The Internet of Medical Things (IoMT) paradigm is becoming mainstream in multiple clinical trials and healthcare procedures. Thanks to innovative technologies, latest-generation communication networks, and state-of-the-art portable devices, IoTM opens up new scenarios for data collection and continuous patient monitoring. Two very important aspects should be considered to make the most of this paradigm. For the first aspect, moving the processing task from the cloud to the edge leads to several advantages, such as responsiveness, portability, scalability, and reliability of the sensor node. For the second aspect, in order to increase the accuracy of the system, state-of-the-art cognitive algorithms based on artificial intelligence and deep learning must be integrated. Sensory nodes often need to be battery powered and need to remain active for a long time without a different power source. Therefore, one of the challenges to be addressed during the design and development of IoMT devices concerns energy optimization. Our work proposes an implementation of cognitive data analysis based on deep learning techniques on resource-constrained computing platform. To handle power efficiency, we introduced a component called Adaptive runtime Manager (ADAM). This component takes care of reconfiguring the hardware and software of the device dynamically during the execution, in order to better adapt it to the workload and the required operating mode. To test the high computational load on a multi-core system, the Orlando prototype board by STMicroelectronics, cognitive analysis of Electrocardiogram (ECG) traces have been adopted, considering single-channel and six-channel simultaneous cases. Experimental results show that by managing the sensory node configuration at runtime, energy savings of at least 15% can be achieved.

Runtime adaptive iomt node on multi-core processor platform

Scrugli M. A.;Meloni P.
;
Sau C.;Raffo L.
2021-01-01

Abstract

The Internet of Medical Things (IoMT) paradigm is becoming mainstream in multiple clinical trials and healthcare procedures. Thanks to innovative technologies, latest-generation communication networks, and state-of-the-art portable devices, IoTM opens up new scenarios for data collection and continuous patient monitoring. Two very important aspects should be considered to make the most of this paradigm. For the first aspect, moving the processing task from the cloud to the edge leads to several advantages, such as responsiveness, portability, scalability, and reliability of the sensor node. For the second aspect, in order to increase the accuracy of the system, state-of-the-art cognitive algorithms based on artificial intelligence and deep learning must be integrated. Sensory nodes often need to be battery powered and need to remain active for a long time without a different power source. Therefore, one of the challenges to be addressed during the design and development of IoMT devices concerns energy optimization. Our work proposes an implementation of cognitive data analysis based on deep learning techniques on resource-constrained computing platform. To handle power efficiency, we introduced a component called Adaptive runtime Manager (ADAM). This component takes care of reconfiguring the hardware and software of the device dynamically during the execution, in order to better adapt it to the workload and the required operating mode. To test the high computational load on a multi-core system, the Orlando prototype board by STMicroelectronics, cognitive analysis of Electrocardiogram (ECG) traces have been adopted, considering single-channel and six-channel simultaneous cases. Experimental results show that by managing the sensory node configuration at runtime, energy savings of at least 15% can be achieved.
Adaptive system; Health information management; Internet of Things; Low-power electronics; Multi-core processing; Neural networks; Remote sensing; Runtime; Wearable sensors
File in questo prodotto:
File Dimensione Formato  
electronics-10-02572.pdf

accesso aperto

Descrizione: articolo online
Tipologia: versione editoriale
Dimensione 980.81 kB
Formato Adobe PDF
980.81 kB Adobe PDF Visualizza/Apri

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/11584/345337
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact