spandsp
0.0.6
Main Page
Related Pages
Classes
Files
File List
File Members
v17rx.h
Go to the documentation of this file.
1
/*
2
* SpanDSP - a series of DSP components for telephony
3
*
4
* v17rx.h - ITU V.17 modem receive part
5
*
6
* Written by Steve Underwood <steveu@coppice.org>
7
*
8
* Copyright (C) 2003 Steve Underwood
9
*
10
* All rights reserved.
11
*
12
* This program is free software; you can redistribute it and/or modify
13
* it under the terms of the GNU Lesser General Public License version 2.1,
14
* as published by the Free Software Foundation.
15
*
16
* This program is distributed in the hope that it will be useful,
17
* but WITHOUT ANY WARRANTY; without even the implied warranty of
18
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19
* GNU Lesser General Public License for more details.
20
*
21
* You should have received a copy of the GNU Lesser General Public
22
* License along with this program; if not, write to the Free Software
23
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
24
*/
25
26
/*! \file */
27
28
#if !defined(_SPANDSP_V17RX_H_)
29
#define _SPANDSP_V17RX_H_
30
31
/*! \page v17rx_page The V.17 receiver
32
\section v17rx_page_sec_1 What does it do?
33
The V.17 receiver implements the receive side of a V.17 modem. This can operate
34
at data rates of 14400, 12000, 9600 and 7200 bits/second. The audio input is a stream
35
of 16 bit samples, at 8000 samples/second. The transmit and receive side of V.17
36
modems operate independantly. V.17 is mostly used for FAX transmission over PSTN
37
lines, where it provides the standard 14400 bits/second rate.
38
39
\section v17rx_page_sec_2 How does it work?
40
V.17 uses QAM modulation, at 2400 baud, and trellis coding. Constellations with
41
16, 32, 64, and 128 points are defined. After one bit per baud is absorbed by the
42
trellis coding, this gives usable bit rates of 7200, 9600, 12000, and 14400 per
43
second.
44
45
V.17 specifies a training sequence at the start of transmission, which makes the
46
design of a V.17 receiver relatively straightforward. The first stage of the
47
training sequence consists of 256
48
symbols, alternating between two constellation positions. The receiver monitors
49
the signal power, to sense the possible presence of a valid carrier. When the
50
alternating signal begins, the power rising above a minimum threshold (-43dBm0)
51
causes the main receiver computation to begin. The initial measured power is
52
used to quickly set the gain of the receiver. After this initial settling, the
53
front end gain is locked, and the adaptive equalizer tracks any subsequent
54
signal level variation. The signal is oversampled to 24000 samples/second (i.e.
55
signal, zero, zero, signal, zero, zero, ...) and fed to a complex root raised
56
cosine pulse shaping filter. This filter has been modified from the conventional
57
root raised cosine filter, by shifting it up the band, to be centred at the nominal
58
carrier frequency. This filter interpolates the samples, pulse shapes, and performs
59
a fractional sample delay at the same time. 192 sets of filter coefficients are used
60
to achieve a set of finely spaces fractional sample delays, between zero and
61
one sample. By choosing every fifth sample, and the appropriate set of filter
62
coefficients, the properly tuned symbol tracker can select data samples at 4800
63
samples/second from points within 0.28 degrees of the centre and mid-points of
64
each symbol. The output of the filter is multiplied by a complex carrier, generated
65
by a DDS. The result is a baseband signal, requiring no further filtering, apart from
66
an adaptive equalizer. The baseband signal is fed to a T/2 adaptive equalizer.
67
A band edge component maximisation algorithm is used to tune the sampling, so the samples
68
fed to the equalizer are close to the mid point and edges of each symbol. Initially
69
the algorithm is very lightly damped, to ensure the symbol alignment pulls in
70
quickly. Because the sampling rate will not be precisely the same as the
71
transmitter's (the spec. says the symbol timing should be within 0.01%), the
72
receiver constantly evaluates and corrects this sampling throughout its
73
operation. During the symbol timing maintainence phase, the algorithm uses
74
a heavier damping.
75
76
The carrier is specified as 1800Hz +- 1Hz at the transmitter, and 1800 +-7Hz at
77
the receiver. The receive carrier would only be this inaccurate if the link
78
includes FDM sections. These are being phased out, but the design must still
79
allow for the worst case. Using an initial 1800Hz signal for demodulation gives
80
a worst case rotation rate for the constellation of about one degree per symbol.
81
Once the symbol timing synchronisation algorithm has been given time to lock to the
82
symbol timing of the initial alternating pattern, the phase of the demodulated signal
83
is recorded on two successive symbols - once for each of the constellation positions.
84
The receiver then tracks the symbol alternations, until a large phase jump occurs.
85
This signifies the start of the next phase of the training sequence. At this
86
point the total phase shift between the original recorded symbol phase, and the
87
symbol phase just before the phase jump occurred is used to provide a coarse
88
estimation of the rotation rate of the constellation, and it current absolute
89
angle of rotation. These are used to update the current carrier phase and phase
90
update rate in the carrier DDS. The working data already in the pulse shaping
91
filter and equalizer buffers is given a similar step rotation to pull it all
92
into line. From this point on, a heavily damped integrate and dump approach,
93
based on the angular difference between each received constellation position and
94
its expected position, is sufficient to track the carrier, and maintain phase
95
alignment. A fast rough approximator for the arc-tangent function is adequate
96
for the estimation of the angular error.
97
98
The next phase of the training sequence is a scrambled sequence of two
99
particular symbols. We train the T/2 adaptive equalizer using this sequence. The
100
scrambling makes the signal sufficiently diverse to ensure the equalizer
101
converges to the proper generalised solution. At the end of this sequence, the
102
equalizer should be sufficiently well adapted that is can correctly resolve the
103
full QAM constellation. However, the equalizer continues to adapt throughout
104
operation of the modem, fine tuning on the more complex data patterns of the
105
full QAM constellation.
106
107
In the last phase of the training sequence, the modem enters normal data
108
operation, with a short defined period of all ones as data. As in most high
109
speed modems, data in a V.17 modem passes through a scrambler, to whiten the
110
spectrum of the signal. The transmitter should initialise its data scrambler,
111
and pass the ones through it. At the end of the ones, real data begins to pass
112
through the scrambler, and the transmit modem is in normal operation. The
113
receiver tests that ones are really received, in order to verify the modem
114
trained correctly. If all is well, the data following the ones is fed to the
115
application, and the receive modem is up and running. Unfortunately, some
116
transmit side of some real V.17 modems fail to initialise their scrambler before
117
sending the ones. This means the first 23 received bits (the length of the
118
scrambler register) cannot be trusted for the test. The receive modem,
119
therefore, only tests that bits starting at bit 24 are really ones.
120
121
The V.17 signal is trellis coded. Two bits of each symbol are convolutionally coded
122
to form a 3 bit trellis code - the two original bits, plus an extra redundant bit. It
123
is possible to ignore the trellis coding, and just decode the non-redundant bits.
124
However, the noise performance of the receiver would suffer. Using a proper
125
trellis decoder adds several dB to the noise tolerance to the receiving modem. Trellis
126
coding seems quite complex at first sight, but is fairly straightforward once you
127
get to grips with it.
128
129
Trellis decoding tracks the data in terms of the possible states of the convolutional
130
coder at the transmitter. There are 8 possible states of the V.17 coder. The first
131
step in trellis decoding is to find the best candidate constellation point
132
for each of these 8 states. One of thse will be our final answer. The constellation
133
has been designed so groups of 8 are spread fairly evenly across it. Locating them
134
is achieved is a reasonably fast manner, by looking up the answers in a set of space
135
map tables. The disadvantage is the tables are potentially large enough to affect
136
cache performance. The trellis decoder works over 16 successive symbols. The result
137
of decoding is not known until 16 symbols after the data enters the decoder. The
138
minimum total accumulated mismatch between each received point and the actual
139
constellation (termed the distance) is assessed for each of the 8 states. A little
140
analysis of the coder shows that each of the 8 current states could be arrived at
141
from 4 different previous states, through 4 different constellation bit patterns.
142
For each new state, the running total distance is arrived at by inspecting a previous
143
total plus a new distance for the appropriate 4 previous states. The minimum of the 4
144
values becomes the new distance for the state. Clearly, a mechanism is needed to stop
145
this distance from growing indefinitely. A sliding window, and several other schemes
146
are possible. However, a simple single pole IIR is very simple, and provides adequate
147
results.
148
149
For each new state we store the constellation bit pattern, or path, to that state, and
150
the number of the previous state. We find the minimum distance amongst the 8 new
151
states for each new symbol. We then trace back through the states, until we reach the
152
one 16 states ago which leads to the current minimum distance. The bit pattern stored
153
there is the error corrected bit pattern for that symbol.
154
155
So, what does Trellis coding actually achieve? TCM is easier to understand by looking
156
at the V.23bis modem spec. The V.32bis spec. is very similar to V.17, except that it
157
is a full duplex modem and has non-TCM options, as well as the TCM ones in V.17.
158
159
V32bis defines two options for pumping 9600 bits per second down a phone line - one
160
with and one without TCM. Both run at 2400 baud. The non-TCM one uses simple 16 point
161
QAM on the raw data. The other takes two out of every four raw bits, and convolutionally
162
encodes them to 3. Now we have 5 bits per symbol, and we need 32 point QAM to send the
163
data.
164
165
The raw error rate from simple decoding of the 32 point QAM is horrible compared to
166
decoding the 16 point QAM. If a point decoded from the 32 point QAM is wrong, the likely
167
correct choice should be one of the adjacent ones. It is unlikely to have been one that
168
is far away across the constellation, unless there was a huge noise spike, interference,
169
or something equally nasty. Now, the 32 point symbols do not exist in isolation. There
170
was a kind of temporal smearing in the convolutional coding. It created a well defined
171
dependency between successive symbols. If we knew for sure what the last few symbols
172
were, they would lead us to a limited group of possible values for the current symbol,
173
constrained by the behaviour of the convolutional coder. If you look at how the symbols
174
were mapped to constellation points, you will see the mapping tries to spread those
175
possible symbols as far apart as possible. This will leave only one that is pretty
176
close to the received point, which must be the correct choice. However, this assumes
177
we know the last few symbols for sure. Since we don't, we have a bit more work to do
178
to achieve reliable decoding.
179
180
Instead of decoding to the nearest point on the constellation, we decode to a group of
181
likely constellation points in the neighbourhood of the received point. We record the
182
mismatch for each - that is the distance across the constellation between the received
183
point and the group of nearby points. To avoid square roots, recording x2 + y2 can be
184
good enough. Symbol by symbol, we record this information. After a few symbols we can
185
stand back and look at the recorded information.
186
187
For each symbol we have a set of possible symbol values and error metric pairs. The
188
dependency between symbols, created by the convolutional coder, means some paths from
189
symbol to symbol are possible and some are not. It we trace back through the possible
190
symbol to symbol paths, and total up the error metric through those paths, we end up
191
with a set of figures of merit (or more accurately figures of demerit, since
192
larger == worse) for the likelihood of each path being the correct one. The path with
193
the lowest total metric is the most likely, and gives us our final choice for what we
194
think the current symbol really is.
195
196
That was hard work. It takes considerable computation to do this selection and traceback,
197
symbol by symbol. We need to get quite a lot from this. It needs to drive the error rate
198
down so far that is compensates for the much higher error rate due to the larger
199
constellation, and then buys us some actual benefit. Well in the example we are looking
200
at - V.32bis at 9600bps - it works out the error rate from the TCM option is like using
201
the non-TCM option with several dB more signal to noise ratio. That's nice. The non-TCM
202
option is pretty reasonable on most phone lines, but a better error rate is always a
203
good thing. However, V32bis includes a 14,400bps option. That uses 2400 baud, and 6 bit
204
symbols. Convolutional encoding increases that to 7 bits per symbol, by taking 2 bits and
205
encoding them to 3. This give a 128 point QAM constellation. Again, the difference between
206
using this, and using just an uncoded 64 point constellation is equivalent to maybe 5dB of
207
extra signal to noise ratio. However, in this case it is the difference between the modem
208
working only on the most optimal lines, and being widely usable across most phone lines.
209
TCM absolutely transformed the phone line modem business.
210
*/
211
212
/*!
213
V.17 modem receive side descriptor. This defines the working state for a
214
single instance of a V.17 modem receiver.
215
*/
216
typedef
struct
v17_rx_state_s
v17_rx_state_t
;
217
218
#if defined(__cplusplus)
219
extern
"C"
220
{
221
#endif
222
223
/*! Initialise a V.17 modem receive context.
224
\brief Initialise a V.17 modem receive context.
225
\param s The modem context.
226
\param bit_rate The bit rate of the modem. Valid values are 7200, 9600, 12000 and 14400.
227
\param put_bit The callback routine used to put the received data.
228
\param user_data An opaque pointer passed to the put_bit routine.
229
\return A pointer to the modem context, or NULL if there was a problem. */
230
SPAN_DECLARE(
v17_rx_state_t
*)
v17_rx_init
(
v17_rx_state_t
*s,
int
bit_rate
,
put_bit_func_t
put_bit
,
void
*user_data);
231
232
/*! Reinitialise an existing V.17 modem receive context.
233
\brief Reinitialise an existing V.17 modem receive context.
234
\param s The modem context.
235
\param bit_rate The bit rate of the modem. Valid values are 7200, 9600, 12000 and 14400.
236
\param short_train TRUE if a short training sequence is expected.
237
\return 0 for OK, -1 for bad parameter */
238
SPAN_DECLARE(
int
)
v17_rx_restart
(
v17_rx_state_t
*s,
int
bit_rate,
int
short_train
);
239
240
/*! Release a V.17 modem receive context.
241
\brief Release a V.17 modem receive context.
242
\param s The modem context.
243
\return 0 for OK */
244
SPAN_DECLARE(
int
)
v17_rx_release
(
v17_rx_state_t
*s);
245
246
/*! Free a V.17 modem receive context.
247
\brief Free a V.17 modem receive context.
248
\param s The modem context.
249
\return 0 for OK */
250
SPAN_DECLARE(
int
)
v17_rx_free
(
v17_rx_state_t
*s);
251
252
/*! Get the logging context associated with a V.17 modem receive context.
253
\brief Get the logging context associated with a V.17 modem receive context.
254
\param s The modem context.
255
\return A pointer to the logging context */
256
SPAN_DECLARE(
logging_state_t
*)
v17_rx_get_logging_state
(
v17_rx_state_t
*s);
257
258
/*! Change the put_bit function associated with a V.17 modem receive context.
259
\brief Change the put_bit function associated with a V.17 modem receive context.
260
\param s The modem context.
261
\param put_bit The callback routine used to handle received bits.
262
\param user_data An opaque pointer. */
263
SPAN_DECLARE(
void
)
v17_rx_set_put_bit
(
v17_rx_state_t
*s,
put_bit_func_t
put_bit,
void
*user_data);
264
265
/*! Change the modem status report function associated with a V.17 modem receive context.
266
\brief Change the modem status report function associated with a V.17 modem receive context.
267
\param s The modem context.
268
\param handler The callback routine used to report modem status changes.
269
\param user_data An opaque pointer. */
270
SPAN_DECLARE(
void
)
v17_rx_set_modem_status_handler
(
v17_rx_state_t
*s,
modem_rx_status_func_t
handler,
void
*user_data);
271
272
/*! Process a block of received V.17 modem audio samples.
273
\brief Process a block of received V.17 modem audio samples.
274
\param s The modem context.
275
\param amp The audio sample buffer.
276
\param len The number of samples in the buffer.
277
\return The number of samples unprocessed.
278
*/
279
SPAN_DECLARE_NONSTD
(
int
) v17_rx(
v17_rx_state_t
*s, const int16_t amp[],
int
len);
280
281
/*! Fake processing of a missing block of received V.17 modem audio samples.
282
(e.g due to packet loss).
283
\brief Fake processing of a missing block of received V.17 modem audio samples.
284
\param s The modem context.
285
\param len The number of samples to fake.
286
\return The number of samples unprocessed.
287
*/
288
SPAN_DECLARE_NONSTD
(
int
) v17_rx_fillin(
v17_rx_state_t
*s,
int
len);
289
290
/*! Get a snapshot of the current equalizer coefficients.
291
\brief Get a snapshot of the current equalizer coefficients.
292
\param s The modem context.
293
\param coeffs The vector of complex coefficients.
294
\return The number of coefficients in the vector. */
295
#if defined(SPANDSP_USE_FIXED_POINTx)
296
SPAN_DECLARE(
int
)
v17_rx_equalizer_state
(
v17_rx_state_t
*s,
complexi_t
**coeffs);
297
#else
298
SPAN_DECLARE(
int
)
v17_rx_equalizer_state
(
v17_rx_state_t
*s,
complexf_t
**coeffs);
299
#endif
300
301
/*! Get the current received carrier frequency.
302
\param s The modem context.
303
\return The frequency, in Hertz. */
304
SPAN_DECLARE(
float
)
v17_rx_carrier_frequency
(
v17_rx_state_t
*s);
305
306
/*! Get the current symbol timing correction since startup.
307
\param s The modem context.
308
\return The correction. */
309
SPAN_DECLARE(
float
)
v17_rx_symbol_timing_correction
(
v17_rx_state_t
*s);
310
311
/*! Get a current received signal power.
312
\param s The modem context.
313
\return The signal power, in dBm0. */
314
SPAN_DECLARE(
float
)
v17_rx_signal_power
(
v17_rx_state_t
*s);
315
316
/*! Set the power level at which the carrier detection will cut in
317
\param s The modem context.
318
\param cutoff The signal cutoff power, in dBm0. */
319
SPAN_DECLARE(
void
)
v17_rx_signal_cutoff
(
v17_rx_state_t
*s,
float
cutoff);
320
321
/*! Set a handler routine to process QAM status reports
322
\param s The modem context.
323
\param handler The handler routine.
324
\param user_data An opaque pointer passed to the handler routine. */
325
SPAN_DECLARE(
void
)
v17_rx_set_qam_report_handler
(
v17_rx_state_t
*s, qam_report_handler_t handler,
void
*user_data);
326
327
#if defined(__cplusplus)
328
}
329
#endif
330
331
#endif
332
/*- End of file ------------------------------------------------------------*/
src
spandsp
v17rx.h
Generated on Wed Oct 2 2013 18:50:52 for spandsp by
1.8.1.2