EMMA-DRONE v0.1

Warning

Pre-implementation. This page describes proposed contracts. Class signatures, parameter types, schema fields, and behavior are subject to change before code lands. Once implementation exists, content here will be regenerated from docstrings or sourced from running tests.

Dataset. EMMA-DRONE-v0.1 covers the task, target domain, input, labels, splits, and the download command; this recipe pins only the rfgen configuration. Source. rfgen (primary).

rfgen recipe

  • rfgen commit: <rfgen-commit> (frozen at release; placeholder until pinned).

  • Emitters: the rfgen emitter_zoo group, drone_rf family (OFDM telemetry at 2.4 and 5.8 GHz), spanning multiple drone models. Emitter selection and placement come from the rfgen scene and placement groups.

  • Receiver array: 8-element ULA (uniform linear array), half-wavelength spacing (spacing_lambda: 0.5), via the rfgen MultiRXConfig.geometry member ULA. ULA serves the azimuth-dominant detection task. Loaded into ArraySpec at load time.

  • Channel environment(s): E_URBAN (Sionna UMa) and E_RURAL (Sionna RMa), pinned through the rfgen channel group.

  • Labeler + label extraction: the rfgen label group writes the drone-emitter class; LabelExtractor maps emitters[].drone_present to the detection target and emitters[].drone_model to the model label.

  • SNR (signal-to-noise ratio) regime: drawn from the rfgen channel.snr_db_range (proposed 0 to 30 dB).

  • Sample counts: approximately 120k scenes total.

  • Splits + seeds: train / dev / holdout seed ranges pinned at release; the holdout is a frozen secret seed, never published. OOD axis: LEAVE_ONE_ENVIRONMENT_OUT, secondary DRONE_MODEL.

Real-capture counterpart

EMMA-REAL-OOD-v0.1. Real drone-RF captures are not wired at v0.1 (synthetic only); the adapter set lands later.

Notes

  • This recipe is a near-monopoly for EMMA: every public drone-RF dataset (DroneRF, DroneDetect / RFUAV) is single-antenna, so multi-antenna drone AoA (angle-of-arrival) and localization are unevaluable in the existing literature.

  • No real-capture drone signatures at v0.1; the sim-to-real gap is asserted by recipe design, not yet measured.

References

  • Media Inhof et al. (DroneRF), 2019. The de-facto single-antenna drone-RF benchmark EMMA moves beyond. (verify)

  • Coluccia et al. (DroneDetect / RFUAV), 2019 to 2025. Competing single-antenna drone-RF datasets. (verify)

  • NVIDIA Sionna: Hoydis et al., “Sionna,” 2023, arXiv:2203.11854. The UMa and RMa models behind the two environments. (verify)

See Also