forked from xdsopl/LDPC
-
Notifications
You must be signed in to change notification settings - Fork 2
/
testbench.cc
243 lines (206 loc) · 8.31 KB
/
testbench.cc
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
/*
LDPC testbench
Copyright 2018 Ahmet Inan <xdsopl@gmail.com>
*/
#include <stdlib.h>
#include <iostream>
#include <iomanip>
#include <random>
#include <cmath>
#include <cassert>
#include <chrono>
#include <cstring>
#include <algorithm>
#include <functional>
#include "testbench.hh"
#include "encoder.hh"
#include "algorithms.hh"
#include "interleaver.hh"
#include "modulation.hh"
#if 0
#include "flooding_decoder.hh"
static const int TRIALS = 50;
#else
#include "layered_decoder.hh"
static const int TRIALS = 25;
#endif
LDPCInterface *create_ldpc(char *standard, char prefix, int number);
Interleaver<code_type> *create_interleaver(char *modulation, char *standard, char prefix, int number);
ModulationInterface<complex_type, code_type> *create_modulation(char *name, int len);
int main(int argc, char **argv)
{
if (argc != 6)
return -1;
typedef NormalUpdate<simd_type> update_type;
//typedef SelfCorrectedUpdate<simd_type> update_type;
//typedef MinSumAlgorithm<simd_type, update_type> algorithm_type;
typedef OffsetMinSumAlgorithm<simd_type, update_type, FACTOR> algorithm_type;
//typedef MinSumCAlgorithm<simd_type, update_type, FACTOR> algorithm_type;
//typedef LogDomainSPA<simd_type, update_type> algorithm_type;
//typedef LambdaMinAlgorithm<simd_type, update_type, 3> algorithm_type;
//typedef SumProductAlgorithm<simd_type, update_type> algorithm_type;
LDPCEncoder<code_type> encode;
LDPCDecoder<simd_type, algorithm_type> decode;
LDPCInterface *ldpc = create_ldpc(argv[2], argv[3][0], atoi(argv[3]+1));
if (!ldpc) {
std::cerr << "no such table!" << std::endl;
return -1;
}
const int CODE_LEN = ldpc->code_len();
const int DATA_LEN = ldpc->data_len();
std::cerr << "testing LDPC(" << CODE_LEN << ", " << DATA_LEN << ") code." << std::endl;
encode.init(ldpc);
decode.init(ldpc);
ModulationInterface<complex_type, code_type> *mod = create_modulation(argv[4], CODE_LEN);
if (!mod) {
std::cerr << "no such modulation!" << std::endl;
return -1;
}
const int MOD_BITS = mod->bits();
assert(CODE_LEN % MOD_BITS == 0);
const int SYMBOLS = CODE_LEN / MOD_BITS;
Interleaver<code_type> *itl = create_interleaver(argv[4], argv[2], argv[3][0], atoi(argv[3]+1));
assert(itl);
value_type SNR = atof(argv[1]);
//value_type mean_signal = 0;
value_type sigma_signal = 1;
value_type mean_noise = 0;
value_type sigma_noise = std::sqrt(sigma_signal * sigma_signal / (2 * std::pow(10, SNR / 10)));
std::cerr << SNR << " Es/N0 => AWGN with standard deviation of " << sigma_noise << " and mean " << mean_noise << std::endl;
value_type code_rate = (value_type)DATA_LEN / (value_type)CODE_LEN;
value_type spectral_efficiency = code_rate * MOD_BITS;
value_type EbN0 = 10 * std::log10(sigma_signal * sigma_signal / (spectral_efficiency * 2 * sigma_noise * sigma_noise));
std::cerr << EbN0 << " Eb/N0, using spectral efficiency of " << spectral_efficiency << " from " << code_rate << " code rate and " << MOD_BITS << " bits per symbol." << std::endl;
std::random_device rd;
std::default_random_engine generator(rd());
typedef std::uniform_int_distribution<int> uniform;
typedef std::normal_distribution<value_type> normal;
auto data = std::bind(uniform(0, 1), generator);
auto awgn = std::bind(normal(mean_noise, sigma_noise), generator);
int BLOCKS = atoi(argv[5]);
if (BLOCKS < 1)
return -1;
void *aligned_buffer = aligned_alloc(sizeof(simd_type), sizeof(simd_type) * CODE_LEN);
simd_type *simd = reinterpret_cast<simd_type *>(aligned_buffer);
code_type *code = new code_type[BLOCKS * CODE_LEN];
code_type *orig = new code_type[BLOCKS * CODE_LEN];
code_type *noisy = new code_type[BLOCKS * CODE_LEN];
complex_type *symb = new complex_type[BLOCKS * SYMBOLS];
for (int j = 0; j < BLOCKS; ++j)
for (int i = 0; i < DATA_LEN; ++i)
code[j * CODE_LEN + i] = 1 - 2 * data();
for (int j = 0; j < BLOCKS; ++j)
encode(code + j * CODE_LEN, code + j * CODE_LEN + DATA_LEN);
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
orig[i] = code[i];
for (int i = 0; i < BLOCKS; ++i)
itl->fwd(code + i * CODE_LEN);
for (int j = 0; j < BLOCKS; ++j)
mod->mapN(symb + j * SYMBOLS, code + j * CODE_LEN);
for (int i = 0; i < BLOCKS * SYMBOLS; ++i)
symb[i] += complex_type(awgn(), awgn());
if (1) {
code_type tmp[MOD_BITS];
value_type sp = 0, np = 0;
for (int i = 0; i < SYMBOLS; ++i) {
mod->hard(tmp, symb[i]);
complex_type s = mod->map(tmp);
complex_type e = symb[i] - s;
sp += std::norm(s);
np += std::norm(e);
}
value_type snr = 10 * std::log10(sp / np);
sigma_signal = std::sqrt(sp / SYMBOLS);
sigma_noise = std::sqrt(np / (2 * sp));
std::cerr << snr << " Es/N0, stddev " << sigma_noise << " of noise and " << sigma_signal << " of signal estimated via hard decision." << std::endl;
}
// $LLR=log(\frac{p(x=+1|y)}{p(x=-1|y)})$
// $p(x|\mu,\sigma)=\frac{1}{\sqrt{2\pi}\sigma}}e^{-\frac{(x-\mu)^2}{2\sigma^2}}$
value_type precision = FACTOR / (sigma_noise * sigma_noise);
for (int j = 0; j < BLOCKS; ++j)
mod->softN(code + j * CODE_LEN, symb + j * SYMBOLS, precision);
for (int i = 0; i < BLOCKS; ++i)
itl->bwd(code + i * CODE_LEN);
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
noisy[i] = code[i];
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
assert(!std::isnan(code[i]));
int iterations = 0;
int num_decodes = 0;
auto start = std::chrono::system_clock::now();
for (int j = 0; j < BLOCKS; j += SIMD_WIDTH) {
int blocks = j + SIMD_WIDTH > BLOCKS ? BLOCKS - j : SIMD_WIDTH;
for (int n = 0; n < blocks; ++n)
for (int i = 0; i < CODE_LEN; ++i)
reinterpret_cast<code_type *>(simd+i)[n] = code[(j+n)*CODE_LEN+i];
int trials = TRIALS;
int count = decode(simd, simd + DATA_LEN, trials, blocks);
++num_decodes;
for (int n = 0; n < blocks; ++n)
for (int i = 0; i < CODE_LEN; ++i)
code[(j+n)*CODE_LEN+i] = reinterpret_cast<code_type *>(simd+i)[n];
if (count < 0) {
iterations += blocks * trials;
std::cerr << "decoder failed at converging to a code word!" << std::endl;
} else {
iterations += blocks * (trials - count);
std::cerr << trials - count << " iterations were needed." << std::endl;
}
}
auto end = std::chrono::system_clock::now();
auto msec = std::chrono::duration_cast<std::chrono::milliseconds>(end - start);
int kbs = (BLOCKS * DATA_LEN + msec.count() / 2) / msec.count();
std::cerr << kbs << " kilobit per second." << std::endl;
float avg_iter = (float)iterations / (float)BLOCKS;
std::cerr << avg_iter << " average iterations per block." << std::endl;
float avg_msec = (float)msec.count() / (float)num_decodes;
std::cerr << avg_msec << " average milliseconds per decode." << std::endl;
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
assert(!std::isnan(code[i]));
int awgn_errors = 0;
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
awgn_errors += noisy[i] * orig[i] < 0;
int quantization_erasures = 0;
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
quantization_erasures += !noisy[i];
int uncorrected_errors = 0;
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
uncorrected_errors += code[i] * orig[i] <= 0;
int decoder_errors = 0;
for (int i = 0; i < BLOCKS * CODE_LEN; ++i)
decoder_errors += code[i] * orig[i] <= 0 && orig[i] * noisy[i] > 0;
float bit_error_rate = (float)uncorrected_errors / (float)(BLOCKS * CODE_LEN);
if (1) {
for (int i = 0; i < CODE_LEN; ++i)
code[i] = code[i] < 0 ? -1 : 1;
itl->fwd(code);
value_type sp = 0, np = 0;
for (int i = 0; i < SYMBOLS; ++i) {
complex_type s = mod->map(code + i * MOD_BITS);
complex_type e = symb[i] - s;
sp += std::norm(s);
np += std::norm(e);
}
value_type snr = 10 * std::log10(sp / np);
sigma_signal = std::sqrt(sp / SYMBOLS);
sigma_noise = std::sqrt(np / (2 * sp));
std::cerr << snr << " Es/N0, stddev " << sigma_noise << " of noise and " << sigma_signal << " of signal estimated from corrected symbols." << std::endl;
}
std::cerr << awgn_errors << " errors caused by AWGN." << std::endl;
std::cerr << quantization_erasures << " erasures caused by quantization." << std::endl;
std::cerr << decoder_errors << " errors caused by decoder." << std::endl;
std::cerr << uncorrected_errors << " errors uncorrected." << std::endl;
std::cerr << bit_error_rate << " bit error rate." << std::endl;
if (0) {
std::cout << SNR << " " << bit_error_rate << " " << avg_iter << " " << EbN0 << std::endl;
}
delete ldpc;
delete mod;
delete itl;
free(aligned_buffer);
delete[] code;
delete[] orig;
delete[] noisy;
delete[] symb;
return 0;
}