Downlink Transmission (MISO case)

[Mio-TW]

Dear Sionna Team,

I follow this example and want to create the downlink MISO transmission.
But I have no idea why "The number of transmitted streams is equal to the number of UT antennas in both uplink and downlink"?
(that is, NUM_STREAMS_PER_TX = NUM_UT_ANT)

This will cause the transmitter to transmit with only one antenna in my simulation.
The following is my codes:

# Configuration
NUM_UT = 1     
NUM_BS = 1
NUM_UT_ANT = 1
NUM_BS_ANT = 32
batch_size = 16
num_bits_per_symbol = 6
coderate = 0.5  

# The number of transmitted streams is equal to the number of UT antennas in both uplink and downlink
num_streams_per_tx = NUM_UT_ANT
rx_tx_association=np.array([[1]])
stream_manager = StreamManagement(rx_tx_association, num_streams_per_tx)

# Resource Grid
resource_grid = ResourceGrid(num_ofdm_symbols=14,
                             fft_size=128,
                             subcarrier_spacing=30e3,
                             num_tx=NUM_BS, 
                             num_streams_per_tx=num_streams_per_tx,
                             cyclic_prefix_length=10,
                             pilot_pattern="kronecker",
                             pilot_ofdm_symbol_indices=[2, 11])
# Setup antenna arrays
ut_array = AntennaArray(num_rows=1,
                        num_cols=int(NUM_UT_ANT/2),
                        polarization="dual",
                        polarization_type="VH",
                        antenna_pattern="38.901",
                        carrier_frequency=carrier_frequency)

bs_array = AntennaArray(num_rows=4,
                        num_cols=4,
                        polarization="dual",
                        polarization_type="VH",
                        antenna_pattern="38.901",
                        carrier_frequency=carrier_frequency)

# Configure the channel model
cdl = CDL(cdl_model="C",
          delay_spread=100e-9,
          carrier_frequency=3.5e9,
          ut_array,
          bs_array,
          direction="downlink",
          min_speed=0.0,
          max_speed=25.0)

ofdm_channel = OFDMChannel(channel_model=cdl,
                           resource_grid=resource_grid,
                           add_awgn=True,
                           normalize_channel=True,
                           return_channel=True)
######################################
## Transmitter
_binary_source = BinarySource()
_n = int(resource_grid.num_data_symbols * num_bits_per_symbol) 
_k = int(_n * coderate) 
_encoder = LDPC5GEncoder(_k, _n)
_mapper = Mapper("qam", num_bits_per_symbol)
_rg_mapper = ResourceGridMapper(resource_grid)

b = _binary_source([batch_size, NUM_BS, num_streams_per_tx, _k])
print("b = ", b.shape)   # (16, 1, 1, 4608)
c = _encoder(b)
print("c = ", c.shape)   # (16, 1, 1, 9216)
x = _mapper(c)
print("x = ", x.shape)   # (16, 1, 1, 1536)
x_rg = _rg_mapper(x)
print("x_rg = ", x_rg.shape)  # (16, 1, 1, 14, 128)

The dimension of "x_rg" is (16, 1, 1, 14, 128) but not (16, 1, 32, 14, 128).
I want to know how to correct.
Thank you so much.

[Mio-TW]

I even printed out the receiver side and found that there are also issues with the dimensions.
The dimension of y and h appears "0".

# Channel
ebno_db = 5
ebno_db_tf = np.array([ebno_db])
ebno_db_tf = tf.convert_to_tensor(ebno_db_tf, dtype=tf.float32)
no = ebnodb2no(ebno_db_tf, num_bits_per_symbol, coderate)
no_ = expand_to_rank(no, tf.rank(x_rg))

# Receiver
y, h = ofdm_channel([x_rg, no_])
print("y = ", y.shape)  # (16, 1, 0, 14, 128)
print("h = ", h.shape)  # (16, 1, 0, 1, 32, 14, 128)

[faycalaa]

Hi,

By having a quick look at your code, I noticed that NUM_UT_ANT is set to 1, and then the ut_array is created with num_cols=int(NUM_UT_ANT/2), which would lead to num_cols=0.

Could that be the issue?

[Mio-TW]

@faycalaa, thank you so much for pointing out this issue!
I modified my code to the following.

ut_array = AntennaArray(num_rows=1,
                        num_cols=1),
                        polarization="single",
                        polarization_type="V",
                        antenna_pattern="38.901",
                        carrier_frequency=carrier_frequency)

And now the dimensions are normal now:

print("y = ", y.shape)  # (16, 1, 1, 14, 128)
print("h = ", h.shape)  # (16, 1, 1, 1, 32, 14, 128)

But I still don't know why the dimension of "x_rg" is (16, 1, 1, 14, 128) instead of (16, 1, 32, 14, 128).
Also, I have no idea why "The number of transmitted streams is equal to the number of UT antennas in both uplink and downlink"?
Do you have any insights on this?

[jhoydis]

Hi @Mio-TW,

You need to understand that the number of streams is different from the number of antennas a device has. In your code, the user terminal has a single antenna. Therefore, it can only send or receive a single stream. The base station has 32 antennas which means that it could send and receive up to 32 streams. However, as there is only a single user terminal with a single antenna, the base station can also send only a single stream to this terminal.

In the downlink, x_rg is the resource grid before precoding by the base station. After precoding, it would have the dimensions that you expect.