youais / gr-turboseti

turboSETI blocks and flowgraph for GNU Radio

A turboSETI block for GNU Radio

This is a GNU Radio Out of Tree (OOT) module for the findET blocks. The aim is to create a GNU Radio block that can perform turboSETI analysis on a numpy.float32 data matrix stored in RAM.

The module currently consists of 3 blocks:

• findET
• findET Buffer
• findET Sink

The findET Sink block seeks to combine the functions of both findET Buffer and findET. For a good example of the intended output of findET Sink, refer to turboseti_multiprocessing_test.py.

This code has been tested on a Linux installation of GNU Radio version 3.8.

Dependencies

• turboseti-stream
• turboSETI
• blimpy
• GNU Radio version 3.8
• gr-ata OOT module (not necessary for this module, but needed to run ATA observations)

Installing gr-turboseti

git clone https://github.com/youais/gr-turboseti.git
cd gr-turboseti
mkdir build
cd build
cmake ../
sudo make install
sudo ldconfig #for Linux users

You may need to specify the install path when running cmake.

Navigating this Github Repo

• /python -- Python source code for blocks
• /grc -- .yml code for blocks' appearance in GNU Radio Companion (GRC)
• /lib -- C++ source code for blocks (not used in this module)
• /examples -- Examples of flowgraph and turboseti_stream output, as well as observation planning code

Background

This project was undertaken as part of the Berkeley SETI Research Center 2021 summer REU. My mentors were Dr. Wael Farah (SETI Institute), and Dr. Steve Croft (Breakthrough Listen, UC Berkeley).

The aim of my project is to develop a SETI data processing pipeline for the Allen Telescope Array (ATA), using GNU Radio. The ATA is a radio interferometer operated by the SETI Institute at the Hat Creek Radio Observatory in California, and consists of 42 fully-steerable antennae, each 6.1m in diameter. Its main science goal is to perform searches for technosignatures, which appear as narrowband signals 'drifting' in frequency.

Currently, the existing data-processing pipeline for the ATA uses custom hardware unavailable to those not on-site. GNU Radio is a free open-source software for developing signal-processing routines, and is used by a large community of amateur radio astronomers and enthusiasts. The implementation of a GNU Radio SETI pipeline will make the search for extraterrestrial intelligence more accesible to smaller radio observatories and citizen scientists.

Pipeline Details

The GNU Radio SETI pipeline is outlined as follows:

1. Radio telescope data from the ATA streams in through a USRP source
2. The data is 'channelised' through a polyphase filterbank (PFB), followed by a Fast Fourier Transform (FFT). This creates a high-resolution spectral product on the order of ~1MHz
3. This product accumulates in findET Buffer for ~60s, to create a data matrix of shape (60, 1e6)
4. The data matrix is then passed to findET, which uses an adapted version of turboSETI (i.e. turboseti_stream) to analyses it for potential technosignatures

Example flowgraph (refer to examples folder for .grc file):

Outcome (as of September 19, 2021)

Working:

• findET block
• findET Buffer block
• 1 channel USRP/File Source flowgraph up to 1 MHz sample rate
• 4 channel USRP/File Source flowgraph up to 4 MHz sample rate

Issues:

• findET Sink block -- current issue: TypeError: cannot pickle 'SwigPyObject' object

Future Work

1. Automate plotting of dynamic spectra of hits (adapt turboSETI's find_event_pipeline and plot_event_pipeline functions)
2. Increase maximum sample rate at which the flowgraphs can run (using multiprocessing?)
3. Observe known technosignature source (e.g. Chang'e 5) with the ATA using the GNU Radio SETI pipeline
4. Turn gr-turboseti into PyPi package
5. Begin ATA observations of interesting stars using the GNU Radio SETI pipeline

I plan to continue working on this project into the academic year.

Acknowledgements

Richard Elkins and Luigi Cruz did a significant amount of work on developing turboseti_stream. Luigi also helped greatly with the structure of the flowgraph, particularly the polyphase filterbank and FFT components. Daniel Estévez and Derek Kozel answered many, many questions about GNU Radio, OOT modules, and using Python multiprocessing. Lastly, Wael Farah was very patient while helping me wade through hordes upon hordes of bugs.

Thank you all!

This project was made possible by funding from Breakthrough Listen.

References

Theory:

• Sheikh et al. 2019, Choosing a Maximum Drift Rate in a SETI Search: Astrophysical Considerations

GNU Radio:

• GNU Radio OOT Modules
• Types of GNU Radio Blocks
• Guided GNU Radio Tutorial for Python Blocks (depreciated for v3.8, but still helpful)
• Using GNU Radio at the ATA