def __init__(self, args, p_array1, p_array2):
super(DEC_LargeCNN2Int, self).__init__()
self.args = args
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.interleaver1 = Interleaver(args, p_array1)
self.deinterleaver1 = DeInterleaver(args, p_array1)
self.interleaver2 = Interleaver(args, p_array2)
self.deinterleaver2 = DeInterleaver(args, p_array2)
self.dec1_cnns = torch.nn.ModuleList()
self.dec2_cnns = torch.nn.ModuleList()
self.dec1_outputs = torch.nn.ModuleList()
self.dec2_outputs = torch.nn.ModuleList()
for idx in range(args.num_iteration):
self.dec1_cnns.append(SameShapeConv1d(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec2_cnns.append(SameShapeConv1d(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec1_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, 1))
else:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
def __init__(self, args, p_array1, p_array2):
# turbofy only for code rate 1/3
super(ENC_interCNN2Int, self).__init__(args)
self.args = args
# Encoder
self.enc_cnn_1 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_1 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_2 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_2 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_3 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_3 = torch.nn.Linear(args.enc_num_unit, 1)
self.interleaver1 = Interleaver(args, p_array1)
self.interleaver2 = Interleaver(args, p_array2)
def __init__(self, args, p_array):
# turbofy only for code rate 1/3
super(ENC_interCNN2Int, self).__init__(args)
self.args = args
# Encoder
self.enc_cnn_1 = SameShapeConv1d(num_layer=args.enc_num_layer, in_channels=args.code_rate_k,
out_channels= args.enc_num_unit, kernel_size = args.dec_kernel_size)
self.enc_linear_1 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_2 = SameShapeConv1d(num_layer=args.enc_num_layer, in_channels=args.code_rate_k,
out_channels= args.enc_num_unit, kernel_size = args.dec_kernel_size)
self.enc_linear_2 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_3 = SameShapeConv1d(num_layer=args.enc_num_layer, in_channels=args.code_rate_k,
out_channels= args.enc_num_unit, kernel_size = args.dec_kernel_size)
self.enc_linear_3 = torch.nn.Linear(args.enc_num_unit, 1)
self.interleaver1 = Interleaver(args, p_array)
seed2 = 1000
rand_gen2 = mtrand.RandomState(seed2)
p_array2 = rand_gen2.permutation(arange(args.block_len))
print('p_array1', p_array)
print('p_array2', p_array2)
self.interleaver2 = Interleaver(args, p_array2)
def __init__(self, args, p_array):
super(TurboAE_decoder1D, self).__init__()
self.args = args
cuda = True if torch.cuda.is_available() else False
self.this_device = torch.device("cuda" if cuda else "cpu")
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
self.dec1_cnns = torch.nn.ModuleList()
self.dec2_cnns = torch.nn.ModuleList()
self.dec1_outputs = torch.nn.ModuleList()
self.dec2_outputs = torch.nn.ModuleList()
for idx in range(args.num_iteration):
self.dec1_cnns.append(SameShapeConv1d(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec2_cnns.append(SameShapeConv1d(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec1_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.code_rate_k))
else:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
# also need some CNN for f
self.ftstart = SameShapeConv2d(num_layer=args.dec_num_layer, in_channels=args.img_channels,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
self.ftend = SameShapeConv2d(num_layer=1, in_channels=args.dec_num_unit,
out_channels= args.img_channels, kernel_size = args.dec_kernel_size)
def __init__(self, args, p_array):
super(ENC_interRNN_sys, self).__init__(args)
self.args = args
# Encoder
self.enc_rnn = torch.nn.GRU(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.enc_rnn_int = torch.nn.GRU(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_int = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.interleaver = Interleaver(args, p_array)
def __init__(self, args, p_array):
# turbofy only for code rate 1/3
super(ENC_turbofy_rate2, self).__init__(args)
self.args = args
# Encoder
self.enc_rnn_1 = torch.nn.GRU(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_1 = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.enc_rnn_2 = torch.nn.GRU(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_2 = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.interleaver = Interleaver(args, p_array)
def __init__(self,
args,
input_size=1,
is_systematic_bit=False,
is_interleave=False,
p_array=[]):
super(CNN_encoder, self).__init__()
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.args = args
self.is_systematic_bit = is_systematic_bit
self.is_interleave = is_interleave
if self.is_interleave:
self.interleaver = Interleaver(args, p_array)
self.p_array = p_array
# Encoder
self.enc_cnn = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=input_size,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_linear = torch.nn.Linear(args.enc_num_unit, 1)
def __init__(self, args, p_array):
super(FTAE_Shareddecoder, self).__init__()
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.args = args
# interleaver
self.p_array = p_array
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
if args.cnn_type == 'dense':
CNNModel = DenseSameShapeConv1d
else:
CNNModel = SameShapeConv1d
self.dec1_cnns = CNNModel(num_layer=args.dec_num_layer,
in_channels=2 + args.num_iter_ft,
out_channels=args.dec_num_unit,
kernel_size=args.dec_kernel_size)
self.dec1_outputs = torch.nn.Linear(args.dec_num_unit,
args.num_iter_ft)
self.dec2_cnns = CNNModel(num_layer=args.dec_num_layer,
in_channels=2 + args.num_iter_ft,
out_channels=args.dec_num_unit,
kernel_size=args.dec_kernel_size)
self.dec2_outputs = torch.nn.Linear(args.dec_num_unit,
args.num_iter_ft)
self.final_outputs = torch.nn.Linear(args.num_iter_ft, 1)
def __init__(self, args, p_array):
super(CCE_Turbo_Encoder1D, self).__init__()
self.args = args
cuda = True if torch.cuda.is_available() else False
self.this_device = torch.device("cuda" if cuda else "cpu")
# Define 1D Network for TurboAE
self.enc_cnn_1 = SameShapeConv1d(num_layer=args.cce_num_layer,
in_channels=args.code_rate_k,
out_channels=args.cce_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_1 = torch.nn.Linear(args.cce_num_unit, 1)
self.enc_cnn_2 = SameShapeConv1d(num_layer=args.cce_num_layer,
in_channels=args.code_rate_k,
out_channels=args.cce_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_2 = torch.nn.Linear(args.cce_num_unit, 1)
self.enc_cnn_3 = SameShapeConv1d(num_layer=args.cce_num_layer,
in_channels=args.code_rate_k,
out_channels=args.cce_num_unit,
kernel_size=args.dec_kernel_size)
self.enc_linear_3 = torch.nn.Linear(args.cce_num_unit, 1)
self.interleaver = Interleaver(args, p_array)
self.norm = torch.nn.BatchNorm1d(self.args.code_rate_n, affine=True)
def __init__(self, args, p_array):
super(FTAE_encoder, self).__init__()
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.args = args
# interleaver
self.p_array = p_array
self.interleaver = Interleaver(args, p_array)
if self.args.cnn_type == 'dense':
CNNModel = DenseSameShapeConv1d
else:
CNNModel = SameShapeConv1d
# Encoder
self.enc_cnn_1 = CNNModel(num_layer=args.enc_num_layer, in_channels=args.code_rate_k,
out_channels= args.enc_num_unit, kernel_size = args.enc_kernel_size)
self.enc_linear_1 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_2 = CNNModel(num_layer=args.enc_num_layer, in_channels=args.code_rate_k + 2,
out_channels= args.enc_num_unit, kernel_size = args.enc_kernel_size)
self.enc_linear_2 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_cnn_3 = CNNModel(num_layer=args.enc_num_layer, in_channels=args.code_rate_k + 4,
out_channels= args.enc_num_unit, kernel_size = args.enc_kernel_size)
self.enc_linear_3 = torch.nn.Linear(args.enc_num_unit, 1)
def __init__(self, args, p_array):
super(DEC_LargeRNN_rate2, self).__init__()
self.args = args
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
self.dec1_rnns = torch.nn.ModuleList()
self.dec2_rnns = torch.nn.ModuleList()
self.dec1_outputs = torch.nn.ModuleList()
self.dec2_outputs = torch.nn.ModuleList()
for idx in range(args.num_iteration):
self.dec1_rnns.append(torch.nn.GRU(1 + args.num_iter_ft, args.dec_num_unit,
num_layers=2, bias=True, batch_first=True,
dropout=args.dropout, bidirectional=True))
self.dec2_rnns.append(torch.nn.GRU(1+ args.num_iter_ft, args.dec_num_unit,
num_layers=2, bias=True, batch_first=True,
dropout=args.dropout, bidirectional=True))
self.dec1_outputs.append(torch.nn.Linear(2*args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(2*args.dec_num_unit, 1))
else:
self.dec2_outputs.append(torch.nn.Linear(2*args.dec_num_unit, args.num_iter_ft))
def __init__(self, args, p_array):
super(DEC_LargeCNN, self).__init__()
self.args = args
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
self.dec1_cnns = torch.nn.ModuleList()
self.dec2_cnns = torch.nn.ModuleList()
self.dec1_outputs = torch.nn.ModuleList()
self.dec2_outputs = torch.nn.ModuleList()
if self.args.encoder == 'TurboAE_rate3_cnn':
CNNLayer = SameShapeConv1d
else:
CNNLayer = DenseSameShapeConv1d
for idx in range(args.num_iteration):
self.dec1_cnns.append(CNNLayer(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec2_cnns.append(CNNLayer(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec1_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, 1))
else:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
def __init__(self, args, p_array):
super(FTAE_decoder, self).__init__()
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.this_device = torch.device("cuda" if use_cuda else "cpu")
self.args = args
# interleaver
self.p_array = p_array
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
# Decoder
self.dec1_cnns = torch.nn.ModuleList()
self.dec2_cnns = torch.nn.ModuleList()
self.dec1_outputs = torch.nn.ModuleList()
self.dec2_outputs = torch.nn.ModuleList()
if args.cnn_type =='dense':
CNNModel = DenseSameShapeConv1d
else:
CNNModel = SameShapeConv1d
for idx in range(args.num_iteration):
if self.args.codec == 'turboae_blockdelay_cnn':
self.dec1_cnns.append(CNNModel(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec2_cnns.append(CNNModel(num_layer=args.dec_num_layer, in_channels=2 + args.num_iter_ft,
out_channels= args.dec_num_unit, kernel_size = args.dec_kernel_size)
)
self.dec1_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.code_rate_k))
else:
self.dec2_outputs.append(torch.nn.Linear(args.dec_num_unit, args.num_iter_ft))
else: # RNN based
self.dec1_cnns.append(torch.nn.GRU(2 + args.num_iter_ft, args.dec_num_unit,
num_layers=args.dec_num_layer, bias=True, batch_first=True,
dropout=0, bidirectional=True)
)
self.dec2_cnns.append(torch.nn.GRU(2 + args.num_iter_ft, args.dec_num_unit,
num_layers=args.dec_num_layer, bias=True, batch_first=True,
dropout=0, bidirectional=True)
)
self.dec1_outputs.append(torch.nn.Linear(2*args.dec_num_unit, args.num_iter_ft))
if idx == args.num_iteration -1:
self.dec2_outputs.append(torch.nn.Linear(2*args.dec_num_unit, args.code_rate_k))
else:
self.dec2_outputs.append(torch.nn.Linear(2*args.dec_num_unit, args.num_iter_ft))
def __init__(self, args, p_array):
# turbofy only for code rate 1/3
super(ENC_interRNN, self).__init__(args)
self.enc_rnns = torch.nn.ModuleList()
self.args = args
# Encoder
if args.enc_rnn == 'gru':
RNN_MODEL = torch.nn.GRU
elif args.enc_rnn == 'lstm':
RNN_MODEL = torch.nn.LSTM
else:
RNN_MODEL = torch.nn.RNN
self.enc_rnn_1 = RNN_MODEL(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_1 = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.enc_rnn_2 = RNN_MODEL(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_2 = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.enc_rnn_3 = RNN_MODEL(1,
args.enc_num_unit,
num_layers=args.enc_num_layer,
bias=True,
batch_first=True,
dropout=0,
bidirectional=True)
self.enc_linear_3 = torch.nn.Linear(2 * args.enc_num_unit, 1)
self.interleaver = Interleaver(args, p_array)
def __init__(self, args, p_array):
super(NeuralTurbofyDec, self).__init__()
self.args = args
self.interleaver = Interleaver(args, p_array)
self.deinterleaver = DeInterleaver(args, p_array)
self.dec_rnn = torch.nn.GRU(args.code_rate_n + args.num_iter_ft - 1 , args.dec_num_unit, num_layers=2, bias=True, batch_first=True,
dropout=args.dropout, bidirectional=True)
self.dec_out = torch.nn.Linear(2*args.dec_num_unit, args.num_iter_ft)
self.dec_final = torch.nn.Linear(args.num_iter_ft, 1)
use_cuda = not args.no_cuda and torch.cuda.is_available()
self.device = torch.device("cuda" if use_cuda else "cpu")
def __init__(self, args, p_array):
# turbofy only for code rate 1/3
super(ENC_interCNN, self).__init__(args)
self.args = args
# Encoder
if self.args.encoder == 'TurboAE_rate3_cnn':
self.enc_cnn_1 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_cnn_2 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_cnn_3 = SameShapeConv1d(num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
else: # Dense
self.enc_cnn_1 = DenseSameShapeConv1d(
num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_cnn_2 = DenseSameShapeConv1d(
num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_cnn_3 = DenseSameShapeConv1d(
num_layer=args.enc_num_layer,
in_channels=args.code_rate_k,
out_channels=args.enc_num_unit,
kernel_size=args.enc_kernel_size)
self.enc_linear_1 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_linear_2 = torch.nn.Linear(args.enc_num_unit, 1)
self.enc_linear_3 = torch.nn.Linear(args.enc_num_unit, 1)
self.interleaver = Interleaver(args, p_array)
