Skip to main content
FOSDEM PWA

Machine Learning on Air: Overview and Tutorial on Open-Source Machine Learning Frameworks for DSP and Radio

K.3.601 | Day 2 | 14:10 - 14:55 | Speakers: Andrej Rode

Machine Learning on Air: Overview and Tutorial on Open-Source Machine Learning Frameworks for DSP and Radio
A picture of a devroom at FOSDEM 2024
Open in browser
Get involved in the conversation!Join the chat

Notes

Abstract

For the past decade, artificial Intelligence (AI) and machine learning (ML) have revolutionized numerous research fields and industries. The machine learning community has not left out software-defined Radio (SDR) and digital signal processing (DSP).

Thankfully, this development has not been done behind closed doors, and plenty of frameworks have been released by research laboratories and industry with an open-source license. To name a few (in no particular order): Sionna (https://github.com/NVlabs/sionna), Commplax (https://github.com/remifan/commplax), MOKka (https://github.com/kit-cel/mokka), scikit-learn & numpy, and maybe some more.

The goal of this talk is to give an overview of existing frameworks combining DSP and ML, and present a short tutorial on some aspects of what is already possible.

Attachments

Speakers

Andrej Rode

Andrej is currently a PhD student at the Communications Engineering Lab (CEL) at the Karlsruhe Institute of Technology (KIT) in Germany. His research focuses on optimizing fiber-optical communication systems with the aid of machine learning. In particular, the physical layer design, including geometric and probabilistic constellation shaping in interplay with digital signal processing for high Baudrate systems are his focus areas.

For the past 10 years, Andrej has been involved with the GNU Radio project, serving multiple different roles, writing code and improving the C++ build system, running the continuous integration services, and running the core IT infrastructure for the project and conference.

Notice: The placeholder video image is licensed under CC BY-SA 4.0. The original image can be found hereChanges made to the image are: Cropped the image to a new ratio, part of the image was cut off.