Below are extracts from what was written to stdout by the demodulator and decoder program for three different recorded wav files. Not show below is the fact that a Redundancy Level of 2 (20% of the outer code symbols being redundant) was used in all three of these cases.
I redirect stdout to a named file so that I can look at it whenever I want, without having to demodulate and decode again.
An explanation of all of the fields of data shown below is part of the documentation that comes with the program, and so will not be repeated here.
The first case, shown in the top third in the image just below, is one in which 2 of the 8 blocks were not decoded successfully. The last column in what is shown below has a "+" character if the block was decoded successfully and something other than a "+" character if the corresponding block was not decoded successfully.
The other two cases shown below, in the middle and bottom thirds of the image, show results from successfully decoding all 8 blocks of their respective recorded wav files.
Look at the column titled "outer code changes". For the case shown in the middle third of the image below, the number of "outer code changes" in the first 4 blocks, and the last block, is comfortably below half of the nominal limit of 32 per block for the Redundancy Level 2 case. However, this is not the case for the other 3 blocks, thus Redundancy Level 2 was an appropriate choice for this transmission.
If conditions are stable, and you consistently use less than half of the nominal limit of the outer code's correcting ability, then you probably want to switch to an outer code with half as much redundancy.
The program can generally correct more errors than the nominal limit of the outer decoder. An example of this is shown in the bottom third of the image just below, in which 3 of the 8 blocks had more than the nominal limit of 32 errors corrected by the outer decoder.
Correction of a program error has resulted in successfully decoding one of the recordings by W8ZCF shown below as "Trailer Not Found". Thus, using the corrected program, 2 of VK3CQE's transmissions of 09 April 2001 recorded by W8ZCF successfully decoded, one had 6 good blocks out of 8, and the other 2 recordings didn't have a recognizable trailer.
Below are results from three different receptions of the same transmission by VK3LM. The recordings by: KB4YZ, W8ZCF, and W9NTP, of this transmission were all successfully decoded (all errors corrected).
The horizontal axis of this plot is the sequential index number of the code blocks, which takes integer values from 0 through 7, in this case. Each code block lasts about 2.5 seconds. The vertical axis of this plot is expressed in terms of the percentage of the outer code symbols (306 symbols = 100%).
A bar graph is plotted showing the percentage of outer code symbols corrected by decoding each of the three recordings of this single transmission.
All three receivers were relatively close to each other, within a 130 mile diameter circle in the United States, compared to their distance to the transmitter, which was in Australia. The plot below shows significant variation in the percentage of outer code symbols that needed to be corrected from block to block. Also, there is significant variation in the percentage of outer code symbols that needed to be corrected from each location for a given block.
Significant variations in number of errors that the outer decoder needs to correct occur with both time and receiver location.
Also note that several instances of the outer decoder successfully correcting more than the nominal number of errors it is expected to be able to correct are shown below (red, green and blue lines extending above the magenta line).
Links to: