RemoteRF supports experiments across cabled loopback setups, indoor over-the-air links, and distributed edge hosts. From a single Python script, you can control one radio, coordinate multiple independent transmitters and receivers, or collect labeled IQ datasets.
Experimental Setups
A RemoteRF deployment can offer different physical connections depending on how its radios are wired and placed:
Single-SDR Cabled Loopback
The TX and RX ports of a single SDR are connected through a coaxial cable and inline attenuator. Because TX and RX share the same local oscillator, there is no carrier frequency offset or multipath fading, making it useful for testing basic transceiver code and hardware scaling.
Cross-Device Cabled Link
An attenuated coaxial cable connects the TX port of one SDR to the RX port of a second SDR. This introduces real frequency and clock offsets between two separate radios while avoiding external interference and multipath.
Indoor Over-the-Air (OTA)
Multiple SDRs with antennas transmit and receive across a lab room, providing line-of-sight and indoor multipath channels for testing synchronization, equalization, and multi-antenna techniques.
Distributed Edge Hosts
Radios connected to remote RemoteRF Hosts (such as Raspberry Pi nodes placed in other rooms or buildings) allow longer-range and non-line-of-sight (NLOS) experiments from a single script.
Multi-SDR & MIMO Setups
1. Coordinating Multiple Independent Radios
Because each driver instance takes its own reservation token, one Python script can control multiple transmitters and receivers at the same time:
from remoterf.drivers.adalm_pluto import adi
sdr_tx = adi.Pluto(token="Pluto_00@ucla") # Primary transmitter
sdr_rx = adi.Pluto(token="Pluto_01@ucla") # Primary receiver
sdr_jammer = adi.Pluto(token="Pluto_02@ucla") # Co-channel interferer
# Configure and run multi-node experiment
...
for dev in (sdr_tx, sdr_rx, sdr_jammer):
dev.tx_destroy_buffer()
dev.rx_destroy_buffer()2. MIMO Experiments
For multi-antenna experiments such as beamforming or spatial multiplexing, reserve a MIMO-capable device such as the ADALM-PLUTO MIMO or USRP 2901:
import numpy as np
from remoterf.drivers import ensure_driver
TOKEN = "Pluto_MIMO_00@ucla"
ensure_driver(token=TOKEN)
from remoterf.drivers.adalm_pluto_mimo import PlutoMIMO
pluto = PlutoMIMO(token=TOKEN)
pluto.sample_rate = int(4e6)
pluto.rx_lo = int(2450e6)
pluto.rx_enabled_channels = [0, 1]
pluto.gain_control_mode_chan0 = "manual"
pluto.gain_control_mode_chan1 = "manual"
pluto.rx_hardwaregain_chan0 = 35
pluto.rx_hardwaregain_chan1 = 35
pluto.rx_buffer_size = 8192
# Returns multi-channel complex samples (shape: [2, 8192])
rx_data = pluto.rx()
pluto.rx_destroy_buffer()Collecting IQ Datasets
You can automate sweeps across frequencies, sample rates, gains, and devices in Python to build labeled IQ datasets:
import time
import numpy as np
from remoterf.drivers.adalm_pluto import adi
sdr_tx = adi.Pluto(token="Pluto_00@ucla")
sdr_rx = adi.Pluto(token="Pluto_01@ucla")
# Define waveforms to transmit
waveform_bank = {
"qpsk": ...,
"16qam": ...,
}
records = []
sdr_tx.sample_rate = sdr_rx.sample_rate = int(2e6)
sdr_tx.tx_lo = sdr_rx.rx_lo = int(915e6)
sdr_tx.tx_cyclic_buffer = True
sdr_rx.gain_control_mode_chan0 = "manual"
sdr_rx.rx_hardwaregain_chan0 = 30
sdr_rx.rx_buffer_size = 4096
for mod_name, waveform in waveform_bank.items():
sdr_tx.tx_destroy_buffer()
sdr_tx.tx(waveform.astype(np.complex64))
for tx_gain in range(-35, -5, 5):
sdr_tx.tx_hardwaregain_chan0 = tx_gain
sdr_rx.rx_destroy_buffer()
iq = sdr_rx.rx().astype(np.complex64)
records.append({
"modulation": mod_name,
"tx_gain_db": tx_gain,
"rx_gain_db": 30,
"fc_hz": 915_000_000,
"fs_hz": 2_000_000,
"timestamp_utc": time.time(),
"iq": iq,
})
sdr_tx.tx_destroy_buffer()
sdr_rx.rx_destroy_buffer()Reproducibility Tips
- Use manual receive gain (
'manual'): Automatic gain control rescales incoming signals dynamically, which affects SNR sweeps and power measurements. Set a fixed manual receive gain for consistent captures. - Match TX scaling and check RX clipping for your radio: Different SDR drivers use different sample conventions (for example,
uhdon USRP uses normalized floats in[-1.0, +1.0], whereaspyadi-iioon ADALM-PLUTO uses integer-scaled DAC/ADC ranges). Check your device guide and verify that received I/Q samples do not clip at the ADC rails during gain sweeps. - Clean up buffers when finished: On radios like the ADALM-PLUTO, call
sdr.rx_destroy_buffer()after changing settings to clear buffered samples, and callsdr.tx_destroy_buffer()at the end of your script so cyclic transmissions stop when your experiment ends. - Record hardware metadata: Save the device handles, carrier frequency, sample rate, RF bandwidth, TX/RX gains, and UTC timestamp alongside each saved IQ array.