Page 9 - 2025S

2 UEC Int’l Mini-Conference No.54

Using ICA Algorithm to Separate Respiration Signals From Different Persons

Based on CW Radars System

Ta Hoai Nam*, Nguyen Huu Son, Kurosawa, Guanghao Sun
Department of Mechanical and Intelligent Systems Engineering
The University of Electro-Communications, Tokyo, Japan
Keywords: Non-contact sensors, vital signs, 24GHz radar,

1. Introduction
Human vital signs detection has attracted more attention in daily life.
Human vital signs detection consists of contact detection and non-
contact detection. CW radar has been widely employed on human
vital signs detection since the short-range radar was first introduced for
healthcare application in the 1970s. Vital sign monitoring using
continuous wave (CW) microwave Doppler radar is gaining attention Figure 2. Priciple of ICA algorithm
due to its simpler.
The aim of blind source separation (BSS) is to separate individual
2. Objectives
source signals from microphone array inputs without prior

information about the sources and the mixing methodology. ICA is

known to be a special case of blind source separation. ICA is capable

of finding the underlying components and sources mixed in observed

“noisy” data.

4. Experimental Setup

The simulation experiment is set up as follows

• Use 2 moving plates with frequencies of 0,3Hz and 0,5Hz
respectively to simulate human breathing, 2 devices are placed
Figure 1. Monitoring vital signs of human
30cm apart.
Most of the reported results ocused on obtaining non-contact • Use 2 radars 24GHz and 10 GHz placed next to each other
respiration measurements for a single subject. In many practical and 30cm apart to measure the simulation signal. The radars are
placed 1 m apart from the 2 moving plates.
applications, the received signal can easily be affected when there is
• Use ADC with sampling frequency of 1KHz and use Labview
more than one subject present in the immediate neighborhood, software on computer to collect data from radar.
especially in a home-based, long-term monitoring application where • Use Matlab to process collected data.
the subjects share a bed with the partner, childrend and maybe their
pets.
3. Method
This situation needs to be taken into account so that the system
consisting of multiple Doppler radars can collect the respiratory signals
of multiple subjects in the room and use signal processing algorithms
to accurately separate them from each other.GECC
The CW multi-Radar system with 24GHz frequency is proposed to
be used because of its low cost and simple operating principle

combined with Independent Component Analysis (ICA) algorithm.
This is a quite popular algorithm and widely used in audio signal Figure 3. Setup Simulation Experiment
processing, applied to the blind source separation problem.

*The author is supported by (GECHA) MEXT Scholarship