Datasets

Having Standard Datasets is Key to Progress in the Field

We’ve attempted to compile a collection of standard datasets which can be used in repeated experimentation, measurement, and comparison by numerous researchers with an interest in the Radio domain.

These datasets allow Machine Learning researchers with new ideas to dive directly into an important research area without the need for compiling new or generating new datasets, and allows for direct comparison to efficacy of prior work.

Existing RadioML Datasets

RadioML 2016.10

A synthetic GNU Radio Generated dataset consisting of 11 modulation (8 digital and 3 analog) at a number of SNR levels – this is an improved version of 2016.04C.

RadioML 2016.04

A synthetic GNU Radio Generated dataset consisting of 11 modulation (8 digital and 3 analog).  This dataset provides three different labeled dataset within

– RadioML2016.04A: High SNR AWGN Only

– RadioML2016.04B: Harsh Fading, LO Offset, Noise

– RadioML2016.04C: Fading, LO Offset, Labeled AWGN Noise/SNR Levels