How many ham systems use compression?

Discussion in 'Ham Radio Discussions' started by KX4Z, Sep 6, 2019.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: MessiPaoloni-1
ad: Left-3
ad: Left-2
ad: L-MFJ
ad: Subscribe
  1. W2JKT

    W2JKT Ham Member QRZ Page

    I think I know why you are asking this question, but I think you need to ask it more specifically...

    You have missed some of the compressed ham systems that would actually support you in your argument, and those are D-STAR, Fusion, and DMR... which all use closed, non-free compression via the AMBE codec.
     
    K0UO likes this.
  2. W0PV

    W0PV Ham Member QRZ Page

    K0UO likes this.
  3. KX4Z

    KX4Z Ham Member QRZ Page

    THANKS!!! I became aware of some QUITE significant myths and misunderstandings held even by some in quite high leadership positions and so i wanted to try and write up something that would do a better job of explaining the true facts. Thank you for pointing out those additional systems, which indeed I failed to include. Aren't all of those somewhat of a "fixed" compression system? They don't change with each new instance, do they? i want to be sure that *I* understand things as well!!!!
     
  4. KX4Z

    KX4Z Ham Member QRZ Page

  5. WB6CXC

    WB6CXC Ham Member QRZ Page

    Yes, JS8 and those others use a static compression table or algorithm. I asked Jordan about his "dense code", since this large table contains a huge variety of words (definitely not ham-centric). Apparently Google can give you a list of word frequencies for general english text, much like the "ETAOIN ..." letter frequency. JS8 assigns these words a Huffman-style variable bit code, and obviously provides a way to handle the not-in-dictionary case.

    JS8 performs two trial encodings of a plain-text message, dense-code and Huffman character-based, and then chooses the most efficient one to send. There are flag bits in the message structure that indicate which compression method is being used. So in that case you might consider the compression to be a little bit dynamic, but it still uses fixed tables, not those built and modified on the fly.
     
  6. SM0XHJ

    SM0XHJ Ham Member QRZ Page

    Audio compression algorithms generally use dynamic filter banks who's coefficients are transmitted as part of the compressed data. If that is the case with the AMBE codec I don't know.
     
  7. KX4Z

    KX4Z Ham Member QRZ Page

    Very interesting! Although the tables may be static, The Huffman encoding probably means that if you miss a portion of it, you lose some ability to decode the remainder ; probably not as much as with LZHUF, but you might lose the entire synchronization of your bits
     
  8. KX4Z

    KX4Z Ham Member QRZ Page

    Wow! Then they are much more like the case I was discussing then I realized !!!! Wished I had known that a few hours ago Such is life!!! We all do the best we can
     
  9. WB6CXC

    WB6CXC Ham Member QRZ Page

    There are ways to re-synchronize Huffman codes when bits are dropped or errored, but given the short frame length I doubt if JS8 does this. Of course, JS8 (and FT8, etc) do use pretty robust forward error correction. There is no ARQ.
     
  10. W0PV

    W0PV Ham Member QRZ Page

    Perhaps no ARQ per se. But the auto-sequencing of WSJT on FT modes during a QSO creates an ARQ-like result, ie, a calling station that does not successfully decode the next in sequence message, also used as an ack, from its QSO partner, will repeat its last xmsn until it does. That in turn stimulates repeats from the other. Only when an "RR73" or equivalent is xmitted and successfully received are both sides satisfied.

    This can be problematic too. Only one station might log a seemingly partial contact or recalls starting from the beginning after only obtaining partial QSO data can result in dupes or unnecessary emissions (QRM).

    To my limited knowledge playing with an early edition, JS8 does not ever do that, pause for an acknowledgement or request a repeat if no successful decode, perhaps a reason it has not caught on as fast.
     
Thread Status:
Not open for further replies.

Share This Page

ad: w5yi