Patient monitoring requires the acquisition of increasingly larger amounts of biosignal data that needs to be managed and transferred with minimum energy consumption. As huge quantities of data are often very redundant, it is possible to reduce their size directly on the edge of the system, right after the acquisition. To do so, subspace analysis can be considered a fundamental tool that can be used to significantly reduce the size of high-dimensional data, thus minimizing the requirements for data transfer. The problem of these methods is that they often come with big memory and computation requirements, as they are ultimately equivalent to the expensive eigenspace evaluation. In order to use subspace analysis methods with the minimum requirements in terms of cost and energy consumption, we here rely on two specialized streaming algorithms for the estimation of the subspace of electroencephalogram (EEG) signals directly after the acquisition on edge devices. The implementation of these state-of-the-art algorithms is tested on a common low-end microcontroller unit (MCU), which is an ideal candidate as edge computing digital hardware platform. The functional performance of these methods is evaluated along with the requirements in terms of computational time, energy consumption and memory footprint.

An MCU Implementation of PCA/PSA Streaming Algorithms for EEG Features Extraction

Marchioni, Alex;Mangia, Mauro;Pareschi, Fabio;Rovatti, Riccardo;Setti, Gianluca
2021

Abstract

Patient monitoring requires the acquisition of increasingly larger amounts of biosignal data that needs to be managed and transferred with minimum energy consumption. As huge quantities of data are often very redundant, it is possible to reduce their size directly on the edge of the system, right after the acquisition. To do so, subspace analysis can be considered a fundamental tool that can be used to significantly reduce the size of high-dimensional data, thus minimizing the requirements for data transfer. The problem of these methods is that they often come with big memory and computation requirements, as they are ultimately equivalent to the expensive eigenspace evaluation. In order to use subspace analysis methods with the minimum requirements in terms of cost and energy consumption, we here rely on two specialized streaming algorithms for the estimation of the subspace of electroencephalogram (EEG) signals directly after the acquisition on edge devices. The implementation of these state-of-the-art algorithms is tested on a common low-end microcontroller unit (MCU), which is an ideal candidate as edge computing digital hardware platform. The functional performance of these methods is evaluated along with the requirements in terms of computational time, energy consumption and memory footprint.
In Proceedings - 2021 IEEE Biomedical Circuits and Systems Conference (BioCAS)
01
05
Prono, Luciano; Marchioni, Alex; Mangia, Mauro; Pareschi, Fabio; Rovatti, Riccardo; Setti, Gianluca
File in questo prodotto:
Eventuali allegati, non sono esposti

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/852253
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact