summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorjaseg <git-bigdata-wsl-arch@jaseg.de>2020-02-28 19:30:27 +0100
committerjaseg <git-bigdata-wsl-arch@jaseg.de>2020-02-28 19:30:27 +0100
commit5effadcbaf66f476c8fffefa2c349676f41c3f52 (patch)
tree946f089be4ced12921ffb0178207c276a1d16aa3
parent89b32316aa46f22a4f967ca7953089a3df0c16e7 (diff)
downloadmaster-thesis-5effadcbaf66f476c8fffefa2c349676f41c3f52.tar.gz
master-thesis-5effadcbaf66f476c8fffefa2c349676f41c3f52.tar.bz2
master-thesis-5effadcbaf66f476c8fffefa2c349676f41c3f52.zip
LDPC decoder fully working
-rw-r--r--controller/fw/ldpc_decoder.c94
-rw-r--r--controller/fw/ldpc_decoder_test.py35
-rw-r--r--controller/fw/test_decoder.py27
3 files changed, 108 insertions, 48 deletions
diff --git a/controller/fw/ldpc_decoder.c b/controller/fw/ldpc_decoder.c
index 10b5a08..fe59d77 100644
--- a/controller/fw/ldpc_decoder.c
+++ b/controller/fw/ldpc_decoder.c
@@ -5,6 +5,9 @@
#include <math.h>
#include <stdio.h>
+
+void gausselimination(size_t n, size_t k, int8_t *A, int8_t *b);
+
void inner_logbp(
size_t m, size_t n,
size_t bits_count, size_t nodes_count, const uint32_t bits_values[], const uint32_t nodes_values[],
@@ -60,6 +63,7 @@ int decode(size_t n, size_t nodes_count, size_t bits_count, uint32_t bits[], int
for (unsigned int n_iter=0; n_iter<maxiter; n_iter++) {
inner_logbp(m, n, bits_count, nodes_count, bits, (uint32_t*)bits_transposed, y, Lq, Lr, n_iter, L_posteriori);
+ /*
float *arrs[3] = {Lq, Lr, L_posteriori};
const char *names[3] = {"Lq", "Lr", "L_posteriori"};
size_t lens[3] = {m*n, m*n, n};
@@ -86,6 +90,7 @@ int decode(size_t n, size_t nodes_count, size_t bits_count, uint32_t bits[], int
}
printf("\n]\n");
}
+ */
for (size_t i=0; i<n; i++)
out[i] = L_posteriori[i] <= 0.0f;
@@ -130,18 +135,18 @@ void inner_logbp(
/* step 1 : Horizontal */
unsigned int bits_counter = 0;
for (size_t i=0; i<m; i++) {
- printf("=== i=%zu\n", i);
+ //printf("=== i=%zu\n", i);
for (size_t p=bits_counter; p<bits_counter+bits_count; p++) {
size_t j = bits_values[p];
- printf("\033[38;5;240mj=%04zd ", j);
+ //printf("\033[38;5;240mj=%04zd ", j);
float x = 1;
if (n_iter == 0) {
for (size_t q=bits_counter; q<bits_counter+bits_count; q++) {
if (bits_values[q] != j) {
- int lcv = Lc[bits_values[q]];
- char *s = lcv < 0 ? "\033[91m" : (lcv > 0 ? "\033[92m" : "\033[94m");
- printf("nij=%04u Lc=%s%3d\033[38;5;240m ", bits_values[q], s, lcv);
+ //int lcv = Lc[bits_values[q]];
+ //char *s = lcv < 0 ? "\033[91m" : (lcv > 0 ? "\033[92m" : "\033[94m");
+ //printf("nij=%04u Lc=%s%3d\033[38;5;240m ", bits_values[q], s, lcv);
x *= tanhf(0.5f * Lc[bits_values[q]]);
}
}
@@ -153,7 +158,7 @@ void inner_logbp(
}
}
- printf("\n==== i=%03zd p=%01zd x=%08f\n", i, p-bits_counter, x);
+ //printf("\n==== i=%03zd p=%01zd x=%08f\n", i, p-bits_counter, x);
float num = 1 + x;
float denom = 1 - x;
@@ -197,31 +202,66 @@ void inner_logbp(
}
}
-/* TODO
-def get_message(tG, x):
- """Compute the original `n_bits` message from a `n_code` codeword `x`.
+/* Compute the original (k) bit message from a (n) bit codeword x.
+ *
+ * tG: (n, k)-matrix
+ * x: (n)-vector
+ * out: (k)-vector
+ */
+void get_message(size_t n, size_t k, int8_t *tG, int8_t *x, int8_t *out) {
- Parameters
- ----------
- tG: array (n_code, n_bits) coding matrix tG.
- x: array (n_code,) decoded codeword of length `n_code`.
+ gausselimination(n, k, tG, x);
- Returns
- -------
- message: array (n_bits,). Original binary message.
+ out[k - 1] = x[k - 1];
+ for (ssize_t i=k-2; i>=0; i--) {
+ out[i] = x[i];
- """
- n, k = tG.shape
+ uint8_t sum = 0;
+ for (size_t j=i+1; j<k; j++)
+ sum ^= tG[i*k + j] * out[j];
- rtG, rx = utils.gausselimination(tG, x)
+ out[i] = !!(out[i] - sum);
+ }
+}
- message = np.zeros(k).astype(int)
+/* Solve linear system in Z/2Z via Gauss Gauss elimination.
+ *
+ * A: (n, k)-matrix
+ * b: (n)-vector
+ */
+void gausselimination(size_t n, size_t k, int8_t *A, int8_t *b) {
+ ssize_t d = k<n ? k : n;
+ for (ssize_t j=0; j<d; j++) {
+
+ ssize_t pivot = -1;
+ for (size_t i=j; i<n; i++) {
+ if (A[i*k + j]) {
+ pivot = i;
+ break;
+ }
+ }
+ if (pivot == -1)
+ continue;
+
+ if (pivot != j) {
+ for (size_t i=0; i<k; i++) {
+ int8_t tmp = A[j*k + i];
+ A[j*k + i] = A[pivot*k + i];
+ A[pivot*k + i] = tmp;
+ }
- message[k - 1] = rx[k - 1]
- for i in reversed(range(k - 1)):
- message[i] = rx[i]
- message[i] -= utils.binaryproduct(rtG[i, list(range(i+1, k))],
- message[list(range(i+1, k))])
+ int8_t tmp = b[j];
+ b[j] = b[pivot];
+ b[pivot] = tmp;
+ }
+
+ for (size_t i=j+1; i<n; i++) {
+ if (A[i*k + j]) {
+ for (size_t p=0; p<k; p++)
+ A[i*k + p] = !!(A[i*k + p] - A[j*k + p]);
+ b[i] = !!(b[i] - b[j]);
+ }
+ }
+ }
+}
- return abs(message)
-*/
diff --git a/controller/fw/ldpc_decoder_test.py b/controller/fw/ldpc_decoder_test.py
index d5cbb24..3b91bba 100644
--- a/controller/fw/ldpc_decoder_test.py
+++ b/controller/fw/ldpc_decoder_test.py
@@ -16,8 +16,8 @@ args = parser.parse_args()
lib = C.CDLL('./ldpc_decoder_test.so')
-n = 384
-nodes, bits = 6, 8
+n = 5*19
+nodes, bits = 17, 19
H, G = pyldpc.make_ldpc(n, nodes, bits, systematic=False, seed=0)
_1, bits_pos, _2 = scipy.sparse.find(H)
_, k = G.shape
@@ -26,18 +26,23 @@ st = np.random.RandomState(seed=0)
test_data = st.randint(0, 2, k)
d = np.dot(G, test_data) % 2
x = (-1) ** d
+x[29:] = 0
bits_pos = bits_pos.astype(np.uint32)
x = x.astype(np.int8)
lib.decode.argtypes = [C.c_size_t, C.c_size_t, C.c_size_t, C.POINTER(C.c_size_t), C.POINTER(C.c_int8), C.POINTER(C.c_int8), C.c_uint]
+lib.get_message.argtypes = [C.c_size_t, C.c_size_t, C.POINTER(C.c_int8), C.POINTER(C.c_int8), C.POINTER(C.c_int8)]
if args.reference:
ref_out = test_decoder.decode(H, x, 3)
print('decoder output:', ref_out, flush=True)
- print('msg reconstruction:', pyldpc.get_message(G, ref_out))
- print('reference decoder: ', np.all(np.equal(pyldpc.get_message(G, ref_out), test_data)), flush=True)
- print(ref_out.dtype)
+ print('msg reconstruction:', test_decoder.get_message(G, ref_out))
+ print('reference decoder: ', np.all(np.equal(test_decoder.get_message(G, ref_out), test_data)), flush=True)
+ np.set_printoptions(linewidth=220)
+ print(test_data)
+ print(test_decoder.get_message(G, ref_out))
+ print(test_decoder.get_message(G, ref_out) ^ test_data)
else:
out = np.zeros(n, dtype=np.uint8)
@@ -48,6 +53,20 @@ else:
out.ctypes.data_as(C.POINTER(C.c_int8)),
25), flush=True)
print('decoder output:', out)
- print('msg reconstruction:', pyldpc.get_message(G, out))
- print('decoder under test:', np.all(np.equal(pyldpc.get_message(G, out.astype(np.int64)), test_data)))
- print(out.dtype)
+ print('msg reconstruction:', test_decoder.get_message(G, out.astype(np.int64)))
+ print('decoder under test:', np.all(np.equal(test_decoder.get_message(G, out.astype(np.int64)), test_data)))
+ np.set_printoptions(linewidth=220)
+ print(test_data)
+ print(test_decoder.get_message(G, out.astype(np.int64)))
+ G = G.astype(np.int8)
+ msg = np.zeros(k, dtype=np.int8)
+ lib.get_message(
+ n, k,
+ G.ctypes.data_as(C.POINTER(C.c_int8)),
+ out.astype(np.int8).ctypes.data_as(C.POINTER(C.c_int8)),
+ msg.ctypes.data_as(C.POINTER(C.c_int8)))
+ print(msg)
+ print(msg ^ test_data)
+
+print('codeword length:', len(x))
+print('data length:', len(test_data))
diff --git a/controller/fw/test_decoder.py b/controller/fw/test_decoder.py
index 806125b..8be5b02 100644
--- a/controller/fw/test_decoder.py
+++ b/controller/fw/test_decoder.py
@@ -43,23 +43,24 @@ def decode(H, y, snr, maxiter=100):
Lr = np.zeros(shape=(m, n, n_messages))
for n_iter in range(maxiter):
- print(f'============================ iteration {n_iter} ============================')
+ #print(f'============================ iteration {n_iter} ============================')
Lq, Lr, L_posteriori = _logbp_numba(bits_hist, bits_values, nodes_hist,
nodes_values, Lc, Lq, Lr, n_iter)
- print("Lq=", Lq.flatten())
- print("Lr=", Lr.flatten())
+ #print("Lq=", Lq.flatten())
+ #print("Lr=", Lr.flatten())
#print("L_posteriori=", L_posteriori.flatten())
- print('L_posteriori=[')
- for row in L_posteriori.reshape([-1, 16]):
- for val in row:
- cc = '\033[91m' if val < 0 else ('\033[92m' if val > 0 else '\033[94m')
- print(f"{cc}{val: 012.6g}\033[38;5;240m", end=', ')
- print()
+ #print('L_posteriori=[')
+ #for row in L_posteriori.reshape([-1, 16]):
+ # for val in row:
+ # cc = '\033[91m' if val < 0 else ('\033[92m' if val > 0 else '\033[94m')
+ # print(f"{cc}{val: 012.6g}\033[38;5;240m", end=', ')
+ # print()
x = np.array(L_posteriori <= 0).astype(int)
product = utils.incode(H, x)
if product:
+ print(f'found, n_iter={n_iter}')
break
if n_iter == maxiter - 1:
@@ -83,13 +84,13 @@ def _logbp_numba(bits_hist, bits_values, nodes_hist, nodes_values, Lc, Lq, Lr,
bits_counter = 0
nodes_counter = 0
for i in range(m):
- print(f'=== i={i}')
+ #print(f'=== i={i}')
ff = bits_hist[i]
ni = bits_values[bits_counter: bits_counter + ff]
bits_counter += ff
for j_iter, j in enumerate(ni):
nij = ni[:]
- print(f'\033[38;5;240mj={j:04d}', end=' ')
+ #print(f'\033[38;5;240mj={j:04d}', end=' ')
X = np.ones(n_messages)
if n_iter == 0:
@@ -97,13 +98,13 @@ def _logbp_numba(bits_hist, bits_values, nodes_hist, nodes_values, Lc, Lq, Lr,
if nij[kk] != j:
lcv = Lc[nij[kk],0]
lcc = '\033[91m' if lcv < 0 else ('\033[92m' if lcv > 0 else '\033[94m')
- print(f'nij={nij[kk]:04d} Lc={lcc}{lcv:> 8f}\033[38;5;240m', end=' ')
+ #print(f'nij={nij[kk]:04d} Lc={lcc}{lcv:> 8f}\033[38;5;240m', end=' ')
X *= np.tanh(0.5 * Lc[nij[kk]])
else:
for kk in range(len(nij)):
if nij[kk] != j:
X *= np.tanh(0.5 * Lq[i, nij[kk]])
- print(f'\n==== {i:03d} {j_iter:01d} {X[0]:> 8f}')
+ #print(f'\n==== {i:03d} {j_iter:01d} {X[0]:> 8f}')
num = 1 + X
denom = 1 - X
for ll in range(n_messages):