Struct node_sync_t¶

What one alignment hypothesis scored, and what the runner-up did. More...

#include <viterbi_core.h>

Public Attributes¶

Type	Name
size_t	errors
size_t	margin
size_t	next
unsigned	phase
size_t	symbols

Detailed Description¶

errors against symbols IS the channel symbol error rate when the hypothesis is right, because in sync the decoder corrects the channel and the re-encoded stream differs from the received one exactly where the channel put an error.

Out of sync the decoder is searching a trellis its input does not lie on. The count then runs at a large fraction of the symbols — but not at a half, and the difference is worth stating because a half is what a coin-flip argument predicts and it is wrong: the decoder is a maximum LIKELIHOOD search, so it finds the codeword that agrees with the misaligned stream as well as any codeword can. Measured on a clean stream: 24 % of symbols for CCSDS K=7 r=1/2, 23 % for the same code uninverted, 18 % for a K=5 r=1/3 — against 0 % for the right alignment, which is the separation the decision actually rests on.

margin is what a caller acts on. The absolute count moves with Es/N0 and says nothing on its own; the DIFFERENCE between the best and the next best is the evidence that the search decided.

Public Attributes Documentation¶

variable errors¶

size_t node_sync_t::errors;

its disagreements

variable margin¶

size_t node_sync_t::margin;

next - errors; 0 when nothing separated

variable next¶

size_t node_sync_t::next;

the best competing hypothesis's

variable phase¶

unsigned node_sync_t::phase;

winning offset, 0 .. c->n-1

variable symbols¶

size_t node_sync_t::symbols;

symbols SCORED per hypothesis, which is fewer than the window — see node_sync_scored_symbols

The documentation for this class was generated from the following file native/inc/viterbi/viterbi_core.h