WSJT-X/doc/user_guide/en/decoder_notes.adoc

=== AP Decoding

With the QRA64 decoder Nico Palermo, IV3NWV, introduced a technique
for decoding with the aid of information that naturally accumulates
during a minimal QSO.  This _a priori_ (AP) information can be
used to increase the sensitivity of the decoder.

When an operator decides to answer a CQ, he already knows his own
callsign and that of his potential QSO partner.  He therefore knows
what to expect for at least 56 of the 72 message bits in a
standard-format response to his call.  The _WSJT-X_ decoders for QRA64
and FT8 can use these and similar AP bits to decode messages
containing them with higher sensitivity than otherwise possible.

We have implemented AP decoding in slightly different ways in QRA64
and FT8.  To provide some explicit examples for users, we provide here
a brief description of the FT8 behavior.

AP decoding attempts effectively set the AP bits to the hypothesized
values, as if they had been received correctly.  The decoder then
proceeds to determine whether the remaining message and parity bits
are consistent with the hypothesized AP bits.  If a codeword is found
that the decoder judges to have high (but not overwhelmingly high)
probability of being correct, a ? character is appended when the
decoded message is displayed.  To avoid misleading spots of occasional 
false decodes, messages so marked are not forwarded to {pskreporter}.

Successful AP decodes are always labeled with an end-of-line indicator
of the form aP, where P is one of the single-digit AP decoding types
listed in Table 1.  For example, an `a2` designator says that the
successful decode used MyCall as hypothetically known information.

[[AP_INFO_TABLE]]
.AP information types
[width="35%",cols="h10,<m20",frame=topbot,options="header"]
|===============================================
|P | Message components
|1 | CQ   &#160; &#160;   ?   &#160; &#160;   ? 
|2 | MyCall &#160; &#160; ?   &#160; &#160;   ? 
|3 | MyCall DxCall &#160; &#160;  ? 
|4 | MyCall DxCall RRR
|5 | MyCall DxCall 73
|6 | MyCall DxCall RR73
|===============================================

Table 2 lists the six possible QSO states that are tracked by the
WSJT-X auto-sequencer, along with the type of AP decoding that would
be attempted in each state.

[[AP_DECODING_TYPES_TABLE]]
.AP decoding types for each QSO state
[width="35%",cols="h10,<m20",frame=topbot,options="header"]
|===========================================
|State        |AP type
|CALLING      |   1, 2
|REPLYING     |   2, 3
|REPORT       |   2, 3
|ROGER_REPORT |   3, 4, 5, 6
|ROGERS       |   3, 4, 5, 6
|SIGNOFF      |   3, 1, 2
|===========================================


=== Decoded Lines

Displayed information accompanying decoded messages generally includes UTC,
signal-to-noise ratio in dB, time offset DT in seconds, and
audio frequency in Hz.  Some modes include additional information such
as frequency offset from nominal (DF), frequency drift (Drift or F1),
or distance (km or mi).

There may also be some cryptic characters with special meanings
summarized in the following Table:

[[DECODED_LINES_TABLE]]
.Notations used on decoded text lines
[width="50%",cols="h,3*^",frame=topbot,options="header"]
|===========================================
|Mode    |Mode character|Sync character|End of line information
|FT8     | ~            |      | ? &#160; aP
|JT4     | $            | *, # | f, fN, dNC
|JT9     | @            |      |
|JT65    | #            |      |
|JT65 VHF| #            | *, # | f, fN, dNC
|QRA64   | :            | *    | R
|ISCAT   |              | *    | M  N  C  T
|MSK144  | &            |      | N  H  E
|===========================================
Sync character::
 `*` - Normal sync +
 `#` - Alternate sync

End of line information::
 `?` - Decoded with lower confidence +
 `a` - Decoded with aid of some a priori (AP) information +
 `C` - Confidence indicator [ISCAT and Deep Search; (0-9,*)] +
 `d` - Deep Search algorithm +
 `E` - Size of MSK eye diagram opening - if negative, the eye is closed + 
 `f` - Franke-Taylor or Fano algorithm +
 `H` - Number of bit errors corrected +
 `M` - Message length (characters) +
 `N` - Number of Rx intervals or frames averaged +
 `P` - Number indicating type of AP information (Table 1, above) +
 `R` - Return code from QRA64 decoder +
 `T` - Length of analyzed region (s)