A Real-World Dataset Generator for Specific Emitter Identification

Abstract

Generating high-quality, real-world, well-labeled datasets for radio frequency machine learning (RFML) applications often proves prohibitively cumbersome and expensive, leading to the low availability of high-fidelity, low-cost datasets. Specific emitter identification (SEI) in particular requires a hardware setup capable of supporting transmitting using many different radios, while automated modulation classification (AMC) performance is primarily driven by SNR, channel effects, and the similarity of modulation types. These factors give rise to the need for scalable methods of inexpensive dataset generation. This paper describes the design considerations and a proof-of-concept implementation of a blind user reconfigurable platform capable of creating SEI and AMC datasets throughout a variety of real-world conditions. This paper additionally describes the reliability and performance of the platform relative to existing real-world data generation methods and compares generated datasets to those already present in the literature. This work also describes the software post-processing steps taken to isolate, label, and cull captured data and transform these into a high-quality dataset.

An Analysis of Radio Frequency Transfer Learning Behavior

“…The collection of a captured dataset suitable for this work required developing a custom collection platform. The Blind-User-Reconfigurable Platform (BURP) is a semi-portable hardware testbed that provides the means of creating robust datasets for up to 100 transmitting devices over a long period of time and in a variety of environments [32]. More specifically, BURP coordinates a large number of Txs and multiple Rxs to automatically transmit, collect, and label data.…”

“…Additional details regarding the design of the experimental hardware setup, performance measurements, and support for potential future RFML experimentation can be found in [32].…”

“…Then, for each experiment performed herein, subsets of the data were selected from either the synthetic or captured master dataset using configuration files containing the desired metadata parameters. The synthetic master dataset is publicly available on IEEE DataPort [35], and the captured master dataset is described in greater detail in [32].…”

Transferring Learned Behaviors between Similar and Different Radios

“…Data are collected using a testbed setup consisting of a transmit-side system and a group of receive-side systems connected through a control and coordination back-plane. The testbed setup is reconfigurable for different collection scenarios and is the subject of a previous work, where its design, construction, and operation are presented in detail [ 38 ].…”

“…The transmit-side hardware consists of a transmitter host computer and an interchangeable complement of 60 transmitters; this number was chosen to evenly distribute transmitters across each of the 12 USB hubs as described in [ 38 ] without causing undue strain on the USB ports of the transmitter host. The chosen transmitters are Great Scott Gadgets Yard Stick One (YS1) radios because of their USB connection, availability in large quantities, and similarity to many other commercial IoT transmitters.…”