Loading ...
Sorry, an error occurred while loading the content.

V4 Interleave

Expand Messages
  • Daniel V. Wolff, Jr.
    Hello Rick. I have a question about the interleave method you are using? I think I understand the term interleave but I d like to be sure, as you ve
    Message 1 of 2 , May 24, 2011
    • 0 Attachment

      Hello Rick.

       

      I have a question about the interleave method you are using?  I think I understand the term “interleave” but I’d like to be sure, as you’ve implemented it.

       

      In the RTTY days of old there used to be something frequently used in commercial and military called a “Time-Division Multiplexer” or also the “Time Diversity Modem” (TDM).  The one we TDM is a mode resembling FEC AMTOR. TDM uses one sub-carrier, but assigns separate data channels to different time slots. The MD-1142 (aka TDM BR 6029C) has two modes of operation both are full duplex (the receiver and transmitter sections are fully independent), in the Internal Multiplexer Mode a single TTY signal is fed to 7 different channels and the signal in each channel is time delayed by 1 second form all the other channels.  In addition there is a pilot tone at 561.1 Hz and the MARK/SPACE FSK shift of each tone was 85 Hz (if I recall).  This mode of operation provides a high probability of getting the message through.  In the External Multiplexer Mode each of the 8 channels is fed or receives a separate data stream so it does not have the redundancy of the IMM. In time order the channels are:

       

      Chan #    Center Freq      Delay sec.

         3             1530 Hz             0

         7             2890 Hz             1

         1               850 Hz             2

         4             1870 Hz             3

         6             2550 Hz             4

         2             1190 Hz             5

         5             2210 Hz             7

         8               561.1 Hz       pilot tone

       

      Each of the channels transmitted the same information but each delayed by 1 second.  The receiving TDM would receive all channels but vote on the best five matching channels to give an output.  The redundant transmitted data on different tone pairs with the delay insured to get the data through on noisy channels. The result was an error free transmission nearly every time.  The TDM worked very well for RTTY and I always thought they’d be good for implementation for the new PSK modes used in ham radio but never saw them. There is an initial 8 second latency from the start of transmission but that was never a problem for a long broadcast. Do you remember the TDM?  They were pretty slick.

       

      Anyway, does your interleave work in a similar manner, assigning the same data to different channels with a delay between channels?  I’d like to understand better how you’ve implemented the interleave process in V4.

       

      Also, could you send me the V4 ARQ version so I can test it in Europe?  Thanks.

       


      73,

       

      Daniel V. Wolff, Jr.
      DL4VCW / KA7AGN / AEM1WF (MARS)
      http://www.qsl.net/dl4vcw/

       

      Classification:  UNCLASSIFIED
      Caveats: NONE

       

       

       

      A few brave soles need to test the new V4 ARQ version

      Posted by: "Rick Muething" rmuething@...   kn6kb

      Mon May 23, 2011 1:32 pm (PDT)



      All,

      I am in the final stages of testing the new V4 ARQ mode. This is a nice
      improvement allowing two stations to exchange error free data (after FEC and
      ARQ) even in pretty poor conditions. I am in the final stages of testing
      and updating the help files for V4 and V4Chat. ARQ is a few percent slower
      than FEC (due to the ACK/NAK) but gives perfect copy (or no copy at all if
      things are really bad!).

      I would like a few (no more than 5) brave individuals with some experience
      with the V4 FEC mode to test out the install, help and test ARQ and FEC
      operation before posting the files for the entire group (there is always
      something that will go wrong!)

      One important thing on this new version: I am going to an interleaved
      transmission scheme (where the symbols are sent in a more randomized order
      and then deinterleaved at the receiver). My testing of this over many
      different channel types suggests it offers some significant improvement
      especially on burst noise, multipath and static crashes over the non
      interleaved version that was used before. The only downside to interleaving
      is that it is NOT compatible with the current posted version so it will
      require an update. The interleaved version is the same speed, same 200 Hz
      bandwidth and sounds exactly the same as before .only the order of symbol
      transmission is changed. The new version 0.2.0.0 will include Monitor, FEC
      and ARQ modes

      If you are game for this and want to try and set up some test FEC or ARQ
      session I would appreciate it.

      If interested contact me directly at rmuethingATcfl.rr.com and I will send
      you a zip package later this week. Thanks.

      73,

      Rick Muething, KN6KB


    • Rick Muething
      Daniel, What you described is truly Time division multiplexing but using redundant interleaved data streams. That is not what is being used in V4. Here
      Message 2 of 2 , May 24, 2011
      • 0 Attachment
        Daniel,
         
        What you described is truly Time division multiplexing but using redundant interleaved data streams.  That is not what is being used in V4.  Here interleave is being used in the common code term where the sequence of code symbols is rearranged (interleaved) to achieve a specific purpose.
         
        The reason for interleaving coded symbols has to do with the type of Error correcting code used and the expected type of channel errors. In V4 a fairly powerful Viterbi code of rate 1/2 K=7 code is used (sometimes called the Viking code after the Viking spacecraft that used it).  The Rate 1/2 means that the code contains 1 FEC (error correcting) bit per each bit of information (in other words the code expands the data by a factor of 2).  The length 7 means that the convolutional code used spreads each symbol out over 7 symbol times in an overlapping sequence).
         
        Now the Viterbi decoder is very powerful on random errors but clusters of errors (2 or more errors in a row) significantly reduce the strength of the FEC code and cause more symbol decode errors.  One way to improve on this is to take the Viterbi encoded symbols (in V4 there are 126 such 2 bit symbols per data frame) and instead of sending them in the normal order (as they come out of the Viterbi encoder) scramble them in some different sequence.  In this version I am using what is called a simple block interleaver to do this.  So instead of sending the Viterbi encoded symbols in normal sequence 1,2,3, ....126.  They are sent “interleaved” like 1,10,19,28,...2,11,20,29....126.  Now the receiving station (knowing the interleave mechanism used) simply re orders the received symbols back to their original Viterbi order BEFORE trying to decode them using the Viterbi decoder. You can see if there were a burst error during transmission say from static where 5 sequential symbols were wiped out this would now be spread in time to the same 5 errors but spread out in time (non adjacent).  The Viterbi decoder handles these more randomly distributed errors much better.  The net result is a decrease in frame error rate for burst, static and short term multipath errors. In simulations over thousands of frames I saw an improvement in frame success from about 50% (no interleave) to near 80% (interleaved) over poor channels (S/N = –5dB, CCIR Multipath poor channel).  That is equivalent to about doubling your transmission power. For random white noise Interleaving is no benefit (but doesn’t harm) but most HF propagation is much worse than white Gaussian noise channels.
         
        The only downside to interleaving is that you must capture and store all the interleaved symbols before decoding. For V4 this is not a concern since there are only 126 symbols per frame and the frame is not decoded until the entire frame is captured.
         
        Hope this helps answer your question. Here is a Wikipedia reference  http://en.wikipedia.org/wiki/Interleaving
         
        Rick Muething, KN6KB
         
        Sent: Tuesday, May 24, 2011 7:45 AM
        Subject: [V4Protocol] V4 Interleave
         
         

        Hello Rick.

        I have a question about the interleave method you are using?  I think I understand the term “interleave” but I’d like to be sure, as you’ve implemented it.

        In the RTTY days of old there used to be something frequently used in commercial and military called a “Time-Division Multiplexer” or also the “Time Diversity Modem” (TDM).  The one we TDM is a mode resembling FEC AMTOR. TDM uses one sub-carrier, but assigns separate data channels to different time slots. The MD-1142 (aka TDM BR 6029C) has two modes of operation both are full duplex (the receiver and transmitter sections are fully independent), in the Internal Multiplexer Mode a single TTY signal is fed to 7 different channels and the signal in each channel is time delayed by 1 second form all the other channels.  In addition there is a pilot tone at 561.1 Hz and the MARK/SPACE FSK shift of each tone was 85 Hz (if I recall).  This mode of operation provides a high probability of getting the message through.  In the External Multiplexer Mode each of the 8 channels is fed or receives a separate data stream so it does not have the redundancy of the IMM. In time order the channels are:

        Chan #    Center Freq      Delay sec.

           3             1530 Hz             0

           7             2890 Hz             1

           1               850 Hz             2

           4             1870 Hz             3

           6             2550 Hz             4

           2             1190 Hz             5

           5             2210 Hz             7

           8               561.1 Hz       pilot tone

        Each of the channels transmitted the same information but each delayed by 1 second.  The receiving TDM would receive all channels but vote on the best five matching channels to give an output.  The redundant transmitted data on different tone pairs with the delay insured to get the data through on noisy channels. The result was an error free transmission nearly every time.  The TDM worked very well for RTTY and I always thought they’d be good for implementation for the new PSK modes used in ham radio but never saw them. There is an initial 8 second latency from the start of transmission but that was never a problem for a long broadcast. Do you remember the TDM?  They were pretty slick.

        Anyway, does your interleave work in a similar manner, assigning the same data to different channels with a delay between channels?  I’d like to understand better how you’ve implemented the interleave process in V4.

        Also, could you send me the V4 ARQ version so I can test it in Europe?  Thanks.


        73,

        Daniel V. Wolff, Jr.
        DL4VCW / KA7AGN / AEM1WF (MARS)
        http://www.qsl.net/dl4vcw/

        Classification:  UNCLASSIFIED
        Caveats: NONE

        A few brave soles need to test the new V4 ARQ version

        Posted by: "Rick Muething" rmuething@...   kn6kb

        Mon May 23, 2011 1:32 pm (PDT)



        All,

        I am in the final stages of testing the new V4 ARQ mode. This is a nice
        improvement allowing two stations to exchange error free data (after FEC and
        ARQ) even in pretty poor conditions. I am in the final stages of testing
        and updating the help files for V4 and V4Chat. ARQ is a few percent slower
        than FEC (due to the ACK/NAK) but gives perfect copy (or no copy at all if
        things are really bad!).

        I would like a few (no more than 5) brave individuals with some experience
        with the V4 FEC mode to test out the install, help and test ARQ and FEC
        operation before posting the files for the entire group (there is always
        something that will go wrong!)

        One important thing on this new version: I am going to an interleaved
        transmission scheme (where the symbols are sent in a more randomized order
        and then deinterleaved at the receiver). My testing of this over many
        different channel types suggests it offers some significant improvement
        especially on burst noise, multipath and static crashes over the non
        interleaved version that was used before. The only downside to interleaving
        is that it is NOT compatible with the current posted version so it will
        require an update. The interleaved version is the same speed, same 200 Hz
        bandwidth and sounds exactly the same as before .only the order of symbol
        transmission is changed. The new version 0.2.0.0 will include Monitor, FEC
        and ARQ modes

        If you are game for this and want to try and set up some test FEC or ARQ
        session I would appreciate it.

        If interested contact me directly at rmuethingATcfl.rr.com and I will send
        you a zip package later this week. Thanks.

        73,

        Rick Muething, KN6KB


      Your message has been successfully submitted and would be delivered to recipients shortly.