t/gr-ieee802-11
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
Files
0f56254df31aa4a06e78d003ec4130e6b18d5105
gr-ieee802-11/lib/utils.cc
T
Gonzalo Arcos 0f56254df3 - Added new Viterbi convolutional decoder to the OFDM Decode Mac block replacing the former IT++ version. This new decoder uses 
SSE2 and improves the performance significantly. The decoder was taken from the gr-dvbt module from GNU Radio.
- Disabled some logging functions in the ofdm_mac and ofdm_decode_signal blocks that were degrading the performance significantly.

[Gonzalo Arcos | Rodrigo Ferreri]
2016-04-20 06:53:49 +02:00

1037 lines
30 KiB
C++
Raw Blame History

/*
* Copyright (C) 2013 Bastian Bloessl <bloessl@ccs-labs.org>
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "utils.h"
#include <cassert>
#include <cstring>
#include <math.h>
// constellations for different modulations (normalized to avg power 1)
const std::complex<double> BPSK_D[2] = {
std::complex<double>(-1.0, 0.0), std::complex<double>(1.0, 0.0)};
const std::complex<double> QPSK_D[4] = {
std::complex<double>(-0.7071, -0.7071), std::complex<double>(-0.7071, 0.7071),
std::complex<double>(+0.7071, -0.7071), std::complex<double>(+0.7071, 0.7071)};
const std::complex<double> QAM16_D[16] = {
std::complex<double>(-0.9487, -0.9487), std::complex<double>(-0.9487, -0.3162),
std::complex<double>(-0.9487, 0.9487), std::complex<double>(-0.9487, 0.3162),
std::complex<double>(-0.3162, -0.9487), std::complex<double>(-0.3162, -0.3162),
std::complex<double>(-0.3162, 0.9487), std::complex<double>(-0.3162, 0.3162),
std::complex<double>(0.9487, -0.9487), std::complex<double>(0.9487, -0.3162),
std::complex<double>(0.9487, 0.9487), std::complex<double>(0.9487, 0.3162),
std::complex<double>(0.3162, -0.9487), std::complex<double>(0.3162, -0.3162),
std::complex<double>(0.3162, 0.9487), std::complex<double>(0.3162, 0.3162)};
const std::complex<double> QAM64_D[64] = {
std::complex<double>(-1.0801, -1.0801), std::complex<double>(-1.0801, -0.7715),
std::complex<double>(-1.0801, -0.1543), std::complex<double>(-1.0801, -0.4629),
std::complex<double>(-1.0801, 1.0801), std::complex<double>(-1.0801, 0.7715),
std::complex<double>(-1.0801, 0.1543), std::complex<double>(-1.0801, 0.4629),
std::complex<double>(-0.7715, -1.0801), std::complex<double>(-0.7715, -0.7715),
std::complex<double>(-0.7715, -0.1543), std::complex<double>(-0.7715, -0.4629),
std::complex<double>(-0.7715, 1.0801), std::complex<double>(-0.7715, 0.7715),
std::complex<double>(-0.7715, 0.1543), std::complex<double>(-0.7715, 0.4629),
std::complex<double>(-0.1543, -1.0801), std::complex<double>(-0.1543, -0.7715),
std::complex<double>(-0.1543, -0.1543), std::complex<double>(-0.1543, -0.4629),
std::complex<double>(-0.1543, 1.0801), std::complex<double>(-0.1543, 0.7715),
std::complex<double>(-0.1543, 0.1543), std::complex<double>(-0.1543, 0.4629),
std::complex<double>(-0.4629, -1.0801), std::complex<double>(-0.4629, -0.7715),
std::complex<double>(-0.4629, -0.1543), std::complex<double>(-0.4629, -0.4629),
std::complex<double>(-0.4629, 1.0801), std::complex<double>(-0.4629, 0.7715),
std::complex<double>(-0.4629, 0.1543), std::complex<double>(-0.4629, 0.4629),
std::complex<double>(1.0801, -1.0801), std::complex<double>(1.0801, -0.7715),
std::complex<double>(1.0801, -0.1543), std::complex<double>(1.0801, -0.4629),
std::complex<double>(1.0801, 1.0801), std::complex<double>(1.0801, 0.7715),
std::complex<double>(1.0801, 0.1543), std::complex<double>(1.0801, 0.4629),
std::complex<double>(0.7715, -1.0801), std::complex<double>(0.7715, -0.7715),
std::complex<double>(0.7715, -0.1543), std::complex<double>(0.7715, -0.4629),
std::complex<double>(0.7715, 1.0801), std::complex<double>(0.7715, 0.7715),
std::complex<double>(0.7715, 0.1543), std::complex<double>(0.7715, 0.4629),
std::complex<double>(0.1543, -1.0801), std::complex<double>(0.1543, -0.7715),
std::complex<double>(0.1543, -0.1543), std::complex<double>(0.1543, -0.4629),
std::complex<double>(0.1543, 1.0801), std::complex<double>(0.1543, 0.7715),
std::complex<double>(0.1543, 0.1543), std::complex<double>(0.1543, 0.4629),
std::complex<double>(0.4629, -1.0801), std::complex<double>(0.4629, -0.7715),
std::complex<double>(0.4629, -0.1543), std::complex<double>(0.4629, -0.4629),
std::complex<double>(0.4629, 1.0801), std::complex<double>(0.4629, 0.7715),
std::complex<double>(0.4629, 0.1543), std::complex<double>(0.4629, 0.4629)};
ofdm_param::ofdm_param(Encoding e) {
encoding = e;
switch(e) {
case BPSK_1_2:
n_bpsc = 1;
n_cbps = 48;
n_dbps = 24;
rate_field = 0x0D; // 0b00001101
break;
case BPSK_3_4:
n_bpsc = 1;
n_cbps = 48;
n_dbps = 36;
rate_field = 0x0F; // 0b00001111
break;
case QPSK_1_2:
n_bpsc = 2;
n_cbps = 96;
n_dbps = 48;
rate_field = 0x05; // 0b00000101
break;
case QPSK_3_4:
n_bpsc = 2;
n_cbps = 96;
n_dbps = 72;
rate_field = 0x07; // 0b00000111
break;
case QAM16_1_2:
n_bpsc = 4;
n_cbps = 192;
n_dbps = 96;
rate_field = 0x09; // 0b00001001
break;
case QAM16_3_4:
n_bpsc = 4;
n_cbps = 192;
n_dbps = 144;
rate_field = 0x0B; // 0b00001011
break;
case QAM64_2_3:
n_bpsc = 6;
n_cbps = 288;
n_dbps = 192;
rate_field = 0x01; // 0b00000001
break;
case QAM64_3_4:
n_bpsc = 6;
n_cbps = 288;
n_dbps = 216;
rate_field = 0x03; // 0b00000011
break;
defaut:
assert(false);
break;
}
}
void
ofdm_param::print() {
std::cout << "OFDM Parameters:" << std::endl;
std::cout << "endcoding :" << encoding << std::endl;
std::cout << "rate_field :" << (int)rate_field << std::endl;
std::cout << "n_bpsc :" << n_bpsc << std::endl;
std::cout << "n_cbps :" << n_cbps << std::endl;
std::cout << "n_dbps :" << n_dbps << std::endl;
}
tx_param::tx_param(ofdm_param &ofdm, int psdu_length) {
psdu_size = psdu_length;
// number of symbols (17-11)
n_sym = (int) ceil((16 + 8 * psdu_size + 6) / (double) ofdm.n_dbps);
// number of bits of the data field (17-12)
n_data = n_sym * ofdm.n_dbps;
// number of padding bits (17-13)
n_pad = n_data - (16 + 8 * psdu_size + 6);
n_encoded_bits = n_sym * ofdm.n_cbps;
}
void
tx_param::print() {
std::cout << "TX Parameters:" << std::endl;
std::cout << "psdu_size: " << psdu_size << std::endl;
std::cout << "n_sym: " << n_sym << std::endl;
std::cout << "n_data: " << n_data << std::endl;
std::cout << "n_pad: " << n_pad << std::endl;
}
void scramble(const char *in, char *out, tx_param &tx, char initial_state) {
int state = initial_state;
int feedback;
for (int i = 0; i < tx.n_data; i++) {
feedback = (!!(state & 64)) ^ (!!(state & 8));
out[i] = feedback ^ in[i];
state = ((state << 1) & 0x7e) | feedback;
}
}
void reset_tail_bits(char *scrambled_data, tx_param &tx) {
memset(scrambled_data + tx.n_data - tx.n_pad - 6, 0, 6 * sizeof(char));
}
int ones(int n) {
int sum = 0;
for(int i = 0; i < 8; i++) {
if(n & (1 << i)) {
sum++;
}
}
return sum;
}
void convolutional_encoding(const char *in, char *out, tx_param &tx) {
int state = 0;
for(int i = 0; i < tx.n_data; i++) {
assert(in[i] == 0 || in[i] == 1);
state = ((state << 1) & 0x7e) | in[i];
out[i * 2] = ones(state & 0155) % 2;
out[i * 2 + 1] = ones(state & 0117) % 2;
}
}
void puncturing(const char *in, char *out, tx_param &tx, ofdm_param &ofdm) {
int mod;
for (int i = 0; i < tx.n_data * 2; i++) {
switch(ofdm.encoding) {
case BPSK_1_2:
case QPSK_1_2:
case QAM16_1_2:
*out = in[i];
out++;
break;
case QAM64_2_3:
if (i % 4 != 3) {
*out = in[i];
out++;
}
break;
case BPSK_3_4:
case QPSK_3_4:
case QAM16_3_4:
case QAM64_3_4:
mod = i % 6;
if (!(mod == 3 || mod == 4)) {
*out = in[i];
out++;
}
break;
defaut:
assert(false);
break;
}
}
}
void interleave(const char *in, char *out, tx_param &tx, ofdm_param &ofdm, bool reverse) {
int n_cbps = ofdm.n_cbps;
int first[n_cbps];
int second[n_cbps];
int s = std::max(ofdm.n_bpsc / 2, 1);
for(int j = 0; j < n_cbps; j++) {
first[j] = s * (j / s) + ((j + int(floor(16.0 * j / n_cbps))) % s);
}
for(int i = 0; i < n_cbps; i++) {
second[i] = 16 * i - (n_cbps - 1) * int(floor(16.0 * i / n_cbps));
}
for(int i = 0; i < tx.n_sym; i++) {
for(int k = 0; k < n_cbps; k++) {
if(reverse) {
out[i * n_cbps + second[first[k]]] = in[i * n_cbps + k];
} else {
out[i * n_cbps + k] = in[i * n_cbps + second[first[k]]];
}
}
}
}
void interleave(const double *in, double *out, tx_param &tx, ofdm_param &ofdm, bool reverse) {
int n_cbps = ofdm.n_cbps;
int first[n_cbps];
int second[n_cbps];
int s = std::max(ofdm.n_bpsc / 2, 1);
for(int j = 0; j < n_cbps; j++) {
first[j] = s * (j / s) + ((j + int(floor(16.0 * j / n_cbps))) % s);
}
for(int i = 0; i < n_cbps; i++) {
second[i] = 16 * i - (n_cbps - 1) * int(floor(16.0 * i / n_cbps));
}
for(int i = 0; i < tx.n_sym; i++) {
for(int k = 0; k < n_cbps; k++) {
if(reverse) {
out[i * n_cbps + second[first[k]]] = in[i * n_cbps + k];
} else {
out[i * n_cbps + k] = in[i * n_cbps + second[first[k]]];
}
}
}
}
void split_symbols(const char *in, char *out, tx_param &tx, ofdm_param &ofdm) {
int symbols = tx.n_sym * 48;
for (int i = 0; i < symbols; i++) {
out[i] = 0;
for(int k = 0; k < ofdm.n_bpsc; k++) {
assert(*in == 1 || *in == 0);
out[i] |= (*in << (ofdm.n_bpsc - k - 1));
in++;
}
}
}
void generate_bits(const char *psdu, char *data_bits, tx_param &tx) {
// first 16 bits are zero (SERVICE/DATA field)
memset(data_bits, 0, 16);
data_bits += 16;
for(int i = 0; i < tx.psdu_size; i++) {
for(int b = 0; b < 8; b++) {
data_bits[i * 8 + b] = !!(psdu[i] & (1 << b));
}
}
}
/* Here starts the SSE2 Viterbi convolutional decoder implementation */
/*
* Copyright 1995 Phil Karn, KA9Q
* Copyright 2008 Free Software Foundation, Inc.
* 2014 Added SSE2 implementation Bogdan Diaconescu
*
* This file is part of GNU Radio
*
* GNU Radio is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 3, or (at your option)
* any later version.
*
* GNU Radio is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with GNU Radio; see the file COPYING. If not, write to
* the Free Software Foundation, Inc., 51 Franklin Street,
* Boston, MA 02110-1301, USA.
*/
/*
* Viterbi decoder for K=7 rate=1/2 convolutional code
* Some modifications from original Karn code by Matt Ettus
* Major modifications by adding SSE2 code by Bogdan Diaconescu
*/
#include <xmmintrin.h>
/* The two generator polynomials for the NASA Standard K=7 code.
* Since these polynomials are known to be optimal for this constraint
* length there is not much point in changing them. But if you do, you
* will have to regenerate the BUTTERFLY macro calls in viterbi()
*/
#define POLYA 0x6d
#define POLYB 0x4f
/* The basic Viterbi decoder operation, called a "butterfly"
* operation because of the way it looks on a trellis diagram. Each
* butterfly involves an Add-Compare-Select (ACS) operation on the two nodes
* where the 0 and 1 paths from the current node merge at the next step of
* the trellis.
*
* The code polynomials are assumed to have 1's on both ends. Given a
* function encode_state() that returns the two symbols for a given
* encoder state in the low two bits, such a code will have the following
* identities for even 'n' < 64:
*
* encode_state(n) = encode_state(n+65)
* encode_state(n+1) = encode_state(n+64) = (3 ^ encode_state(n))
*
* Any convolutional code you would actually want to use will have
* these properties, so these assumptions aren't too limiting.
*
* Doing this as a macro lets the compiler evaluate at compile time the
* many expressions that depend on the loop index and encoder state and
* emit them as immediate arguments.
* This makes an enormous difference on register-starved machines such
* as the Intel x86 family where evaluating these expressions at runtime
* would spill over into memory.
*/
static union branchtab27 { unsigned char c[32]; __m128i v[2];} Branchtab27_sse2[2];
// Maximum number of traceback bytes
#define TRACEBACK_MAX 24
// Metrics for each state
static unsigned char mmresult[64] __attribute__((aligned(16))) = {0};
// Paths for each state
static unsigned char ppresult[TRACEBACK_MAX][64] __attribute__((aligned(16))) = {{0,0}};
// Position in circular buffer where the current decoded byte is stored
static int store_pos = 0;
#define BUTTERFLY(i,sym) { \
int m0,m1,m2,m3;\
/* ACS for 0 branch */\
m0 = state[i].metric + mets[sym]; /* 2*i */\
m1 = state[i+32].metric + mets[3 ^ sym]; /* 2*i + 64 */\
if(m0 > m1){\
next[2*i].metric = m0;\
next[2*i].path = state[i].path << 1;\
} else {\
next[2*i].metric = m1;\
next[2*i].path = (state[i+32].path << 1)|1;\
}\
/* ACS for 1 branch */\
m2 = state[i].metric + mets[3 ^ sym]; /* 2*i + 1 */\
m3 = state[i+32].metric + mets[sym]; /* 2*i + 65 */\
if(m2 > m3){\
next[2*i+1].metric = m2;\
next[2*i+1].path = state[i].path << 1;\
} else {\
next[2*i+1].metric = m3;\
next[2*i+1].path = (state[i+32].path << 1)|1;\
}\
}
unsigned char d_Partab[] = {
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
0, 1, 1, 0, 1, 0, 0, 1,
0, 1, 1, 0, 1, 0, 0, 1,
1, 0, 0, 1, 0, 1, 1, 0,
};/* Parity lookup table */
/* Convolutionally encode data into binary symbols */
unsigned char
d_encode(unsigned char *symbols,
unsigned char *data,
unsigned int nbytes,
unsigned char encstate)
{
int i;
while(nbytes-- != 0){
for(i=7;i>=0;i--){
encstate = (encstate << 1) | ((*data >> i) & 1);
*symbols++ = d_Partab[encstate & POLYA];
*symbols++ = d_Partab[encstate & POLYB];
}
data++;
}
return encstate;
}
/* Viterbi decoder */
int
d_viterbi(unsigned long *metric, /* Final path metric (returned value) */
unsigned char *data, /* Decoded output data */
unsigned char *symbols, /* Raw deinterleaved input symbols */
unsigned int nbits, /* Number of output bits */
int mettab[2][256] /* Metric table, [sent sym][rx symbol] */
){
unsigned int bitcnt = 0;
int mets[4];
long bestmetric;
int beststate,i;
struct viterbi_state state0[64],state1[64],*state,*next;
state = state0;
next = state1;
/* Initialize starting metrics to prefer 0 state */
state[0].metric = 0;
for(i=1;i<64;i++)
state[i].metric = -999999;
state[0].path = 0;
for(bitcnt = 0;bitcnt < nbits;bitcnt++){
/* Read input symbol pair and compute all possible branch
* metrics
*/
mets[0] = mettab[0][symbols[0]] + mettab[0][symbols[1]];
mets[1] = mettab[0][symbols[0]] + mettab[1][symbols[1]];
mets[2] = mettab[1][symbols[0]] + mettab[0][symbols[1]];
mets[3] = mettab[1][symbols[0]] + mettab[1][symbols[1]];
symbols += 2;
BUTTERFLY(0,0); BUTTERFLY(1,1); BUTTERFLY(2,3); BUTTERFLY(3,2);
BUTTERFLY(4,3); BUTTERFLY(5,2); BUTTERFLY(6,0); BUTTERFLY(7,1);
BUTTERFLY(8,0); BUTTERFLY(9,1); BUTTERFLY(10,3); BUTTERFLY(11,2);
BUTTERFLY(12,3); BUTTERFLY(13,2); BUTTERFLY(14,0); BUTTERFLY(15,1);
BUTTERFLY(16,2); BUTTERFLY(17,3); BUTTERFLY(18,1); BUTTERFLY(19,0);
BUTTERFLY(20,1); BUTTERFLY(21,0); BUTTERFLY(22,2); BUTTERFLY(23,3);
BUTTERFLY(24,2); BUTTERFLY(25,3); BUTTERFLY(26,1); BUTTERFLY(27,0);
BUTTERFLY(28,1); BUTTERFLY(29,0); BUTTERFLY(30,2); BUTTERFLY(31,3);
/* Swap current and next states */
if(bitcnt & 1){
state = state0;
next = state1;
} else {
state = state1;
next = state0;
}
// ETTUS
//if(bitcnt > nbits-7){
/* In tail, poison non-zero nodes */
//for(i=1;i<64;i += 2)
// state[i].metric = -9999999;
//}
/* Produce output every 8 bits once path memory is full */
if((bitcnt % 8) == 5 && bitcnt > 32){
/* Find current best path */
bestmetric = state[0].metric;
beststate = 0;
for(i=1;i<64;i++){
if(state[i].metric > bestmetric){
bestmetric = state[i].metric;
beststate = i;
}
}
#ifdef notdef
printf("metrics[%d] = %d state = %lx\n",beststate,
state[beststate].metric,state[beststate].path);
#endif
*data++ = state[beststate].path >> 24;
}
}
/* Output remaining bits from 0 state */
// ETTUS Find best state instead
bestmetric = state[0].metric;
beststate = 0;
for(i=1;i<64;i++){
if(state[i].metric > bestmetric){
bestmetric = state[i].metric;
beststate = i;
}
}
if((i = bitcnt % 8) != 6)
state[beststate].path <<= 6-i;
*data++ = state[beststate].path >> 24;
*data++ = state[beststate].path >> 16;
*data++ = state[beststate].path >> 8;
*data = state[beststate].path;
//printf ("BS = %d\tBSM = %d\tM0 = %d\n",beststate,state[beststate].metric,state[0].metric);
*metric = state[beststate].metric;
return 0;
}
void
d_viterbi_chunks_init(struct viterbi_state* state) {
// Initialize starting metrics to prefer 0 state
int i;
state[0].metric = 0;
state[0].path = 0;
for(i=1;i<64;i++)
{
state[i].metric = 0;//-999999;
state[i].path = 0;
}
}
void
d_viterbi_chunks_init_sse2(__m128i *mm0, __m128i *pp0) {
// Initialize starting metrics to prefer 0 state
int i, j;
for (i = 0; i < 4; i++)
{
mm0[i] = _mm_setzero_si128();
pp0[i] = _mm_setzero_si128();
}
int polys[2] = { POLYA, POLYB };
for(i=0; i < 32; i++)
{
Branchtab27_sse2[0].c[i] = (polys[0] < 0) ^ d_Partab[(2*i) & abs(polys[0])] ? 1 : 0;
Branchtab27_sse2[1].c[i] = (polys[1] < 0) ^ d_Partab[(2*i) & abs(polys[1])] ? 1 : 0;
}
for (i = 0; i < 64; i++)
{
mmresult[i] = 0;
for (j = 0; j < TRACEBACK_MAX; j++)
ppresult[j][i] = 0;
}
}
void
d_viterbi_butterfly8(unsigned char *symbols, int mettab[2][256], struct viterbi_state *state0, struct viterbi_state *state1)
{
unsigned int bitcnt;
int mets[4];
struct viterbi_state *state, *next;
state = state0;
next = state1;
// Operate on 16 symbols (8 bits) at a time
for(bitcnt = 0;bitcnt < 8;bitcnt++){
// Read input symbol pair and compute all possible branch metrics
mets[0] = mettab[0][symbols[0]] + mettab[0][symbols[1]];
mets[1] = mettab[0][symbols[0]] + mettab[1][symbols[1]];
mets[2] = mettab[1][symbols[0]] + mettab[0][symbols[1]];
mets[3] = mettab[1][symbols[0]] + mettab[1][symbols[1]];
symbols += 2;
// These macro calls were generated by genbut.c
BUTTERFLY(0,0); BUTTERFLY(1,1); BUTTERFLY(2,3); BUTTERFLY(3,2);
BUTTERFLY(4,3); BUTTERFLY(5,2); BUTTERFLY(6,0); BUTTERFLY(7,1);
BUTTERFLY(8,0); BUTTERFLY(9,1); BUTTERFLY(10,3); BUTTERFLY(11,2);
BUTTERFLY(12,3); BUTTERFLY(13,2); BUTTERFLY(14,0); BUTTERFLY(15,1);
BUTTERFLY(16,2); BUTTERFLY(17,3); BUTTERFLY(18,1); BUTTERFLY(19,0);
BUTTERFLY(20,1); BUTTERFLY(21,0); BUTTERFLY(22,2); BUTTERFLY(23,3);
BUTTERFLY(24,2); BUTTERFLY(25,3); BUTTERFLY(26,1); BUTTERFLY(27,0);
BUTTERFLY(28,1); BUTTERFLY(29,0); BUTTERFLY(30,2); BUTTERFLY(31,3);
// Swap current and next states
if(bitcnt & 1){
state = state0;
next = state1;
} else {
state = state1;
next = state0;
}
}
}
void
d_viterbi_butterfly2(unsigned char *symbols, int mettab[2][256], struct viterbi_state *state0, struct viterbi_state *state1)
{
int mets[4];
struct viterbi_state *state, *next;
state = state0;
next = state1;
// Operate on 4 symbols (2 bits) at a time
// Read input symbol pair and compute all possible branch metrics
#if 0
if (symbols[0] == 2)
{
mets[0] = mettab[0][symbols[1]];
mets[1] = mettab[1][symbols[1]];
mets[2] = mettab[0][symbols[1]];
mets[3] = mettab[1][symbols[1]];
}
else if (symbols[1] == 2)
{
mets[0] = mettab[0][symbols[0]];
mets[1] = mettab[1][symbols[0]];
mets[2] = mettab[0][symbols[0]];
mets[3] = mettab[1][symbols[0]];
}
else
#endif
{
mets[0] = mettab[0][symbols[0]] + mettab[0][symbols[1]];
mets[1] = mettab[0][symbols[0]] + mettab[1][symbols[1]];
mets[2] = mettab[1][symbols[0]] + mettab[0][symbols[1]];
mets[3] = mettab[1][symbols[0]] + mettab[1][symbols[1]];
}
// These macro calls were generated by genbut.c
BUTTERFLY(0,0); BUTTERFLY(1,1); BUTTERFLY(2,3); BUTTERFLY(3,2);
BUTTERFLY(4,3); BUTTERFLY(5,2); BUTTERFLY(6,0); BUTTERFLY(7,1);
BUTTERFLY(8,0); BUTTERFLY(9,1); BUTTERFLY(10,3); BUTTERFLY(11,2);
BUTTERFLY(12,3); BUTTERFLY(13,2); BUTTERFLY(14,0); BUTTERFLY(15,1);
BUTTERFLY(16,2); BUTTERFLY(17,3); BUTTERFLY(18,1); BUTTERFLY(19,0);
BUTTERFLY(20,1); BUTTERFLY(21,0); BUTTERFLY(22,2); BUTTERFLY(23,3);
BUTTERFLY(24,2); BUTTERFLY(25,3); BUTTERFLY(26,1); BUTTERFLY(27,0);
BUTTERFLY(28,1); BUTTERFLY(29,0); BUTTERFLY(30,2); BUTTERFLY(31,3);
state = state1;
next = state0;
// Read input symbol pair and compute all possible branch metrics
#if 0
if (symbols[2] == 2)
{
mets[0] = mettab[0][symbols[3]];
mets[1] = mettab[1][symbols[3]];
mets[2] = mettab[0][symbols[3]];
mets[3] = mettab[1][symbols[3]];
}
else if (symbols[3] == 2)
{
mets[0] = mettab[0][symbols[2]];
mets[1] = mettab[1][symbols[2]];
mets[2] = mettab[0][symbols[2]];
mets[3] = mettab[1][symbols[2]];
}
else
#endif
{
mets[0] = mettab[0][symbols[2]] + mettab[0][symbols[3]];
mets[1] = mettab[0][symbols[2]] + mettab[1][symbols[3]];
mets[2] = mettab[1][symbols[2]] + mettab[0][symbols[3]];
mets[3] = mettab[1][symbols[2]] + mettab[1][symbols[3]];
}
// These macro calls were generated by genbut.c
BUTTERFLY(0,0); BUTTERFLY(1,1); BUTTERFLY(2,3); BUTTERFLY(3,2);
BUTTERFLY(4,3); BUTTERFLY(5,2); BUTTERFLY(6,0); BUTTERFLY(7,1);
BUTTERFLY(8,0); BUTTERFLY(9,1); BUTTERFLY(10,3); BUTTERFLY(11,2);
BUTTERFLY(12,3); BUTTERFLY(13,2); BUTTERFLY(14,0); BUTTERFLY(15,1);
BUTTERFLY(16,2); BUTTERFLY(17,3); BUTTERFLY(18,1); BUTTERFLY(19,0);
BUTTERFLY(20,1); BUTTERFLY(21,0); BUTTERFLY(22,2); BUTTERFLY(23,3);
BUTTERFLY(24,2); BUTTERFLY(25,3); BUTTERFLY(26,1); BUTTERFLY(27,0);
BUTTERFLY(28,1); BUTTERFLY(29,0); BUTTERFLY(30,2); BUTTERFLY(31,3);
}
void
d_viterbi_butterfly2_sse2(unsigned char *symbols, __m128i *mm0, __m128i *mm1, __m128i *pp0, __m128i *pp1)
{
int i;
__m128i *metric0, *metric1;
__m128i *path0, *path1;
metric0 = mm0;
path0 = pp0;
metric1 = mm1;
path1 = pp1;
// Operate on 4 symbols (2 bits) at a time
__m128i m0, m1, m2, m3, decision0, decision1, survivor0, survivor1;
__m128i metsv, metsvm;
__m128i shift0, shift1;
__m128i tmp0, tmp1;
__m128i sym0v, sym1v;
sym0v = _mm_set1_epi8(symbols[0]);
sym1v = _mm_set1_epi8(symbols[1]);
for (i = 0; i < 2; i++)
{
if (symbols[0] == 2)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v);
metsv = _mm_sub_epi8(_mm_set1_epi8(1),metsvm);
}
else if (symbols[1] == 2)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v);
metsv = _mm_sub_epi8(_mm_set1_epi8(1),metsvm);
}
else
{
metsvm = _mm_add_epi8(_mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v),_mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v));
metsv = _mm_sub_epi8(_mm_set1_epi8(2),metsvm);
}
m0 = _mm_add_epi8(metric0[i], metsv);
m1 = _mm_add_epi8(metric0[i+2], metsvm);
m2 = _mm_add_epi8(metric0[i], metsvm);
m3 = _mm_add_epi8(metric0[i+2], metsv);
decision0 = _mm_cmpgt_epi8(_mm_sub_epi8(m0,m1),_mm_setzero_si128());
decision1 = _mm_cmpgt_epi8(_mm_sub_epi8(m2,m3),_mm_setzero_si128());
survivor0 = _mm_or_si128(_mm_and_si128(decision0,m0),_mm_andnot_si128(decision0,m1));
survivor1 = _mm_or_si128(_mm_and_si128(decision1,m2),_mm_andnot_si128(decision1,m3));
shift0 = _mm_slli_epi16(path0[i], 1);
shift1 = _mm_slli_epi16(path0[2+i], 1);
shift1 = _mm_add_epi8(shift1, _mm_set1_epi8(1));
metric1[2*i] = _mm_unpacklo_epi8(survivor0,survivor1);
tmp0 = _mm_or_si128(_mm_and_si128(decision0,shift0),_mm_andnot_si128(decision0,shift1));
metric1[2*i+1] = _mm_unpackhi_epi8(survivor0,survivor1);
tmp1 = _mm_or_si128(_mm_and_si128(decision1,shift0),_mm_andnot_si128(decision1,shift1));
path1[2*i] = _mm_unpacklo_epi8(tmp0, tmp1);
path1[2*i+1] = _mm_unpackhi_epi8(tmp0, tmp1);
}
metric0 = mm1;
path0 = pp1;
metric1 = mm0;
path1 = pp0;
sym0v = _mm_set1_epi8(symbols[2]);
sym1v = _mm_set1_epi8(symbols[3]);
for (i = 0; i < 2; i++)
{
if (symbols[2] == 2)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v);
metsv = _mm_sub_epi8(_mm_set1_epi8(1),metsvm);
}
else if (symbols[3] == 2)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v);
metsv = _mm_sub_epi8(_mm_set1_epi8(1),metsvm);
}
else
{
metsvm = _mm_add_epi8(_mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v),_mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v));
metsv = _mm_sub_epi8(_mm_set1_epi8(2),metsvm);
}
m0 = _mm_add_epi8(metric0[i], metsv);
m1 = _mm_add_epi8(metric0[i+2], metsvm);
m2 = _mm_add_epi8(metric0[i], metsvm);
m3 = _mm_add_epi8(metric0[i+2], metsv);
decision0 = _mm_cmpgt_epi8(_mm_sub_epi8(m0,m1),_mm_setzero_si128());
decision1 = _mm_cmpgt_epi8(_mm_sub_epi8(m2,m3),_mm_setzero_si128());
survivor0 = _mm_or_si128(_mm_and_si128(decision0,m0),_mm_andnot_si128(decision0,m1));
survivor1 = _mm_or_si128(_mm_and_si128(decision1,m2),_mm_andnot_si128(decision1,m3));
shift0 = _mm_slli_epi16(path0[i], 1);
shift1 = _mm_slli_epi16(path0[2+i], 1);
shift1 = _mm_add_epi8(shift1, _mm_set1_epi8(1));
metric1[2*i] = _mm_unpacklo_epi8(survivor0,survivor1);
tmp0 = _mm_or_si128(_mm_and_si128(decision0,shift0),_mm_andnot_si128(decision0,shift1));
metric1[2*i+1] = _mm_unpackhi_epi8(survivor0,survivor1);
tmp1 = _mm_or_si128(_mm_and_si128(decision1,shift0),_mm_andnot_si128(decision1,shift1));
path1[2*i] = _mm_unpacklo_epi8(tmp0, tmp1);
path1[2*i+1] = _mm_unpackhi_epi8(tmp0, tmp1);
}
}
void
d_viterbi_butterfly_sse2(unsigned char *symbols, __m128i *mm0, __m128i *mm1, __m128i *pp0, __m128i *pp1)
{
int i;
__m128i *metric0, *metric1;
__m128i *path0, *path1;
metric0 = mm0;
path0 = pp0;
metric1 = mm1;
path1 = pp1;
for (i = 0; i < 2; i++)
printf("symbols[%i]: %i\n", i, symbols[i]);
// Operate on 4 symbols (2 bits) at a time
__m128i m0, m1, m2, m3, decision0, decision1, survivor0, survivor1;
__m128i metsv, metsvm;
__m128i shift0, shift1;
__m128i tmp0, tmp1;
__m128i sym0v, sym1v;
sym0v = _mm_set1_epi8(symbols[0]);
sym1v = _mm_set1_epi8(symbols[1]);
for (i = 0; i < 2; i++)
{
#if 1
if (symbols[0] == 2)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v);
}
else if (symbols[1] == 0)
{
metsvm = _mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v);
}
else
#endif
metsvm = _mm_add_epi8(_mm_xor_si128(Branchtab27_sse2[0].v[i],sym0v),_mm_xor_si128(Branchtab27_sse2[1].v[i],sym1v));
metsv = _mm_sub_epi8(_mm_set1_epi8(2),metsvm);
m0 = _mm_add_epi8(metric0[i], metsv);
m1 = _mm_add_epi8(metric0[i+2], metsvm);
m2 = _mm_add_epi8(metric0[i], metsvm);
m3 = _mm_add_epi8(metric0[i+2], metsv);
decision0 = _mm_cmpgt_epi8(_mm_sub_epi8(m0,m1),_mm_setzero_si128());
decision1 = _mm_cmpgt_epi8(_mm_sub_epi8(m2,m3),_mm_setzero_si128());
survivor0 = _mm_or_si128(_mm_and_si128(decision0,m0),_mm_andnot_si128(decision0,m1));
survivor1 = _mm_or_si128(_mm_and_si128(decision1,m2),_mm_andnot_si128(decision1,m3));
shift0 = _mm_slli_epi16(path0[i], 1);
shift1 = _mm_slli_epi16(path0[2+i], 1);
shift1 = _mm_add_epi8(shift1, _mm_set1_epi8(1));
metric1[2*i] = _mm_unpacklo_epi8(survivor0,survivor1);
tmp0 = _mm_or_si128(_mm_and_si128(decision0,shift0),_mm_andnot_si128(decision0,shift1));
metric1[2*i+1] = _mm_unpackhi_epi8(survivor0,survivor1);
tmp1 = _mm_or_si128(_mm_and_si128(decision1,shift0),_mm_andnot_si128(decision1,shift1));
path1[2*i] = _mm_unpacklo_epi8(tmp0, tmp1);
path1[2*i+1] = _mm_unpackhi_epi8(tmp0, tmp1);
}
for (i = 0; i < 4; i++)
{
mm0[i] = metric1[i];
pp0[i] = path1[i];
}
}
unsigned char
d_viterbi_get_output(struct viterbi_state *state, unsigned char *outbuf) {
// Produce output every 8 bits once path memory is full
// if((bitcnt % 8) == 5 && bitcnt > 32) {
// Find current best path
unsigned int i, beststate = 0;
int bestmetric;
bestmetric = state[0].metric;
beststate = 0;
for(i=1;i<64;i++)
{
if(state[i].metric > bestmetric) {
bestmetric = state[i].metric;
beststate = i;
}
}
*outbuf = state[beststate].path >> 24;
return bestmetric;
}
unsigned char
d_viterbi_get_output_sse2(__m128i *mm0, __m128i *pp0, int ntraceback, unsigned char *outbuf) {
// Find current best path
int i;
int bestmetric, minmetric;
int beststate = 0;
int pos = 0;
// Implement a circular buffer with the last ntraceback paths
store_pos = (store_pos + 1) % ntraceback;
// TODO - find another way to extract the value
for (i = 0; i < 4; i++)
{
_mm_store_si128((__m128i *) &mmresult[i*16], mm0[i]);
_mm_store_si128((__m128i *) &ppresult[store_pos][i*16], pp0[i]);
}
// Find out the best final state
bestmetric = mmresult[beststate];
minmetric = mmresult[beststate];
for (i = 1; i < 64; i++)
{
if (mmresult[i] > bestmetric)
{
bestmetric = mmresult[i];
beststate = i;
}
if (mmresult[i] < minmetric)
minmetric = mmresult[i];
}
// Trace back
for (i = 0, pos = store_pos; i < (ntraceback - 1); i++)
{
// Obtain the state from the output bits
// by clocking in the output bits in reverse order.
// The state has only 6 bits
beststate = ppresult[pos][beststate] >> 2;
pos = (pos - 1 + ntraceback) % ntraceback;
}
// Store output byte
*outbuf = ppresult[pos][beststate];
// Zero out the path variable
// and prevent metric overflow
for (i = 0; i < 4; i++)
{
pp0[i] = _mm_setzero_si128();
mm0[i] = _mm_sub_epi8(mm0[i], _mm_set1_epi8(minmetric));
}
return bestmetric;
}
/* end of new functions */
Reference in New Issue View Git Blame Copy Permalink