the chi-squared statistics, so it is not too much of a problem in practice.
Your argument about spread spectrum being used covertly is valid — that is one advantage of spread spectrum. However, note that in that very same example the military is also using it successfully for their own communication. So it is hidden only from those not in the know, but perfectly visible to the intended receiver. In a military application they will use a spreading sequence that is generated cryptograpically to make interception difficult. But in civilian (or interstellar) communication the idea is to make it as transparent as possible, so that anybody or everybody can intercept it successfully. It is like the difference between shipping hundred dollar bills in an armored truck or by bicycle courier — both methods work well for the bank but the bicycle is advantageous to the crook.
The argument for SS is actually not primarily RFI. Rather, what I am looking for is fundamental principles that make strong statements about the nature of the signal. RFI does, noise doesnt. So RFI is wonderfully useful as a principle upon which to ground signal design that the TX and REC can use as an anchor to coordinate with one another without actually coordinating.
You make valid criticism that the receiver has to search over some new dimensions — time duration, bandwidth, etc. However, that is true of ANY information bearing signal, not just SS. Furthermore, when you consider other impairments (other than noise and RFI) additional information about the signal falls out. I will be posting another paper soon that examines plasma dispersion in the interstellar medium. It turns out that this impairment tells us a lot about exactly what bandwidth and symbol period to look for based on observable physical parameters. This observation applies to any information-bearing signal, including SS.
To Jean-Pierre, let me clarify that a signal can be detected at any SNR, it doesnt have to be high. With a matched filter, the signal level at the output is equal to the total signal energy. Now suppose we keep signal energy constant, but increase the bandwidth and/or time duration of the signal. The noise that overlaps the signal in time and frequency is growing, but the detectability is staying fixed. Thus, it is possible to detect that signal at any (arbitrarily small) SNR. That is a foundation of the covert nature of SS in military applications. They deliberately use an SNR so small that the enemy would not even know that the communication was happening without the secret information necessary to recover the exact signal waveform.
I like the idea of spread spectrum since it’s simple as playing the keys on a piano at RF frequencies (or optical) to produce notes or chords.
RFI does present some issues. Tracing the signal might not be so easy for RF but one could filter by frequency and then polarization. Reverse engineering RFI might prove interesting into it’s component sources. One will probably find all sorts of signal pieces that can be traced to earth, moon and sun effects or space weather in general. Earth’s magnetic field alone generates some interesting RFI.
In deep space, one might use the signal as a sensor. Information about the path and what the signal came across in the path could be discerned by a sophisticated receiver. As a sensor, information-bearing signals could be sent in order to discern the strength and other analog signal characteristics. Microwave signals were used in the 20th century to determine fog conditions between stations (10-100 miles, typically 35). Spread spectrum is perfect for use as a sensor for weather and detecting objects in the path similar to that of the trip light for a garage door opener; the path is clear or blocked.
Multipath SS on a grand scale would require knowledge of mass-in-space from stars and planets down to moons and comets.
If you are distinguishing between RFI and noise, then perhaps one should breakdown the noise even further to the components or ground. RFI may be present in the noise since components and various grounds also act as antennas. While noise signal may have a higher level of sources, one might be able to correlate and perhaps synchronize a RF signal received thru both RFI and noise sources.
The idea of pencil beam communication is one of efficiency and privacy. Navy directional signaling employs a signal lamp. One could tighten the signal further with a pencil beam RF or simply use a maser, laser or other coherent radiation device. The reason for a signal lamp was privacy, not efficiency nor speed. History has many examples of signal fires by tower or simply by smoke. But a focused signal on earth was primarily for privacy. Is privacy an issue for aliens in space?
Pencil beams suffer as a limited line of sight communication. Bending the beam on a consistent basis requires somewhat expensive systems for over the horizon or around sun or other stellar gravity well. Dark mass and dark matter probably add to the signal power and energy requirements and may exclude the use of certain frequencies.
Pencil beam comm suffers from one other assumption. What if the receiving party is NOT at the required location, or even along the path. The end result is that someone on earth – no matter how low the possibility – might receive the signal. So pencil beam communication requires in part that one senses the location first. In radar, this is the process of search, acquire and track.
If one speaks of pencil beam RF in space, the mind boggles at the energy requirements for a Spread Spectrum transmission let alone a regular one.
Over planetary and stellar distances, dispersion and diffusion issues alone will require a system with soliton capability and multipath options all the while assuming the beam is correctly aimed.
Most likely we will detect communication pieces of a planetary level cellular system before we detect a single source comm. These remnants will not be pure sine waves but tortured waves that may be shifted in frequency and may require piecing together wavelets and making the right assumptions in order just to get a signal.
Most likely, earthlings will detect a sensor signal probing the planet, moon or solar system before an intelligent information signal is detected. That sensor signal may take many forms far beyond the limitations of purely RF below the optical range or even above the terahertz range.
Dave, I don’t think we misunderstand each other on the white noise question. Perhaps I didn’t clearly say what I meant. It’s just that one starts with “noise” and then we apply spread spectrum parameters to it. If you choose right you find a candidate signal. I say “candidate” since it is perfectly possible to decode noise into what turns out to be a false positive.
I also already understood the point re pi and randomness. What I now realize I’m confused about is whether you are still talking about h(t) or the information the channel contains. This relates to what I said in the previous paragraph. The point about mathematically “expected” coding schemes apply to both.
It takes additional work to determine if, after applying a guess at the spread spectrum parameters, whether the decoded signal is noise or information. This may not be easily apparent. As anyone who has fed noise into a data receiver of one sort or another finds out, eventually, is that it can be decoded into something that looks meaningful. Except that if it is truly noise, continuing the decoding will (usually) quickly show that this is a false positive.
Regarding your point about military communications, where both ends know the algorithm (also true in commercial wireless systems), of course your point is valid. That is why I explicitly restricted my discussion to SETI where the receiver does not know the algorithm (or even whether there is an algorithm).
I still think you are understating the enormity of searching the multi-dimensional space of spread spectrum parameters, so perhaps we will just have to agree to disagree for the present.
Your forthcoming paper on effects of the ISM sounds intriguing. I know of ionospheric effects in terrestrial and ground-to-space effects such as Faraday rotation so it will be interesting to understand what to expect from interstellar communications through a thin plasma.
@ David Messerschmitt: Thanks for your clarification, I am much less qualified than you so probably I am making a big mistake here so please pardon my hubris :-)
.
[DM] “To Jean-Pierre, let me clarify that a signal can be detected at any SNR, it doesn’t have to be high. With a matched filter, the signal level at the output is equal to the total signal energy.”
[JPLR] My understanding is that given energy detected over a carrier you try to solve the problem to find if it does transmit symbols or not. So I agree with your sentence above: With a *matched* filter at the output there is the energy of the signal plus the energy of the noise. The interesting thing is : “to have a matched filter” I will come back to it later.
[DM] Now suppose we keep signal energy constant, but increase the bandwidth and/or time duration of the signal. The noise that overlaps the signal in time and frequency is growing, but the detectability is staying fixed.
[JPLR] OK there is no change in the S/N ratio: There is more noise energy but there is more signal energy. Note this is not the case with frequency-hopping spread spectrum. But it is the case for direct-sequence spread spectrum which is the subject of your paper: “a signal with a noise-like character and much higher bandwidth than necessary”. DSSS is useful to share a channel with other emitters at the expense of energy increase which is not the case in FSSS. FHSS offers the same functionality without having the need of a large channel, it is used in Wi-Fi and Bluetooth.
[DM] Thus, it is possible to detect that signal at any (arbitrarily small) SNR.
[JPLR] Ron S also tells it but added a condition: The lower the S/N ratio the longer it takes to send a signal. I think the proper technical name for this is “process gain”.
In my understanding it implies ***the original S/N ratio has to be over unity***
, this is not my field but I expect some people would strongly complain against a S/N ration arbitrary near zero.
It seems from Wikipedia there are other issues about S/N ratio in process gain: http://en.wikipedia.org/wiki/Process_gain.
I agree with Ron S but I also said that the word “noise” is misleading, there are at least two notions of noise. One for the transmission of symbols over a carrier and you are referring to this (you consider the output of a *matched* filter). The other is about detecting the existence of carrier over the electromagnetic noise without considering symbol transmission. Perhaps it’s more a filter problem than a symbol transmission problem (and indeed you don’t address the filter problem anyway because the paper is about matched filters).
[DM] That is a foundation of the covert nature of SS in military applications. They deliberately use an SNR so small that the enemy would not even know that the communication was happening without the secret information necessary to recover the exact signal waveform.
[JPLR] Here two notions of S/N ratio are IMO mixed, and that was the main point in my first message. OK the communication is scrambled, you even write in the above sentence (and I agree with you) that it’s by cryptographic means so therefore it’s not because it uses spread spectrum technology. But it’s still possible to know that something is happening because someone is pouring a lot of energy over a carrier. For SETI that would be enough, we don’t have to decode a communication that anyway we won’t understand: Symbols give no clues about semantics/languages).
PS: Other considerations poorly linked to your kind answer but with a link to SETI quest:
What does mean using arbitrary low S/N ratio in the context of SETI enabled communication? As an emitter how can you know it would be low energy at the receiver if you don’t know where is the receiver?
Narrow filters for low S/N ratio can work only if a huge part of the surrounding spectrum have the same S/N ratio as highlighted by the LightSquared vs GPS debacle, meaning the only place we could listen to interstellar civilization is in the interstellar media far away from any solar system.
JPLR: “In my understanding it implies ***the original S/N ratio has to be over unity***, this is not my field but I expect some people would strongly complain against a S/N ration arbitrary near zero.”
Let me respond to this since, in part, it was addressed at what I said.
As I already said, it is misleading to say a signal is “above” or “below” the noise since noise is not at a static amplitude. In fact it shows a high statistical variance. We can speak of “average” noise or “peak” noise, but this refers to some method of time averaging (integration). Noise becomes a consideration in any receiver when the signal is low enough that the noise can affect accurate reception of the signal (symbol detection and decoding).
Nothing magic happens when the average signal amplitude slips from positive to negative with respect to the average noise amplitude; the impact of noise changes continuously as signal amplitude declines.
As the noise increasingly dominates the signal, successful recovery of the signal, including symbol decoding, requires either or both of narrow filters and longer integration times. These necessarily reduce the maximum symbol rate (though by different mechanisms). Once the receiver’s “filters” reduce the maximum symbol rate below that contained in the transmission, you will suffer high degradation in symbol reception.
If you can negotiate symbol rate with the transmitter, that’s good. If not (such as for SETI), the best you can hope for is to detect that there is a signal but not what it encodes, if anything.
@Ron S: I completely agree with you, Ron. English is not my native language so perhaps I was not very clear, but what you wrote in the last sentence of your last post summarizes well my initial point. As anyway we won’t understand the meaning of a SETI intercepted communication (sometimes it requires decades of effort to understand some long forgotten human languages), detecting there is an artificial signal is enough which BTW is anything but a trivial task.