Deep-Waveform: A Learned OFDM Receiver Based on Deep Complex-valued Convolutional Networks

The BirdsEye RL/RF project enables localization of mobile radio frequency targets, e.g., drones operators, via commericial off-the-shelf sensors.

Automated RF analyzer

A toolkit for simulating stochastic and/or deterministic radio frequency aggregate spectrum (in both in-phase/quadrature and image formats) for testing sensing algorithms (e.g. detection, parameter estimation, classification).

Low-cost NFC security: authenticating tags via RF fingerprinting with Triplet and SoftTriple losses, effectively combating cloning with low-end SDR devices and a special card slot design.

Radio frequency interference classification via convolutional neural network.

Autonomous CW collector for the Web-888 SDR: records raw narrowband IQ of on-air Morse into MorseBase, a provenance-tracked training corpus where every capture knows the firmware, commit, band weights and dependency versions that produced it. The supervisor is deliberately LLM-free — collection survives the agent being absent, slow, or wrong.

RF-based drone model classification on the DroneDetect dataset. Features a lossless Parquet/DuckDB pipeline, spectral EDA, and an interpretable normalized Welch PSD + LDA classifier achieving 97% segment / 100% recording accuracy, benchmarked against a Spectrogram CNN.

DroneRF is an open-source PyTorch-based 1D CNN system that detects and classifies UAV RF transmissions directly from raw IQ signal samples.