EMMA-FP-v0.1¶

v0.1 · proposed-contract

Task. E-ID-FP - see Tasks. What you predict. which device emitted the signal. Target domain. DEFENSE - defense emitter identification (the DARPA RFMLS, RF Machine Learning Systems, lineage).

Input¶

Raw complex multi-antenna I/Q (in-phase and quadrature), shape (num_rx, 2, N) float32, with per-scene metadata. 8-element uniform linear array (ULA).

Labels¶

A device identifier; scored leave-one-unit-out so unseen devices are held out.

Splits¶

train and dev are public and regenerable; holdout is private and scored by prediction submission. About 60,000 scenes.

Get it¶

emma download --dataset EMMA-FP-v0.1 --split dev --out ./scenes

Metric¶

Top1Accuracy (top-1 closed-set accuracy) and EER (equal error rate). See Metrics.

What this tests¶

Scored leave-one-unit-out, which isolates device identity from capture session and receiver front-end, the confound that existing RF fingerprinting datasets do not separate.

How it is made¶

Generated by rfgen. The exact configuration (commit, emitters, channel, label extraction, seeds) is in the developer recipe.