4

tl;dr:

Question: Have deep-space spacecraft always use some form of spread-spectrum for data downlink?

note: I'm looking for some insight into why, and any possible exceptions, not just a "yes" or "no". Thanks!

I was telling someone that the use of spread-spectrum in one form or another has become almost universal in civilian wireless communication and signaling (cell phone voice and 3G/4G/5G, WiFi, Bluetooth, GPS, ZigBee, LoRa (e.g. Understanding the relationship between LoRa chips, chirps, symbols and bit) etc. and it is interesting to note that WiFi and its resistance to multi-path interference has its origins in a couple of radio astronomers from Australia's CSRIO!

In April 2009, 14 technology companies agreed to pay CSIRO \$1 billion for infringements on CSIRO patents. This led to Australia labeling Wi-Fi as an Australian invention, though this has been the subject of some controversy. CSIRO won a further \$220 million settlement for Wi-Fi patent-infringements in 2012 with global firms in the United States required to pay the CSIRO licensing rights estimated to be worth an additional \$1 billion in royalties. In 2016, the wireless local area network Test Bed was chosen as Australia's contribution to the exhibition A History of the World in 100 Objects held in the National Museum of Australia.

I believe that the Gold code used in GPS can trace part of its origins to NASA's deep-space communications as well, though I can't find a reference for that right now. Note also that the importance of the Gold code is for timing reconstruction using correlation in the same way that NASA used it for range/rate, as well as for signal/noise benefits; see this and this answer for example.

Spread spectrum has many advantages, and one of them is signal-to-noise, which can be understood in terms of the Shannon-Hartley Theorem (see Am I using Shannon-Hartley Theorem and thermal noise correctly here?), and I believe that Voyager's data downlink always employs a bandwidth wider than than its bits-per-second rate would require, but I'm not sure; see the last paragraph in this answer to "How to calculate data rate of Voyager 1?" for example.

uhoh
  • 148,791
  • 53
  • 476
  • 1,473
  • Just to note, instead of having to add the "$xyz" in a code block, you can add a back slash (\) in front of the dollar sign ($xyz) to prevent it from rendering as Latex or whatever funky script it's rendering as. – Edlothiad Mar 02 '18 at 09:32
  • @Edlothiad excellent! So it's the same as the MathJax \$ in some other sites (e.g. Electronics) just backwards. – uhoh Mar 02 '18 at 11:30
  • 2
    I don't know MathJax, I just know the \ means I can stop things from rendering, like *\*waves** allows me to make the word "waves" italic and keep the second set of asterisks: **waves** – Edlothiad Mar 02 '18 at 11:47
  • @Edlothiad left is MathJax in this site, right is MathJax in Electronics SE. https://i.stack.imgur.com/GpXKs.png It's the "same thing except backwards". – uhoh Mar 03 '18 at 05:43
  • Well if that isn’t just confusing... – Edlothiad Mar 03 '18 at 06:55
  • Are you sure Voyager uses spread spectrum techniques? Voyager uses phase modulation and this document makes no mention of spread-spectrum. – Hobbes Mar 12 '18 at 12:48
  • @Hobbes No, and if you re-read the last sentence, which is the last paragraph as well, I was careful not to say it did. I'm drawing upon my memory, and this is exactly why I am asking the question "Have deep-space spacecraft always used some form...". There was also a pseudo random code somewhere in the reading I did about the Voyagers RF system for this unanswered question, but now that I think carefully, I'm not sure if it was in the downlink data channel, or a Gold code generated on Earth for the two-way doppler range-rate measurements. – uhoh Mar 12 '18 at 15:32
  • Your sentence "I'll do some handwaving here and just estimate the bandwidth used by Voyager's spread-spectrum transmission to be about 1 kHz, " - does indicate Voyager uses spread spectrum. If that was not your intent, you'll have to rephrase that sentence. – Hobbes Mar 12 '18 at 15:37
  • @Hobbes Oh, I think you're talking about a sentence that appears on another page, my answer posted back in January, that I've linked to here. It seems that I was sure then, and have grown less sure since. Why don't you leave a comment below that answer, and then I'll edit the answer and refer to your comment? Thanks by the way! – uhoh Mar 12 '18 at 15:45
  • @uhoh: I appreciate your gentle reminder. This book is resting on my shelf and it might have the answer. I just don't have the time to investigate right now. – Andreas Mar 12 '18 at 20:34
  • @Andreas thanks! I don't know what a.co is; does the book have a title and/or author? – uhoh Mar 12 '18 at 22:21
  • 1
    @uhoh: Oops, link works for me (shortened Amazon URL). The title is "Deep Space Communications" / Jim Taylor(editor), Wiley 2016. Covers selected NASA missions. – Andreas Mar 12 '18 at 22:28

1 Answers1

4

Pseudo-random codes are used for ranging, i.e. to calculate the round-trip time of a signal. That takes up some bandwidth, so I would consider that to be spread-spectrum. That has been in use since, I guess, the 60's.

I am not aware of any applications of spread-spectrum techniques in deep space data communications (as opposed to tracking). Sure, they will use wide bandwidths, just because they can. So the modulation scheme is chosen for performance, not for minimizing spillage into adjacent bands as is often a concern on Earth. However spread spectrum means not just a wide bandwidth, but also some sort of time-dependent coding or other structure used at the transmitter and then necessarily duplicated at the receiving end in order to decode the signal through correlation. That is what I am not aware of in deep space comm.

For reference, the definition of spread spectrum as quoted in this paper:

Spread spectrum is a means of transmission in which the signal occupies a bandwidth in excess of the minimum necessary to send the information; the band spread is accomplished by means of a code which is independent of the data, and a synchronized reception with the code at the receiver is used for despreading and subsequent data recovery.

Mark Adler
  • 58,160
  • 3
  • 171
  • 251
  • If I understand correctly, Shannon-Harltey sez (and by "sez" I mean I'm paraphrasing) that if you spread the spectrum of your signal out across a bandwidth significantly larger than you need (with some constraints on how you do it) you can improve your signal to noise. Now I don't know if that alone counts as a "spread-spectrum* technique, but I think it does. The coding can be just a way to do the spreading so that the receiver will know how to decode. It doesn't have to be a "secret" code. This is just my own take though – uhoh Mar 12 '18 at 19:38
  • If there is another term that is used for intentionally spreading a signal out over a larger range of frequencies in order to take advantage of Shannon-Hartley and the improved S/N, and the term is something other than "spread-spectrum", then that is the term that I would like to use in my question. So far, I don't know of any other term. – uhoh Mar 12 '18 at 19:43
  • I think a difficulty here is that spread-spectrum offers several different benefits at the same time. It offers S/N improvement, but also immunity to interference by other artificial signals, and some immunity to multi-path interference as well. I think this is a really interesting and under-appreciated question. – uhoh Mar 12 '18 at 20:07
  • 1
    The key here is that in spread spectrum the 'a code which is independent of the data'. If you increase the bandwidth with a code that is dependant on the data (i.e. an error correcting code) then increasing the bandwidth helps SNR; if it is independent then it just shuffles the noise and gives no advantage. Spread spectrum is great against narrowband interference, but doesn't help against broadband noise. It is also good for tracking (hence its use for ranging & GPS). And for multiple access - but that's not currently a problem for deep space missions... – Paul Norridge Mar 13 '18 at 17:42
  • @PaulNorridge I'll look into that but I'm not sure all of that is true; I am pretty sure that Shanon-Hartley does not require a code that "depends on the data" (such as an error correction code) in order to improve S/N. Give me a few days to post a better reply though, I'm swamped at the moment. In the mean time if you can find a link to a clear explanation of why that is so, I'd greatly appreciate it! – uhoh Mar 19 '18 at 07:36
  • @PaulNorridge see here, here and here for everything I currently know about LORA, where as far as I know the frequency modulation sequence (on top of other coding) is used explicitly for improved S/N à la Shannon-Hartley. (datasheet and applications note). First linked Q also has helpful in-depth video and ppt. – uhoh Mar 19 '18 at 07:44
  • @PaulNorridge the paper Understanding the limits of LoRaWAN is helpful as well. – uhoh Mar 19 '18 at 07:50
  • 1
    @uhoh The key point glossed over in that application note (leaving aside the questionable maths) is that the bandwidth increase only helps when the SNR is fixed. But normally an increase in bandwidth also increases the noise by the same factor, so you have no benefit. As far as I can see, LoRa just increases the SNR by effectively lowering the rate of the useful data, so increasing the energy per data bit. But that isn't dependant on spread spectrum. You could decrease the data rate on an unspread signal and see the same effect. – Paul Norridge Mar 19 '18 at 22:47
  • @PaulNorridge can you give me a link or some reference that supports this and explains it in more depth? Also, would it be possible to articulate more clearly just which aspect of the maths is questionable, and what that question would be? If you see something that is wrong, please point it out. Thanks! – uhoh Mar 19 '18 at 23:11
  • 1
    @uhoh On the maths: although the conclusion that C/B is proportional to S/N is correct, the explanations of how equation 2 & 3 are derived don't make sense (or, at least, they are very confused). – Paul Norridge Mar 20 '18 at 23:26
  • 1
    @uhoh More importantly, C is the potential capacity of the link, not the guaranteed performance. So, "only the transmitted signal bandwidth need be increased" is not correct; you need some way to exploit that bandwidth effectively. And, you are relying on the S/N not changing. If you look at the wikipedia page for S–H there is an example showing why C is typically independent of bandwidth in low SNR. A similar argument works for spread spectrum. – Paul Norridge Mar 20 '18 at 23:27