Abstract
This article describes the main features in the implementation of a Multichannel Continuous Wavelet Transform (MCWT) on the Nvidia Jetson Single Board Computers (SBCs) using a General-Purpose Graphical Processing Unit (GPGPU). Two MCWT implementations are considered. The first implementation of an algorithm computes CWT in time-domain using convolution. The second implementation computes CWT in frequency-domain using Fast Fourier Transform (FFT). Algorithms implemented in C++ programming language using CUDA framework. Both methods are compared with each other on different hardware with various configurations. For GPGPU hardware SBCs Nvidia Jetson Nano and Nvidia Jetson TX2 were used. Also, SBCs Raspberry Pi 2 and Pi 3 were used for comparison with GPGPUs. Implemented MCWT intended to operate in real-time for biomedical signal processing.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
