Instrumentation & Measurement Magazine 24-5 - 43

substrate are used to guarantee the high flexibility of the whole
system. The substrate is made of polymeric materials such
as Polydimethylsiloxane (PDMS), which has its Young's
modulus in the order of a few MPa, leading to a mechanical
compatibility with the POF and embedded structure. As the
LEDs are activated and deactivated in a predefined sequence,
in an on-off keying modulation, there is no simultaneous activation
of multiple LEDs, and the region at which the LED
is positioned on the fiber can be regarded as a measurement
point. As the light source is positioned perpendicular to the fiber,
only the disturbances (thermal, mechanical or both) close
to the light source is detected when this light source is active.
For example, considering Fig. 1, if the disturbance is applied
on the region of LED 1, a significant optical power variation
occurs only when LED 1 is active, i.e., when LEDs 2 to n are
active, there is no significant optical power variation. Thus,
the technique proposed in [16] allows the development of a
quasi-distributed sensing array using the intensity variation
principle, which has the advantages of lower cost, higher portability
and simplicity on signal acquisition and processing
when compared with other quasi-distributed sensing approaches.
Such a system can perform a multipoint assessment
of a single parameter or the multiparameter assessment, where
each sensor can be subjected to the variation of one parameter
such as angle, force and temperature.
The system's scalability also enables using a higher number
of sensors in the same fiber with a few centimeters of distance
between the sensors (or perpendicularly positioned LEDs).
Actually, the distance between LEDs has to be smaller than
the area covered by the LED's light, which depends on the distance
between LED and POF as well as the LED characteristics.
Thus, for arrays with high spatial resolution, i.e., sensors with
sub-centimeter distance between each other, the LED coupling
as well as sensor structure need to be optimized to avoid
crosstalk between sensors. It is also worth noting that the LED
acquisition frequency also plays an important role on the dynamics
of the measured signal. As the acquisition frequency of
each LED can be higher than 50 Hz (i.e., each LED is activated
50 times per second), the system is capable of detecting the dynamics
of human movement.
Another limitation of the technique is the maximum number
of sensors that can be multiplexed. Since there is a lateral
section on the fiber to expose a limited part of its core, there
is a higher power attenuation in this region. Thus, if multiple
lateral sections are performed, there is an even higher signal
attenuation. However, it is worth noting that systems with 15
multiplexed sensors were achieved using light sources with
optical power lower than 1.5 mW [17]. If a higher number of
sensors is needed, light sources with higher optical powers
can be used.
Optical Fiber-embedded Multiplexed
Sensor System Applications
The multiplexing technique for intensity variation sensors
enables many applications in movement analysis. There is
the possibility for POF-embedded systems in 3D printed
August 2021
structures as well as in textiles for wearable and non-wearable
applications. In addition, the capabilities of multiparameter
and multipoint sensing are investigated, where the multipoint
sensing is analyzed in two scenarios for angle variations: a system
with 3 degrees of freedom (DoF) with angle variation in
a single plane and a system with angle variations in multiple
planes, where the orientation angles (roll, pitch and yaw) are
analyzed.
Fig. 2 presents the experimental setup used in the multiparameter
application, which is further depicted in [16]. In this
case, there is a three sensor array in which the positions are
subjected to variations in transverse force, temperature and
angle, and the results in time-domain show a small crosstalk
between sensors when the acquired optical power is analyzed.
Therefore, the angle, force and temperature can be estimated
from the optical power variation following their characterizations
and sensitivities.
The multiparameter sensing results also indicate the capability
of the multiplexing technique to detect multiple events
in sensors with different sensitivities. As discussed in [16],
the root mean squared errors (RMSE) are 2.87° for angle assessment
(resulting in 3.18% considering the whole angular
range), 3.47 N for the transverse force assessment and 0.75 °C
for temperature measurement. The force and temperature estimations
result in relative errors of 5.78% and 2.50% for force
and temperature, respectively, when the whole range of variation
on each parameter is considered. These low errors indicate
not only the feasibility of the system presented in [16], but also
its reliability, which motivate the evaluation of the system for
multipoint assessment of a single parameter. Since human
movement includes angular movements in multiple planes
and with a multitude of DoFs, both capabilities, i.e., the multiple
DoFs in a single plane and angular movements in multiple
planes, are explored [16], [18].
For the angle assessment in different movement planes,
the tests were performed with the angular movement combination
in different orientation planes, i.e., combination of
different roll-pitch-yaw movements (Fig. 3). As the sensors are
positioned with different orientations, it is possible to distinguish
the movements of each pair of planes by analyzing the
responses of all three sensors of the array. As shown on the left
side of Fig. 3, the system has three motors, one for each rotation
plane, where the movement with two planes is obtained by activating
two motors at the same time. The movements on yaw
plane occurred from 0 to 50° in 10° steps and from 0 to 30° in 5°
steps on roll plane, whereas the movement in pitch plane occurred
from 0 to 20° in 5° steps. First, a calibration is performed
on single plane angular movements, i.e., angle variations only
on one plane (roll, then pitch and, finally, yaw planes), to obtain
the sensor sensitivity with respect to each plane. Then, the
tests are performed on the combination between two planes,
where the responses of all three sensors are obtained and the
angles of each plane are estimated from the sensitivities of each
plane obtained in the single plane characterization. The insets
in Fig. 3 show the estimated angle (in y-axis) as a function
of the reference angle (obtained from the motors positions)
IEEE Instrumentation & Measurement Magazine
43
Instrumentation & Measurement Magazine 24-5

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 24-5
