Forward Error Correction

Discussion in 'Ham Radio Discussions' started by KX4Z, Jul 11, 2019.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: Left-3
ad: Subscribe
ad: l-BCInc
ad: L-MFJ
  1. KX4Z

    KX4Z Ham Member QRZ Page

    I think that most of the world understands that some combination of forward error correction and some form of acknowledgement/request turns out to be optimal....unless the information is of zero value to get across anyway.

    You can see that in amateur radio from CW, through AM, SSB, and into the digital realms as well. You see it in virtually every contest, and likely in many many ordinary QSO's. I'm not an expert, but I understand it is utilized in a ton of commercial systems as well. It is pretty standard. Of course, you CANNOT do it if you have only a one-way channel, like a broadcast radio station, and then you switch to FEC....and my picture still completely drops out on our TV every time my packet station comes on because the two antennas are very near each there is a complete loss of message at that point. Since the TV isn't of that great an importance, we don't have to get that terribly upset (tho I did add some filters to improve it a bit and may yet work more on it) -- complete overload of the TV preamp up there....

    I have no idea why something that has, gee, maybe 100 years of history behind it, is now dismissed by "experts". Experts who end up being corrected by those who actually know how some systems work, as well. If you've got a better way to do it --- then my hat is off to you--- go build your product, get it on the market, patent it (or give it away if that is your desire, it is your stuff, not mine!) -- and then the world will be a better place. You may or may not be able to retire, depending on whether you agree with intellectual property rights or not, or are just incredibly generous....

    easy to talk about -- -harder to actually bring something to market!!


  2. SM0AOM

    SM0AOM Ham Member QRZ Page

    You are entirely right that the Shannon capacity limit strictly is valid for time-invariant channels only, or at least for some form of stationary state.

    Among other causes, this is one of reasons that it is more difficult to design an HF modem than other modems, where the channel is much less time-variant and less dispersive.

    Also, there quickly turns up compromises among the depth of interleavers, block lengths and amounts of synchronising or training sequences.
    The error distribution often is such that errors occur in bursts, which makes choices of FEC and time-diversity parameters much harder. Interleavers may make this better, by breaking up error bursts into individual bit errors at the expense of increased latency.

    Taken together, the practically achievable amounts of throughput for a given amount of signal power and channel bandwidth is much lower than the Shannon limit will predict, even if an optimal code should be devised.

    Designers of HF communications have to handle these conflicting requirements, and this has spurred many protocol designs that aim to work around the limitations, and to use the channel optimally.

    One problem is that short data blocks with a lot of parity in each are wasteful of bandwidth when the channel is good, and their error correction capabilities are not taxed.

    Soon became adaptive schemes devised, where the modulation formats, block lengths and parity vs data distributions can be adjusted to suit the actual channel properties. Such schemes are the key of modern HF data communications, but are dependent on a return channel in some form.

    If there is no return channel in any form, such as in vision/sound broadcasting and in some military communications, the only tractable solution is to make some form of "intelligent guess" about the channel characteristics and an expected signal-to-noise ratio, and adapt the allowed FEC overhead and modulation formats to the length, permitted latency and importance of the messages.

    In conventional broadcasting, this is quite easy, but not in military communications. Untold numbers of man-hours have been spent into devising schemes for reliable message distribution to radio-silent users such as submarines. The major world powers today use a mixture of techniques, ranging from low-speed VLF to high-capacity HF or VHF circuits.

    N0TZU likes this.
  3. KX4Z

    KX4Z Ham Member QRZ Page

    Well, yet another time I have learned a great deal from you! Thank you so much
  4. K2CAJ

    K2CAJ XML Subscriber QRZ Page

    The simplest case in Shannon's seminal paper has simple assumptions (a discrete memoryless channel), but Shannon's theory has been applied effectively to continuous-time channels, time-varying channels, rapidly fading channels, multiple access channels, channels with side information, channels with feedback, etc etc.

    As you can imagine, there has always been huge commercial interest in maximizing throughput, so you can fill trucks with all the papers and dissertations devoted to maximizing capacity of real-world multiple-access wireless communications channels. Thus there is no single set of assumptions under which Information theory applies.

    Those manufacturers all use Shannon's theory. Whether or not a channel has feedback, modern digital communications use coding methods and channel models informed by Shannon information theory.

    And of course the proof of the pudding is in the eating: we are surrounded by smart phones, wireless devices, digital televisions, and entropy-coded media like DVDs that clearly work very well.
  5. K0IDT

    K0IDT Ham Member QRZ Page

    I was specific enough on the first try when I responded to Matthew, your reading skills haven't improved.

    Do you mean your latest FCC filing? I guess I really should thank you for that. Your use/overuse of "spying" and "snooping" implies a degree of secrecy or privacy in the only mode you used for the "proof". You proved something but I'm not sure what. You should be ashamed of yourself for dragging in the other two guys for witnesses.

    Please stop with the viewer excuse. If you knew half of what was going on behind the scenes you may want to rethink who you associate with and try to defend. Winlink putting the viewer up was seen as a sign of a good faith effort
    according to some, but if you dig deep enough you'll find it was a CYA until things blow over, then it's back to business as usual--a common carrier email service.

    Gordon it's pointless to try and discuss anything further with you, you're right even when you're wrong. Look for a reply on the ECFS.
  6. KX4Z

    KX4Z Ham Member QRZ Page

    There’s nothing wrong with Shannon‘s theory. You just didn’t address the issues that I raised, which Karl-Arne also addressed.
  7. KX4Z

    KX4Z Ham Member QRZ Page

    Yes, it may be pointless. There are things waiting to be admitted.
    1. There is no encryption routine in any of John Wiseman's code; no key. How would anyone in their right mind call it encrypted or anything close?
    2 The method to decode winlink fairly straightforwardly has now been proposed and published. What is missing is an explanation why self-proclaimed "experts" who claimed this was a national security issue....didn't do it. Or any of the others who hold those views.
    3. The claims of significant interfenrence from USA 97.221(c) stations -- have been proven false with actual data for the WINLINK users, and no other large group has been identified to replace them as possible interfering stations.... Waiting to be admitted.
    4. The claims that ARQ is not needed...have been refuted by a world expert. Waiting to be admitted.
    5. The fact that improvements in efficiency may quite wll make decoding more difficult -- I believe that a proclaimed expert (to his credit) is now acknowledging that -- so the efforts to oppose progress in the radio art may possibly weaken.
    6. The ability to read any and all winlink emails on a web portal has been demonstrated; so now the FCC can do whatever it pleases.

  8. W6EM

    W6EM Ham Member QRZ Page

    Dr. Giddy, you simply are amazing! You and your buddies including Greg Sa_rat conduct bedroom-distance Winlink connects and brag that you've done it! You've copied a Winlink-compressed message pass. Not once, but now five times!!! Oh, and in your latest White Paper brag session, you conveniently include that all it takes is a missing BIT to throw off the SCS modem's monitor mode into it all looking like scrambled eggs. At least you were honest enough to include that caveat.

    Yes, others in Winlink hierarchy, like Rick Muething, have explained long ago that it takes PERFECT COPY of a Winlink exchange, meaning no corrupted or missing BITS to decode it. Do you really think trying to sell a paperback book about how you can sit in your bedroom and prove that it can be decoded will convince anybody? A foolish errand, IMO.

    HF is not solid copy. Fading, multipath, etc., produce missing BITS all the time. Of course, not a problem across a bedroom. Effectively encrypted it still remains in the real world of third party copy.

    Nice try. See you in the Proceeding records.

  9. KX4Z

    KX4Z Ham Member QRZ Page

    And good afternoon to you, also!

    Yes -- all we had to do was show ONCE that Winlink can be intercepted. That alone proves it is NOT encrypted. It is certainly compressed, as you'll find explained over and over and over. Gee, I'm not an expert on SCS modems, but if they don't get a copy that the FEC (forward error correction) can fix...they must get a retry (just like someone in a contest). Honesty is important here -- so getting the facts out is very, very useful.

    That's why it makes zero difference how far we had to communicate -- our goal was to prove it is possible. Since you know that 50,000 emails move through this system roughly a month -- it is pretty obvious that it can work. Now you have to have enough receivers so that you can re-assemble a compete packet stream. That is pretty well explained in Chapter 1 of the text, and I think in some FCC filings as well. Diversity receiver systems have been around since at least 1944 (I put a photo of one in the text, just so everyone could see).

    You are certainly welcome to call it foolish to show how things claimed....are false. You can certainly explain why HF has problems -- that's the whole reason that WINMOR and PACTOR and other modes were valuable creations for ham radio. (Pactor III multiple modes explained here: Protocol.pdf ) But you have failed completely to justify any use of the word encryption. Difficult -- yes. Requires work -- yes. But encrypted? No.

    Here is a definition of encryption:
    In computing, encryption is the method by which plaintext or any other type of data is converted from a readable form to an encoded version that can only be decoded by another entity if they have access to a decryption key

    Go ahead and show us the encryption routine.
    So, honesty is certainly a great quality. More power to you! I believe I explained very carefully what is required. What is why it hasn't already been done by the people who told us that our national security depended on it. You were not one of them, were you?

  10. KX4Z

    KX4Z Ham Member QRZ Page

    Folks, I know this back and forth has not been fun reading....painful, actually. But we are actually making progress.

    Because Lee, W6EM just made a VERY STARTLING STATEMENT. He basically now agrees with what many of us have been trying to get into the record. Although he demands to still use the words "effectively encrypted" -- he actually gave the entire argument away with his words before that:

    "HF is not solid copy. Fading, multipath, etc, produce missing BITS all the time."

    And there you have it. Lee is agreeing with me that this is now just an engineering problem. If you can get a good signal -- he is now agreeing that you can read WINLINK. And here are the things that he knows you have going for you:

    a) There is FEC inside the communications -- so there are actually times when you, the monitoring station, will catch the packet -- and the intended receiver will not, and you'll watch as THEY ask for a repeat! Because of the error correction you have a far better chance than just if one bit is missing -- that is the whole point that several people above have been ardently making: FEC is a huge help!
    b) And you have the advantage of putting together a diversity network, even if all you use are SDR's! The intended recipient probably has only ONE receiver -- you can have 2, 4, 6, 8 -- as many as you can get to help you! You may have a receiver with PERFECT copy by chance, or you may have several poor ones. But you have the ability to check every one of their packets and see who got it. The intended recipient -- DOESNT.​

    There is nothing new about that -- I pointed out a vacuum tube diversity receiver from 1944.

    This is huge progress. Lee has now pointed out the truth -- this is now just an engineering project. There is nothing TRULY encrypted -- the only real issue is getting receivers that can allow you to package up a complete stream of packets. The reason WE had trouble was we couldn't do any of that, were dependent on a crazy modem busy bouncing between Transmit and Receive, real fast, and slowly losing sync. You won't have that problem at all with your Monitoring System.

    Lee has moved this WAY farther forward. It is now a discussion of an engineering project. Can it be done? What are the probabilities? And it is time that some people who think it worth doing, begin! This could be a great masters' or PhD project. Hooray!!!



Share This Page