100GBit/s RF sample offload for RFSoC using GNU Radio and PYNQ

Šiaučiulis, Marius and Northcote, David and Goldsmith, Josh and Crockett, Louise H. and Kaladė, Šarūnas; (2023) 100GBit/s RF sample offload for RFSoC using GNU Radio and PYNQ. In: 2023 21st IEEE Interregional NEWCAS Conference (NEWCAS). IEEE Interregional NEWCAS Conference (NEWCAS) . IEEE, GBR, pp. 1-5. ISBN 9798350300246 (https://doi.org/10.1109/NEWCAS57931.2023.10198070)

[thumbnail of Siauciulis-etal-NEWCAS2023-100GBit-s-RF-sample-offload-for-RFSoC]
Preview
Text. Filename: Siauciulis_etal_NEWCAS2023_100GBit_s_RF_sample_offload_for_RFSoC.pdf
Accepted Author Manuscript
License: Strathprints license 1.0

Download (1MB)| Preview

Abstract

Modern software defined radio systems are capable of multi-gigabit-per-second sampling rates producing unprecedented amounts of digitized RF data. In applications such as wideband spectrum sensing and machine learning algorithms for cognitive radio, prototyping, and instrumentation, it is often impractical to process the acquired data locally in real-time. This motivates the need for a high speed connection to offload data to an accelerator application running on a secondary processing resource. In this paper, we present a novel hardware and software co-design using the AMD RFSoC 4x2 platform, PYNQ and GNU Radio projects. The demonstrated system is capable of continuous 80GBit/s offload in a 100GBit/s channel, utilising a GPU acceleration to rapidly process the Fast Fourier Transforms of a full 2GHz bandwidth RF signal at 60 frames per second.