Computations of cross-correlation functions on a single board Raspberry Pi computer

V. A. Faerman, M. P. Shvetsov, A. V. Tsavnin

Research output: Contribution to journalConference articlepeer-review

1 Citation (Scopus)


The paper discusses the implementation of correlation algorithm for time delay estimation on a Raspberry Pi single-board computer. The implemented correlation algorithm is based on Fourier transform. In the course of the study, we applied two alternative solutions for the software implementation of discrete Fourier transform. The first solution stands on FFTW library and uses general-purpose quad-core ARM Cortex A53 processing unit. The alternative method uses VideoCore IV graphic processing unit and is implemented via firmware GPU_FFT library. We have performed a computational experiment on a Raspberry Pi 3B to determine which solution is more preferable for the implementation of correlator. After a comparative study we figured out that estimated processing time is highly dependent on computations parameters and input signals. For small FFT window sizes CPU is proved to be a preferable option. However, for large FFT windows GPU allows significantly accelerating the computations. At some point, you can achieve even better performance by using batching and GPU for direct FFT and CPU for inverse FFT. According with the results, we have concluded that both alternatives have their own potential advantages and particular drawback. We also establish, that Raspberry Pi 3 B computer with HiFiberry extension can be used as a real-time correlator for audio signals.

Original languageEnglish
Article number012004
JournalJournal of Physics: Conference Series
Issue number1
Publication statusPublished - 25 Aug 2020
Event10th International Conference on High-Performance Computing Systems and Technologies in Scientific Research, Automation of Control and Production, HPCST 2020 - Barnaul, Russian Federation
Duration: 15 May 202016 May 2020

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Fingerprint Dive into the research topics of 'Computations of cross-correlation functions on a single board Raspberry Pi computer'. Together they form a unique fingerprint.

Cite this