Forward Error Correction

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

ad: L-HROutlet
ad: l-rl
ad: L-MFJ
ad: Subscribe
ad: Left-2
ad: Left-3
ad: MessiPaoloni-1
  1. KK5JY

    KK5JY Premium Subscriber QRZ Page

    For such applications, FEC isn't the appropriate tool. For such applications, you use either a digital signature or at least a good CRC on the message, which is transmitted with the message, which tells you if you got the entire message correctly or not.
     
    K2CAJ, KU4X, W0PV and 1 other person like this.
  2. KV6O

    KV6O Ham Member QRZ Page

    There are methods/tools that allow you to insure the message is complete, unmolested (due to corruption or by intent), and enables the authentication of the sender. These methods are cryptographic in nature, but are NOT encryption, so use on amateur bands is completely acceptable.
     
  3. KX4Z

    KX4Z Ham Member QRZ Page

    So.... what do you do when you DID NOT get the entire message correctly? Quit? At least that digital signature might tell you that the message was NOT to connection the B+ wire to lug #123....but that something had gotten lost in the middle....but how do you ever get it FIXED?

    On voice, or CW, you simply ask for a FILL when you know you have gotten a corrupted message. What do you do with FEC?
     
  4. KX4Z

    KX4Z Ham Member QRZ Page

    Hi, -- are those the "digital signature" or similar that another person cited? But again, the same question I asked him -- what do you DO when you know the message is incomplete or molested -- how do you solve that with FEC when it has already failed?
     
  5. KX4Z

    KX4Z Ham Member QRZ Page

    Am I correct that the autosquencing in FT8 demands an acknowledgement ("R") before proceeding? That system uses quite a bit of FEC

    "Forward error correction (FEC) in FT8 uses a low-density parity check (LDPC) code with 75 information bits, a 12-bit cyclic redundancy check (CRC), and 87 parity bits making a 174-bit codeword."

    but is still apparently dependent (just like voice and CW operators) on acknowledgement from the other end that the message was received correctly. Am I missing anything here? I just don't see FEC's effectiveness as being 100%, and it seems that some form of acknowledgement is almost universally required. (i was getting ready for an Extra class course we are teaching and there were questions on this sort of stuff and it piqued my thought process. E8C01 How is Forward Error Correction implemented? )
     
  6. SM0AOM

    SM0AOM Ham Member QRZ Page

    If you do not have a return channel, you have to live with the situation. If there is a return channel, you can send a message back and ask for a retransmission of the whole or parts of the message.

    Actually, this was a major discussion point in the childhood of the GMDSS, when a change from Morse and telephony in distress traffic to various forms of digital messaging was foreseen.

    At this time (early-80s) the major mode for MF/HF communications except 'phone and Morse was CCIR 476 ARQ/FEC radiotelex.

    Using ARQ precluded communications between more than two participants at a time, and distress follow-up and SAR coordination more often than not involved several participants.

    'Phone and Morse have the advantages of being real-time, but is dependent on operator proficiency and a decently good RF channel, and FEC has a phasing period which adds delay and overhead to short messages, but instead works well with low S/N values and also indicate errors.

    A lot of international effort went into preparing procedures that catered for the increased delays and overhead when CCIR 476 FEC was to be used in distress communications.

    73/
    Karl-Arne
    SM0AOM
     
    Last edited: Jul 12, 2019
  7. KX4Z

    KX4Z Ham Member QRZ Page

    Great point, Karl-Arne! If you have a return channel, it seems obvious to get a fill..... but one advantage of FEC alone is that it remains a broadcast technique allowing multiple recipients. My marine VHF radios have an auto-distress call digital button which probably has FEC built in -- but then the message is very very short and repetitive, and is a broadcast type thing: "so-and-so is in distress at lat/longitude: xxxxxxx " over and over. FEC seems appropriate for that.

    But FEC alone -- still does not seem to be able to avoid the consequences of dropout. Add some digital signature and you can at least RECOGNIZE the dropout....but still cannot ffix it. You are telling me that just like CW and voice, digital was using some form of fills (ARQ) from the beginning. There is a question about ARQ in the extra class stuff i'm getting ready to teach, also. The use of FEC alone seems to be something best suited for one-way communications like TV or my auto-distress button on my boat radios.
     
  8. SM0AOM

    SM0AOM Ham Member QRZ Page

    The "distress button" on a marine radio releases a "canned message" using the ITU-R 493 and 541 Digital Selective Calling (DSC) protocols (which I am a co-designer of in the ITU-R study groups).

    These protocols are derivatives of the FEC mode in the
    CCIR 476 radiotelex system but expanded to handle 7-bit binary data.

    To increase the detection probability, the DSC distress message is repeated five times (0r ten times taking time-diversity into account), and in the case of MF/HF multi-frequency distress alerting, repeated five times on each frequency in a sequential pattern (a "distress call attempt") until acknowledged.

    Some use of FEC, in the form of an error-detection code or system, is essential in any ARQ system, in order to find out if the message has been corrupted or not.

    In the usual parlance, "ARQ" is reserved for system where the error detection and the request for repeats are automatic processes, but it is not strictly essential.

    73/
    Karl-Arne
    SM0AOM
     
    KA2CZU and KX4Z like this.
  9. AI3V

    AI3V Ham Member QRZ Page

    Back when I worked for Comsat, we had 2 basic types of fec on the 64kb/sec channels we provided.

    A "7/8 rate " codec (COder DECoder) that sent 7 data bits and 1 fec bit, so the customer got 56 kb/sec

    And a "1/2 rate" codec that sent 4 data bits and 4 fec bits, so the customer got 32kb/sec.

    The 7/8 was used with stations that had 100 foot diameter antennas, and the 1/2 used for 30 foot diameter antennas.

    There are as many fec schemes as there are possible bit combinations.

    Rege
     
    KX4O likes this.
  10. KV6O

    KV6O Ham Member QRZ Page

    This is being discussed at a high level here... FEC is usually used in blocks or packets to allow for corrections to corruption in the packet. If a block cannot be corrected, then (assuming a 2 way link), the block can be requested to be resent. At the end of all of this, all the blocks are assembled into the message. A signature can be used to authenticate the entire message. The entire message has to be there for the signature to be computed against what was sent, so it's "proof" that the message is complete and unmolested. If error correction fails, or the authentication of the message fails, you would have to request that it be resent. That's why breaking a larger message into blocks makes sense - you can perform these checks at a smaller level and resend only what needs to be resent.
     
    Last edited: Jul 12, 2019
    KX4O likes this.

Share This Page

ad: M2Ant-1