100 lines
3.6 KiB
Plaintext
100 lines
3.6 KiB
Plaintext
|
<HTML><HEAD>
|
||
|
|
|
||
|
|
<TITLE> Detailed Decoding Trace Information </TITLE>
|
||
|
|
|
||
|
|
|
||
|
|
</HEAD><BODY>
|
||
|
|
|
||
|
|
<H1> Detailed Decoding Trace Information </H1>
|
||
|
|
|
||
|
|
The <B>-T</B> option to <A HREF="decoding.html#decode"><TT>decode</TT></A>
|
||
|
|
causes a detailed trace of information on the process of decoding to
|
||
|
|
be written to standard output, in a multi-column format, with the first
|
||
|
|
line containing the headers for each column. This format is suitable
|
||
|
|
for reading into S-Plus or R.
|
||
|
|
|
||
|
|
The first column is always the number of the block being decoded;
|
||
|
|
several lines may be output for each block. The other columns vary
|
||
|
|
with the decoding method used, as described below.
|
||
|
|
|
||
|
|
|
||
|
|
<H2>Enum-bit and Enum-block decoding methods</H2>
|
||
|
|
|
||
|
|
For source messages with <I>K</I> bits, 2<SUP><I>K</I></SUP> lines are output
|
||
|
|
for each block, containing the following information:
|
||
|
|
<BLOCKQUOTE>
|
||
|
|
<TABLE>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>block</B> </td>
|
||
|
|
<td>The number of the block, from zero</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>decoding</B> </td>
|
||
|
|
<td>A possible decoding for the message bits, expressed as a hexadecimal
|
||
|
|
number. The other bits are determined by the message bits.</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>likelihood</B> </td>
|
||
|
|
<td>The likelihood for this decoding (ie, the probability of obtaining
|
||
|
|
the data received if this was the message sent).</td></tr>
|
||
|
|
</TABLE>
|
||
|
|
</BLOCKQUOTE>
|
||
|
|
|
||
|
|
For these methods, the number of "iterations" (output with the
|
||
|
|
<B>-t</B> option) is always 2<SUP><I>K</I></SUP>.
|
||
|
|
|
||
|
|
|
||
|
|
<H2>Prprp decoding method</H2>
|
||
|
|
|
||
|
|
Each block results in one line of output for the initial state (based
|
||
|
|
on individual likelihood ratios), and one line for each subsequent
|
||
|
|
iteration, containing the following information:
|
||
|
|
<BLOCKQUOTE>
|
||
|
|
<TABLE>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>block</B> </td>
|
||
|
|
<td>The number of the block, from zero</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>iter</B> </td>
|
||
|
|
<td>The number of an iteration, zero for the initial state.</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>changed</B> </td>
|
||
|
|
<td>The number of bits in the decoding that differ from the bit that would
|
||
|
|
be chosen based just on the likelihood ratio for that bit. Bits whose
|
||
|
|
likelihood ratios are exactly one contribute 0.5 to this
|
||
|
|
count. (Likelihood ratios of exactly one can arise when the output
|
||
|
|
of an AWGN channel rounds to exactly 0.00.)</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>perrs</B> </td>
|
||
|
|
<td>The number of parity check errors in the current tentative
|
||
|
|
decoding.</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>loglik</B> </td>
|
||
|
|
<td>The log likelihood of the current tentative decoding.</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>Eperrs</B> </td>
|
||
|
|
<td>The expected number of parity check errors in a decoding found
|
||
|
|
by randomly picking a value for each bit, independently, according
|
||
|
|
to the current bit probabilities.</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>Eloglik</B> </td>
|
||
|
|
<td>The expected log likelihood of a decoding found by randomly picking
|
||
|
|
a value for each bit, independently, according to the current bit
|
||
|
|
probabilities.
|
||
|
|
</td></tr>
|
||
|
|
<tr align="left" valign="top">
|
||
|
|
<td> <B>entropy</B> </td>
|
||
|
|
<td>The entropy (in bits) of the distribution defined by the current bit
|
||
|
|
probablities, assumed to apply to bits independently.</td></tr>
|
||
|
|
</TABLE>
|
||
|
|
</BLOCKQUOTE>
|
||
|
|
|
||
|
|
The number of "iterations" (output with the <B>-t</B> option) is
|
||
|
|
the obvious count of probability propagation iterations. The
|
||
|
|
initial state does not count as an iteration.
|
||
|
|
|
||
|
|
<HR>
|
||
|
|
|
||
|
|
<A HREF="decoding.html">Back to decoding documentation</A><BR>
|
||
|
|
<A HREF="index.html">Back to index for LDPC software</A>
|
||
|
|
|
||
|
|
</BODY></HTML>
|