OFDM-Sense

OFDM-Sense is a Joint Communication and Sensing (JCAS) research platform developed as the 2025/26 URI ELECOMP Capstone project under Dr. Guoyi Xu. It implements an OFDM transceiver on a USRP X310 software-defined radio (SDR) to explore the use of OFDM waveforms for simultaneous wireless communication and device localization via time-difference-of-arrival (TDOA).

Publication

Our paper, "2D Localization Leveraging OFDM Signals," has been accepted for publication in the 2026 IEEE ORSS proceedings.

O'Malley Sherlock*, Royaljohn Southamavong*, and Guoyi Xu, "2D Localization Leveraging OFDM Signals," in 2026 IEEE ORSS, 2026. (to appear)

A link will be added once available.

Table of Contents

• Publication
• How it Works
• Prerequisites
• Installation
• Configuration
• Workflow
• Localization Experiments
• Simulation
• Project Structure

---

How it Works

A transmitter sends an OFDM packet (preamble + data symbols) from a USRP X310. One or more receivers capture the signal. The receiver pipeline:

1. Synchronization Schmidl-Cox algorithm detects packet start and estimates coarse CFO
2. Channel estimation pilot symbols estimate frequency-domain channel response
3. Equalization & demodulation 16-QAM symbols are recovered and evaluated (EVM, BER, SER)
4. Delay estimation matched filter + sub-sample interpolation measures propagation delay
5. Localization TDOA across multiple receivers feeds a least-squares multilateration solver

---

Prerequisites

Hardware

• USRP X310
• External 10 MHz reference clock (shared across all devices for synchronization)
• 10 GbE connection per USRP

Software

• Python >= 3.8
• UHD (USRP Hardware Driver) required to build the C++ control binary
• CMake >= 3.8 and a C++14 compiler

---

Installation

1. Clone the repository

git clone https://github.com/osherlock1/OFDM-Sense.git
cd OFDM-Sense

2. Build the C++ USRP control binary

mkdir build && cd build
cmake ..
make
cd ..

This produces build/TXRX_FROM_FILE, which the Python scripts call via subprocess.

3. Install the Python package

python -m venv venv
source venv/bin/activate       # Windows: venv\Scripts\activate
pip install -r requirements.txt

4. Verify the installation

pytest

---

Configuration

USRP Hardware (configs/usrp_settings.yaml)

Copy and edit this file to match your hardware setup:

build_path: "./build/TXRX_FROM_FILE"
tx_addr: "addr0=192.168.30.2,addr1=192.168.40.2"
rx_addr: "addr0=192.168.30.2,addr1=192.168.40.2"

subdev: "A:0 B:0"
tx_channel_idx: "1"
rx_channel_idx: "0,1,2"

tx_rate: 100e6
rx_rate: 100e6
tx_freq: 60e6
rx_freq: 60e6

tx_gain: 0
rx_gain: 0
ref: "external"   # external 10 MHz reference clock

Update tx_addr / rx_addr to match your USRP IP addresses. The ref: "external" field requires a shared 10 MHz clock source connected to all devices.

---

Workflow

Quick start (simulation only - no hardware needed)

# Generate a synthetic OFDM packet with noise
python scripts/generate_packet.py --n_symb 30 --snr 20

# Unpack and evaluate
python scripts/unpack_rx.py --sim --plot

Hardware transfer

# 1. Generate the transmit packet
python scripts/generate_packet.py --n_symb 30

# 2. Verify the USRP is reachable (sends a sine wave and checks CFO)
python scripts/test_sin_wave.py

# 3. Run the transfer
python scripts/run_transfer.py

# 4. Unpack and evaluate the received signal
python scripts/unpack_rx.py --plot

---

Localization Experiments

See scripts/experiment_scripts/README.md for a full tutorial. The short version:

# 1. Edit EXPERIMENT_NAME, ROAMING_DEVICES, FIXED_DEVICES at the top of collect_raw_data.py

# 2. Collect data (prompts for device positions)
python scripts/experiment_scripts/collect_raw_data.py --runs 5 --experiments_dir ./experiments

# 3. Process raw .dat files into CSV
python scripts/experiment_scripts/process_experiment.py --experiment_pth ./experiments/my_experiment

# 4. Run multilateration
python scripts/localization/multilateration.py --experiment ./experiments/my_experiment --devices RX3ch1 --anchor ANCHORch0

---

Simulation

Run a Monte Carlo TDOA localization simulation without any hardware:

python scripts/simulation/monte_carlo.py --sigma-ns 0.1 --trials 1000

The simulation uses an interactive drag-and-drop UI to reposition TX and RX nodes and recompute localization error in real time.

---

Project Structure

OFDM-Sense/
├── src/ofdm/               # Installable Python library
│   ├── core/               # Waveform, preamble, payload construction
│   ├── channel/            # Channel estimation, CFO correction, delay
│   ├── modulation/         # 16-QAM
│   ├── processing/         # RX pipeline, batch processing
│   ├── simulation/         # TDOA geometry, solver, Monte Carlo
│   ├── utils/              # USRP config, data generation, evaluation
│   ├── viz/                # Plotting utilities
│   └── config.py           # OFDMConfig dataclass (N=256, CP=64, Fs=100MHz)
├── scripts/                # Runnable entry-point scripts
│   ├── experiment_scripts/ # Data collection and processing workflows
│   ├── localization/       # Multilateration
│   ├── simulation/         # Monte Carlo simulation
│   ├── delay/              # Delay estimation and calibration
│   └── image_demo/         # Image transmission demo
├── usrp_control_files/     # C++ UHD driver source
├── tests/                  # pytest test suite
├── configs/                # YAML hardware configuration
├── data_files/             # Reference packet data
├── notebooks/              # Analysis notebooks
└── CMakeLists.txt          # C++ build system