First OTA Run

Before the First OTA Run

• Prepare two backend nodes. In downlink-only mode, BS uses one TX plus one sensing RX antenna path and UE uses one RX path. If duplex/uplink is enabled, UE also needs a TX antenna/RF chain and BS needs an RX antenna/RF chain for the uplink; in FDD mode the radios must also support the configured uplink carrier.
• Start from the sample YAML that is closest to your hardware. The repository includes X310/B210 examples, but it is not limited to those USRP models.
• Copy runtime YAMLs into build/, because both binaries read BS.yaml or UE.yaml from their working directory.
• If the frontend runs on another machine, point the Python viewer at the backend with --host <backend-ip> or the viewer’s Backend IP field. default_out_ip is only for destination-style output IPs, not UDP/ZMQ listen addresses.
• If you care about stable real-time behavior, run scripts/set_performance.bash, then apply CPU isolation with sudo ../scripts/isolate_cpus.py (asks BS/UE/BS+UE and, when enough kernel-isolated CPUs exist, optionally configures NIC IRQ isolation/RX-TX splitting), and only then launch via sudo ../scripts/isolate_cpus.py run ....

Typical Usage Example

1. Startup of the BS

sudo -s
cd build
# For X310:
cp ../config/BS_X310.yaml BS.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./BS

# For B210:
cp ../config/BS_B210.yaml BS.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./BS

# For B210 duplex:
cp ../config/BS_B210_Duplex.yaml BS.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./BS

If you are using a separate frontend computer, point the monostatic viewer at the BS backend IP with --host or the viewer’s Backend IP field.

2. Startup of the UE

sudo -s
cd build
# For X310:
cp ../config/UE_X310.yaml UE.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./UE

# For B210:
cp ../config/UE_B210.yaml UE.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./UE

# For B210 duplex:
cp ../config/UE_B210_Duplex.yaml UE.yaml
sudo ../scripts/isolate_cpus.py
sudo ../scripts/isolate_cpus.py run ./UE

If you are using a separate frontend computer, point the bistatic viewer at the UE backend IP with --host or the viewer’s Backend IP field. Set default_out_ip only for decoded/debug outputs that should be sent to another machine.

3. Send video on the BS computer

Confirm that BS and UE are running and UE has synchronized with BS. Open another terminal on the BS computer. Put video.mp4 in the current directory, or replace video.mp4 in the command below with the actual video file path.

The command below uses NVIDIA NVENC encoding, a 60 Mb/s video target, and a 70 Mb/s UDP sending limit for an established 100 MS/s X310 link. For a first test or a configuration with a lower sample rate, change all four 60M values to 3M and bitrate=70000000 to bitrate=4000000.

ffmpeg -re -stream_loop -1 -i video.mp4 \
  -map 0:v:0 -map '0:a:0?' \
  -c:v h264_nvenc -preset p4 -tune ll -rc cbr \
  -b:v 60M -minrate:v 60M -maxrate:v 60M -bufsize:v 60M \
  -g 30 -bf 0 -pix_fmt yuv420p -c:a aac -b:a 128k \
  -f mpegts -mpegts_flags +resend_headers -muxdelay 0 -muxpreload 0 \
  "udp://127.0.0.1:50000?pkt_size=1316&bitrate=70000000&burst_bits=10528&buffer_size=4194304"

ffmpeg sends the video and any audio to the local BS program on input port 50000. If it runs on another computer, replace 127.0.0.1 with the BS computer’s IP. Match the port to network_output.udp_input_port in BS.yaml.

For detailed video streaming settings, see Video and UDP Workflows.

4. Play video on the UE computer

Set network_output.udp_output_ip to 127.0.0.1 and network_output.udp_output_port to 50001 in UE.yaml. Open another terminal on the UE computer and run:

ffplay -fflags nobuffer -flags low_delay -framedrop -sync ext \
  "udp://127.0.0.1:50001?fifo_size=65536&overrun_nonfatal=1&buffer_size=16777216"

The UE program sends the received video data to local port 50001, where ffplay receives and plays it. For playback on another computer, follow the address and port settings in Video and UDP Workflows.

5. Run monostatic frontend

python3 ./scripts/plot_sensing_fast.py

This maintained viewer auto-selects CUDA, MLX, Intel GPU, or CPU backends.

6. Run bistatic frontend

python3 ./scripts/plot_bi_sensing_fast.py

This maintained viewer auto-selects CUDA, MLX, Intel GPU, or CPU backends.