def conv_encode(message_bits, trellis, termination = 'term', puncture_matrix=None):
"""
Encode bits using a convolutional code.
Parameters
----------
message_bits : 1D ndarray containing {0, 1}
Stream of bits to be convolutionally encoded.
trellis: pre-initialized Trellis structure.
termination: {'cont', 'term'}, optional
Create ('term') or not ('cont') termination bits.
puncture_matrix: 2D ndarray containing {0, 1}, optional
Matrix used for the puncturing algorithm
Returns
-------
coded_bits : 1D ndarray containing {0, 1}
Encoded bit stream.
"""
k = trellis.k
n = trellis.n
total_memory = trellis.total_memory
rate = float(k)/n
code_type = trellis.code_type
if puncture_matrix is None:
puncture_matrix = np.ones((trellis.k, trellis.n))
number_message_bits = np.size(message_bits)
if termination == 'cont':
inbits = message_bits
number_inbits = number_message_bits
number_outbits = int(number_inbits/rate)
else:
# Initialize an array to contain the message bits plus the truncation zeros
if code_type == 'rsc':
inbits = message_bits
number_inbits = number_message_bits
number_outbits = int((number_inbits + k * total_memory)/rate)
else:
number_inbits = number_message_bits + total_memory + total_memory % k
inbits = np.zeros(number_inbits, 'int')
# Pad the input bits with M zeros (L-th terminated truncation)
inbits[0:number_message_bits] = message_bits
number_outbits = int(number_inbits/rate)
outbits = np.zeros(number_outbits, 'int')
if puncture_matrix is not None:
p_outbits = np.zeros(number_outbits, 'int')
else:
p_outbits = np.zeros(int(number_outbits*
puncture_matrix[0:].sum()/np.size(puncture_matrix, 1)), 'int')
next_state_table = trellis.next_state_table
output_table = trellis.output_table
# Encoding process - Each iteration of the loop represents one clock cycle
current_state = 0
j = 0
for i in range(int(number_inbits/k)): # Loop through all input bits
current_input = bitarray2dec(inbits[i*k:(i+1)*k])
current_output = output_table[current_state][current_input]
outbits[j*n:(j+1)*n] = dec2bitarray(current_output, n)
current_state = next_state_table[current_state][current_input]
j += 1
if code_type == 'rsc' and termination == 'term':
term_bits = dec2bitarray(current_state, trellis.total_memory)
term_bits = term_bits[::-1]
for i in range(trellis.total_memory):
current_input = bitarray2dec(term_bits[i*k:(i+1)*k])
current_output = output_table[current_state][current_input]
outbits[j*n:(j+1)*n] = dec2bitarray(current_output, n)
current_state = next_state_table[current_state][current_input]
j += 1
j = 0
for i in range(number_outbits):
if puncture_matrix[0][i % np.size(puncture_matrix, 1)] == 1:
p_outbits[j] = outbits[i]
j = j + 1
return p_outbits
def _acs_traceback(r_codeword, trellis, decoding_type,
path_metrics, paths, decoded_symbols,
decoded_bits, tb_count, t, count,
tb_depth, current_number_states):
k = trellis.k
n = trellis.n
number_states = trellis.number_states
number_inputs = trellis.number_inputs
branch_metric = 0.0
next_state_table = trellis.next_state_table
output_table = trellis.output_table
pmetrics = np.empty(number_inputs)
index_array = np.empty([number_states, 2], 'int')
# Loop over all the current states (Time instant: t)
for state_num in range(current_number_states):
# Using the next state table find the previous states and inputs
# leading into the current state (Trellis)
number_found = _where_c(next_state_table, number_states, number_inputs, state_num, index_array)
# Loop over all the previous states (Time instant: t-1)
for i in range(number_found):
previous_state = index_array[i, 0]
previous_input = index_array[i, 1]
# Using the output table, find the ideal codeword
i_codeword = output_table[previous_state, previous_input]
i_codeword_array = dec2bitarray(i_codeword, n)
# Compute Branch Metrics
if decoding_type == 'hard':
branch_metric = hamming_dist(r_codeword.astype(int), i_codeword_array.astype(int))
elif decoding_type == 'soft':
neg_LL_0 = np.log(np.exp(r_codeword) + 1) # negative log-likelihood to have received a 0
neg_LL_1 = neg_LL_0 - r_codeword # negative log-likelihood to have received a 1
branch_metric = np.where(i_codeword_array, neg_LL_1, neg_LL_0).sum()
elif decoding_type == 'unquantized':
i_codeword_array = 2*i_codeword_array - 1
branch_metric = euclid_dist(r_codeword, i_codeword_array)
# ADD operation: Add the branch metric to the
# accumulated path metric and store it in the temporary array
pmetrics[i] = path_metrics[previous_state, 0] + branch_metric
# COMPARE and SELECT operations
# Compare and Select the minimum accumulated path metric
path_metrics[state_num, 1] = pmetrics.min()
# Store the previous state corresponding to the minimum
# accumulated path metric
min_idx = pmetrics.argmin()
paths[state_num, tb_count] = index_array[min_idx, 0]
# Store the previous input corresponding to the minimum
# accumulated path metric
decoded_symbols[state_num, tb_count] = index_array[min_idx, 1]
if t >= tb_depth - 1:
current_state = path_metrics[:,1].argmin()
# Traceback Loop
for j in reversed(range(1, tb_depth)):
dec_symbol = decoded_symbols[current_state, j]
previous_state = paths[current_state, j]
decoded_bitarray = dec2bitarray(dec_symbol, k)
decoded_bits[t - tb_depth + 1 + (j - 1) * k + count:t - tb_depth + 1 + j * k + count] = decoded_bitarray
current_state = previous_state
paths[:,0:tb_depth-1] = paths[:,1:]
decoded_symbols[:,0:tb_depth-1] = decoded_symbols[:,1:]
def __init__(self, memory, g_matrix, feedback=None, code_type='default', polynomial_format='MSB'):
[self.k, self.n] = g_matrix.shape
self.code_type = code_type
self.total_memory = memory.sum()
self.number_states = pow(2, self.total_memory)
self.number_inputs = pow(2, self.k)
self.next_state_table = np.zeros([self.number_states,
self.number_inputs], 'int')
self.output_table = np.zeros([self.number_states,
self.number_inputs], 'int')
if isinstance(feedback, int):
warn('Trellis will only accept feedback as a matrix in the future. '
'Using the backwards compatibility version that may contain bugs for k > 1 or with LSB format.',
DeprecationWarning)
if code_type == 'rsc':
for i in range(self.k):
g_matrix[i][i] = feedback
# Compute the entries in the next state table and the output table
for current_state in range(self.number_states):
for current_input in range(self.number_inputs):
outbits = np.zeros(self.n, 'int')
# Compute the values in the output_table
for r in range(self.n):
output_generator_array = np.zeros(self.k, 'int')
shift_register = dec2bitarray(current_state,
self.total_memory)
for l in range(self.k):
# Convert the number representing a polynomial into a
# bit array
generator_array = dec2bitarray(g_matrix[l][r],
memory[l] + 1)
# Loop over M delay elements of the shift register
# to compute their contribution to the r-th output
for i in range(memory[l]):
outbits[r] = (outbits[r] + \
(shift_register[i + l] * generator_array[i + 1])) % 2
output_generator_array[l] = generator_array[0]
if l == 0:
feedback_array = (dec2bitarray(feedback, memory[l] + 1)[1:] * shift_register[0:memory[l]]).sum()
shift_register[1:memory[l]] = \
shift_register[0:memory[l] - 1]
shift_register[0] = (dec2bitarray(current_input,
self.k)[0] + feedback_array) % 2
else:
feedback_array = (dec2bitarray(feedback, memory[l] + 1) *
shift_register[
l + memory[l - 1] - 1:l + memory[l - 1] + memory[l] - 1]).sum()
shift_register[l + memory[l - 1]:l + memory[l - 1] + memory[l] - 1] = \
shift_register[l + memory[l - 1] - 1:l + memory[l - 1] + memory[l] - 2]
shift_register[l + memory[l - 1] - 1] = \
(dec2bitarray(current_input, self.k)[l] + feedback_array) % 2
# Compute the contribution of the current_input to output
outbits[r] = (outbits[r] + \
(np.sum(dec2bitarray(current_input, self.k) * \
output_generator_array + feedback_array) % 2)) % 2
# Update the ouput_table using the computed output value
self.output_table[current_state][current_input] = \
bitarray2dec(outbits)
# Update the next_state_table using the new state of
# the shift register
self.next_state_table[current_state][current_input] = \
bitarray2dec(shift_register)
else:
if polynomial_format == 'MSB':
bit_order = -1
elif polynomial_format in ('LSB', 'Matlab'):
bit_order = 1
else:
raise ValueError('polynomial_format must be "LSB", "MSB" or "Matlab"')
if feedback is None:
feedback = np.identity(self.k, int)
if polynomial_format in ('LSB', 'Matlab'):
feedback *= 2**memory.max()
max_values_lign = memory.max() + 1 # Max number of value on a delay lign
# feedback_array[i] holds the i-th bit corresponding to each feedback polynomial.
feedback_array = np.zeros((max_values_lign, self.k, self.k), np.int8)
for i in range(self.k):
for j in range(self.k):
binary_view = dec2bitarray(feedback[i, j], max_values_lign)[::bit_order]
feedback_array[:max_values_lign, i, j] = binary_view[-max_values_lign-2:]
# g_matrix_array[i] holds the i-th bit corresponding to each g_matrix polynomial.
g_matrix_array = np.zeros((max_values_lign, self.k, self.n), np.int8)
for i in range(self.k):
for j in range(self.n):
binary_view = dec2bitarray(g_matrix[i, j], max_values_lign)[::bit_order]
g_matrix_array[:max_values_lign, i, j] = binary_view[-max_values_lign-2:]
# shift_regs holds on each column the state of a shift register.
# The first row is the input of each shift reg.
shift_regs = np.empty((max_values_lign, self.k), np.int8)
# Compute the entries in the next state table and the output table
for current_state in range(self.number_states):
for current_input in range(self.number_inputs):
current_state_array = dec2bitarray(current_state, self.total_memory)
# Set the first row as the input.
shift_regs[0] = dec2bitarray(current_input, self.k)
# Set the other rows based on the current_state
idx = 0
for idx_mem, mem in enumerate(memory):
shift_regs[1:mem+1, idx_mem] = current_state_array[idx:idx + mem]
idx += mem
# Compute the output table
outputs_array = np.einsum('ik,ikl->l', shift_regs, g_matrix_array) % 2
self.output_table[current_state, current_input] = bitarray2dec(outputs_array)
# Update the first line based on the feedback polynomial
np.einsum('ik,ilk->l', shift_regs, feedback_array, out=shift_regs[0])
shift_regs %= 2
# Update current state array and compute next state table
idx = 0
for idx_mem, mem in enumerate(memory):
current_state_array[idx:idx + mem] = shift_regs[:mem, idx_mem]
idx += mem
self.next_state_table[current_state, current_input] = bitarray2dec(current_state_array)