Welcome and introduction to the SDR/DSP devroom, some personal highlights of the past year and program description.
Abstract: The Opulent Voice Protocol (OVP) is an open-source digital voice protocol designed for bandwidth-constrained radio communications, including satellite and terrestrial amateur radio links. Developed through the peer-reviewed research and development process at Open Research Institute, OVP addresses the critical need for high-quality voice communication protocols that are freely implementable without licensing restrictions.
Built around the 16 kbps Opus voice codec, OVP delivers superior voice quality that exceeds existing amateur digital voice modes while seamlessly integrating voice, keyboard chat, and data in a unified protocol. This eliminates the need for separate, clunky packet data modes. This talk will explore OVP's architecture, performance characteristics, and design trade-offs. The first implementation target for the modem is the PLUTO SDR and can be found here https://github.com/OpenResearchInstitute/pluto_msk
Key Technical Features in the Reference Implementation: Minimum shift keying modulation has constant envelope and no phase discontinuities. Optimized for low SNR conditions, with forward error correction and flywheel synchronization. Efficient bandwidth utilization suitable for 70 cm and above amateur bands. Current hardware implementation on FPGA enables a future open source ASIC design.
We'll cover: The architectural decisions behind OVP's design, showing how domain modeling of the radio channel shaped protocol choices. Audio quality comparisons between OVP and legacy digital voice modes. The integrated communication model, which allows voice, chat, and data to coexist in a single protocol. Performance analyses. Integration with existing SDR platforms and open-source radio stacks. Lessons learned from deploying OVP over the air. The peer-review process and how open collaboration improved the protocol.
Human-radio interface project is here: https://github.com/OpenResearchInstitute/interlocutor Processor-side codebase (Xilinx/AMD 7010 Zynq) is here: https://github.com/OpenResearchInstitute/dialogus Satellite simulator is here: https://github.com/OpenResearchInstitute/locus
This talk is relevant to anyone interested in software-defined radio, open hardware communications systems, space technology, or building robust protocols for constrained environments.
At a time when Global Navigation Satellite System (GNSS) signal spoofing and jamming has never been easier, time and frequency has become an ubiquitous commodity most distributed communication infrastructure rely on. Returning to the pre-space era of long range communication using very low frequency (VLF) signals, we investigate some of the remaining VLF time and frequency transfer signals. Despite their long communication range, the need for bulky antennas and low VLF noise environment makes the direct reception of these signals impractical. In this presentation, we collect timestamped records of VLF signals collected throughout the world from the KiwiSDR networl, and assess the performance of the broadcast signals and the receivers.
WSDR is a web-based platform for real-time signal processing, application development, and custom workflow creation—all in a plug-and-play environment. Built on WebAssembly, WebUSB, and WebSockets, it supports even demanding workloads, including running a full open-source cellular network directly in the browser with all DSP executed on the frontend. In this session, we’ll show how WSDR simplifies building and deploying custom applications.
The Parks-McClellan (Remez) algorithm is a filter design algorithm that is optimal in the sense that it minimizes the maximum error between the desired and realized transfer functions. Many implementations of this algorithm exist, including in GNU Radio and SciPy. However, some of these have issues such as numerical stability for some filter design problems. I will give a summary of the Remez algorithm, why there are different possible implementations, and why some may be better than others. I will also explain how to use this algorithm for some practical filter design problems. The talk is intended partly as publicity for the pm-remez Python/Rust modern implementation by the speaker and partly as a tutorial on filter design.
ZigRadio is a lightweight flow graph signal processing framework built with Zig that features ergonomic syntax, minimal dependencies, easy cross-compilation, and seamless integration into host applications. This talk introduces the project, discusses aspects of the Zig language leveraged by the framework, provides examples of standalone radio receivers as well as integrated applications, and outlines the future roadmap.
Since 2022, CERN uses White Rabbit to distribute the radio frequency signal in the SPS accelerator and the LHC accelerator is planned to also use White Rabbit for the next run (2030).
This talk will give a short overview of the CERN accelerator complex, how RF is used to accelerate protons and what is White Rabbit. It will then discuss why the RF phase is very important for an accelerator, how clocks are synchronized, how RF is computed, digitally distributed and locally regenerated using dedicated electronics.
Module developed for this project: https://gitlab.cern.ch/be-cem-edl/chronos/wr2rf-vme/-/wikis/home
White Rabbit (WR) is a digital synchronization protocol over Gb Ethernet whose development is centralized by CERN with contributions from high energy physics communities including accelerators and detectors. The sub-ns synchronization capability provided by WR makes it well suited for distributed-SDR system synchronization, but the phase detection mechanism requires two tunable oscillators hardly found in generic FPGA boards. Thanks to the advances of FPGA internal clocking circuitry, WR has been demonstrated to run on generic boards not fitted with these peripherals: we demonstrate WR synchronization of the (low cost) AcornCLE215 board and Enjoy Digital's M2SDR board with some limitations over the dedicated hardware and induced by the AD936x RF frontent.
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.
PlutoSDR is running a minimal Linux distribution as the flash memory is limited to 32MB. Recent clones support SD card booting which extends these capabilities to several Gigabytes. The purpose of the presentation is to show how to take advantage of this extra storage to use the platform in a new way. - Debian 12 based - Software pre-installed (Python, GNU Radio, gpredict, maia-sdr api...) - Toolchain to achieve it : - Buildroot : https://github.com/F5OEO/tezuka_fw - rootfs : https://github.com/F5OEO/adi-kuiper-gen - FPGA : https://github.com/F5OEO/maia-sdr/tree/sweep - Web interface to use it, no need to install software on PC - FPGA DSP support (FFT, decimation...) - Optimizing GNU Radio flows for smoother performance - Use case: Web transceiver https://github.com/F5OEO/Remote-SDR-Tezuka
At the Dwingeloo Radio Telescope, we've developed a suite of tools that stream IQ data from SDRs using VRT (VITA 49.0 Radio Transport) over ZeroMQ to multiple clients simultaneously. This architecture enables us to run various applications—including SigMF recording, spectral analysis, pulsar dedispersion, correlation, and more—on the same data stream in real time.
Using this setup, we have successfully received signals from Voyager 1, conducted lunar radar experiments in a bi-static configuration with Astropeiler Stockert, and even achieved a Venus radar bounce.
These tools are highly generic and have found applications beyond radio astronomy. In this talk, we'll provide an overview of the design philosophy and practical usage of these tools, illustrated with examples from our work at the Dwingeloo Radio Telescope.
https://github.com/tftelkamp/vrt-iq-tools/
The MAX2771 chip provides broadband (<44 MHz) signal processing and digitization in the lower (1.1-1.3 GHz) and upper (1.5-1.65 GHz) L-band. Initially dedicated to GNSS signal reception, two such chips clocked by a common frequency reference can be used for passive radar or direction of arrival measurements. In this presentation, we tackle the challenges of PLL setpoint drift leading to phase variations despite the common clock, the impact of low ADC resolution on the recorded signal characteristics, and on correlation calculation. These results are not confined to this particular chip, but valid for any low SNR recordings, including for radioastronomical measurement systems. We demonstrate the use of this setup on passive radar measurements using various ground based and spaceborne sources of opportunity.