26–30 Sept 2022
Capital Hilton
US/Eastern timezone
All GRCon talks are now available to watch at https://www.youtube.com/GNURadioProject

gr-plasma: A New GNU Radio-based Tool for Software-defined Radar

29 Sept 2022, 15:20
30m
Presidential Ballroom (Capital Hilton)

Presidential Ballroom

Capital Hilton

Paper (with talk) High Performance SDR Applications Main Track

Speaker

Shane Flandermeyer (School of Electrical and Computer Engineering, Advanced Radar Research Center, University of Oklahoma)

Description

Building experimental radar systems has traditionally required expensive custom hardware. Such designs are typically difficult to program, preventing easy experimentation with new radar signal processing algorithms. While software-defined radio hardware is not inherently designed to operate as a radar, there have been numerous documented examples of software-defined radar implementations using commercial-off-the-shelf radio hardware – particularly from the GNU Radio community. In this paper and the accompanying talk, the gr-plasma out-of-tree module is presented as a convenient way to implement and test radar systems in GNU Radio. The module is an implementation of the signal processing in Plasma DSP, a separate companion library written in C++ that can be used outside the GNU Radio ecosystem. In addition to describing the operation and utilization of gr-plasma we will highlight the technical challenges we encountered in developing this software. In doing so, we hope to provide resources and motivation to enable other developers to easily use and further develop the gr-plasma architecture.

There are two key challenges with software-defined radar that differ from most communications protocols: transmit/receive synchronization and ultra-reliable high-speed duplex operation. Radar systems operate by transmitting electromagnetic waves into the environment and measuring reflections. To achieve localization accuracy, the transmit and receive processes must be synchronized down to the sample level. Additionally, the resolution of a radar sensor is inversely proportional to the bandwidth of waveform. Therefore, the software must be capable of both transmitting and receiving high-data rate streams to maximize radar performance. All of these challenges must be met in a flexible, easy-to-use software architecture to enable experimentation by both academics and hobbyists alike. We will describe the several approaches we have examined to achieve these data rates as well as compare their performance on both 10 gigabit ethernet and USB 3 links.

To date, we have focused development on pulsed radar operation, which is inherently bursty. Due to this property, gr-plasma uses GNU Radio's message passing interface to packetize each pulse into a protocol data unit (PDU). Consequently, downstream blocks may easily operate on individual pulses (e.g., for range processing) or on an integer number of pulses in a coherent processing interval. Unlike tagged stream approaches, the length of each pulse is not limited by the buffer size of each block, which becomes a limiting factor at high sample rates even when the pulse repetition interval (PRI) is small.

The gr-plasma module includes tools for natively generating linear FM (LFM) and arbitrary polyphase-coded FM (PCFM) waveforms, as well as any other arbitrary waveform desired from a file. These waveforms can be transmitted and received through UHD-compatible software-defined radios using a custom USRP Radar block. Rather than requiring a time-stamped PDU for every pulse, the block repeatedly transmits the data from the last PDU it has received and handles all pulse timing internally. This reduces the number of messages that must be processed during streaming while simplifying the design of the block. This architecture also facilitates “cognitive operation” with low latency. For example, an adaptive waveform controller block can use the results of downstream processing to send a message with an updated waveform that is processed on a sub-millisecond time scale. Moreover, the radar block can sustain similar sample rates in GNU Radio as in pure UHD because it does not incur any overhead from the GNU Radio scheduler. This functionality is key for research in radar operation in congested spectrum, such as dynamic spectrum access (DSA) applications. Finally, the module includes a custom file sink block for saving data from large PDUs, which also saves metadata that complies with the SigMF v1.0.0 standard or select extensions. This final feature is of key significance as software-defined radar research involves pulse-to-pulse changes in the radar waveform (e.g., for spectral co-existence, radar-embedded communications, or other emerging research topics) that challenge traditional metadata formats. In the final paper and presentation, the utility and structure of each block will be discussed, challenges that arose during development will be described, and the results of range-doppler processing will be shown for data collected with gr-plasma and an Ettus X310 in an open-air test. Depending on travel costs and conference setup, an attempt will be made for a live demonstration of the gr-plasma framework using a laptop, two low-cost ultra wideband Vivaldi antennas, and an Ettus radio (either an X310 or B210). This demonstration would be arranged with the conference organizers to ensure spectrum authorization and space to demonstrate.

Talk Length 30 Minutes
Acknowledge Acknowledge In-Person

Primary author

Shane Flandermeyer (School of Electrical and Computer Engineering, Advanced Radar Research Center, University of Oklahoma)

Co-authors

Rylee Mattingly (School of Electrical and Computer Engineering, Advanced Radar Research Center, University of Oklahoma) Dr Justin Metcalf (School of Electrical and Computer Engineering, Advanced Radar Research Center, University of Oklahoma)

Presentation materials