Synchronous multi-signal acquisition for WBSNs using gold-code based joint-compressive sensing

Biomedical signals (BMSs) which produce valuable information about the condition of various physiological parameters are measured through multi-lead wired biomedical acquisition systems. In order to process real-time synchronous signals from the various parts of the human body, portable devices with...

Full description

Saved in:
Bibliographic Details
Published in:2016 IEEE Biomedical Circuits and Systems Conference (BioCAS) pp. 236 - 239
Main Authors: Rahimi, Arya Alex, Carreira, Leonardo Bosco, Gupta, Subhanshu
Format: Conference Proceeding
Language:English
Published: IEEE 01.10.2016
Subjects:
Biosignal
Compressed sensing
Computer architecture
dynamic thresholding
Electrocardiography
Gold
Gold Code
Hardware
Joint Compressive Sensing
multi-domain
multi-domain CS
multi-lead
multi-modal
secure random number generator
Sensors
Sparse matrices
synchronous multi-channel ECG
WBSN
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Biomedical signals (BMSs) which produce valuable information about the condition of various physiological parameters are measured through multi-lead wired biomedical acquisition systems. In order to process real-time synchronous signals from the various parts of the human body, portable devices with multi-channel bio-potential acquisition are needed. We present a multi-channel compressive sensing (CS) technique to enable synchronous acquisition of multi-channel bio-signals for Wearable Body Sensor Networks (WBSNs). The proposed technique exploits inter- and intra-channel sparsities inherent in a multi-channel ensemble to enable higher power savings compared to traditional CS. Simultaneous single-pass reconstruction of multiple signals that are sparse in time and frequency domain are shown using the same front-end hardware and minimal reconstruction overhead. The analysis of coding techniques for the sensing matrix is presented with Gold Codes, which are shown as the best option for a multi-channel WBSN framework achieving near zero cross-correlation between each column of the sensing matrix as well as each input channel. For a 4X compression, the proposed Joint-Compressive Sensing (Joint-CS) technique is shown to require 4% less thresholding with 30 dB better SNR than conventional CS with Bernoulli codes. In addition, an on-chip Gold-Code seed initialization circuit is demonstrated that further strengthens the SNR gains as compared to conventional pseudo-random number based seed initialization, demonstrating a stand-alone Joint-CS implementation.
DOI:10.1109/BioCAS.2016.7833775