Multiplexing technique for quasi-distributed sensors arrays in polymer optical fiber intensity variation-based sensors

Optics and Laser Technology 111 (2019) 81–88

Contents lists available at ScienceDirect

Optics and Laser Technology journal homepage: www.elsevier.com/locate/optlastec

Full length article

Multiplexing technique for quasi-distributed sensors arrays in polymer optical fiber intensity variation-based sensors

T

⁎

Arnaldo G. Leal-Juniora, , Camilo R. Díaza, Carlos Marquesb, Maria José Pontesa, Anselmo Frizeraa a

Telecommunications Laboratory (LABTEL), Electrical Engineering Department, Federal University of Espírito Santo, Fernando Ferrari avenue, 29075-910 Vitória, ES, Brazil b Instituto de Telecomunicações and Physics Department & I3N, Universidade de Aveiro, Campus Universitário de Santiago, 3810-193 Aveiro, Portugal

H I GH L IG H T S

technique for intensity variation-based sensors. • Multiplexing approach for the analysis of maximum number of sensors and sensitivities. • Analytical validation for 3-DOF angle measurement using three-sensor array. • Experimental • Technique validation for quasi-distributed array in multi-parameter applications.

A R T I C LE I N FO

A B S T R A C T

Keywords: Optical fiber sensors Polymer optical fiber Quasi-distributed sensors Multi-parameter

This paper presents a multiplexing technique for polymer optical fiber (POF) intensity variation-based sensors. The technique relies on the side-coupling between the light source and the POF lateral section. Thus, each sensor has its own light source which has its activation as well as its signal acquisition (made with two photodetectors, one at each end of the POF) controlled by a microcontroller. With this technique, a matrix with the number of columns equal to the number of photodetectors multiplied by the number of light sources (or sensors) is obtained. This enables the decoupling of the response of each sensor. The presented analytical approach shows a tradeoff between the number of sensors and their sensitivities (considering the dynamic range of each sensor). In addition, experimental results show the feasibility of the technique to measure angles in a 3°-of-freedom (DOF) systems with errors as low as 3°. Furthermore, the proposed approach was also tested in multi-parameter applications, where temperature, angle and force were estimated with errors up to 5% with an array with 3 POF sensors. The results presented in this paper can pave the way for novel applications of quasi-distributed sensors with intensity variation-based sensors both in multi-DOF systems and in multi-parameter applications with the additional advantages of lower cost than fiber Bragg gratings-based systems and lower spatial resolution than distributed sensors.

1. Introduction The development of optical fiber-based sensor systems has led to the widespread of polymer optical fibers (POFs), which were previously employed in short communications systems [1]. In sensor applications, POFs present important advantages over their silica counterparts, which include higher flexibility, fracture toughness and higher strain limits. In addition, POFs have lower Young’s modulus that results in stress, force and pressure sensors with higher sensitivity [2]. Similarly,

POFs present higher thermo-optic coefficient than silica fibers, resulting in higher sensitivity for temperature sensors [2]. The aforementioned advantages enable the development of POF sensors for different parameters, which include liquid level [3], temperature [4], angle [5], humidity [6], biochemical analytes [7], pressure [8], acceleration [9], refractive index [10] and force [11]. The lower processing temperatures of polymers has enabled the development of fibers with different materials, each one with their own advantages and disadvantages. Thus, POFs are fabricated with humidity

⁎

Corresponding author. E-mail addresses: [email protected] (A.G. Leal-Junior), [email protected] (C.R. Díaz), [email protected] (C. Marques), [email protected] (M.J. Pontes), [email protected] (A. Frizera). https://doi.org/10.1016/j.optlastec.2018.09.044 Received 6 August 2018; Received in revised form 14 September 2018; Accepted 18 September 2018 0030-3992/ © 2018 Elsevier Ltd. All rights reserved.

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

in a POF sensors with lateral section similar to the one presented in [38]. However, in this case, a control system acquires the signal of each photodetector when each LED is active, which result in a matrix with P × D vectors, where D is the number of LEDs and P the number of photodetectors. In this way, it is possible to decouple the effects of different parameters in each lateral section of the fiber. The sensor is experimentally validated for 3 degree of freedom (DOF) applications in the simultaneous measurement of temperature, force and angle. The experimental validation also includes two different scenarios: the response of three sensors for the same parameter and with similar sensitivity; and different sensors for different parameters (temperature, force and angle). Since the lack of multiplexing capabilities is regarded as an important issue in intensity variation-based sensors, the technique proposed in this paper enables the advancement of both quasi-distributed sensor techniques and intensity variation sensors. Furthermore, comparing with the conventional technologies for distributed (or quasi-distributed) sensing architectures such as OFDR, backscatter reflectometry, microwave photonics, FBG, the proposed multiplexing technique presents much lower cost and simplicity on the signal processing. In addition, it presents potentially higher spatial resolution than OFDR and OTDR techniques, since the resolution in the proposed technique is limited to the physical distance between the fibers lateral sections.

insensitive cyclic olefin copolymers (COC) such as TOPAS 5013 [12] and Zeonex 480R [13]; low-loss polymers that include cyclic transparent amorphous fluoropolymers (CYTOPs) [14]; polycarbonate (PC), which presents higher strain limits [15]. However, the most employed polymer in POF fabrication (to date) is polymethyl methacrylate (PMMA) [16]. In addition, there are microstructured POFs that present few-mode or single mode operation depending on the geometry of the microstructures [17], which is an advantageous feature for grating inscription in fibers [18]. Solid core POFs, especially the ones with larger diameters and multimode operation, have the advantage of simpler fabrication [19], which lead to lower cost fibers. Furthermore, POFs with larger diameters enable the application of low-cost plastic connector that generally results in lower cost systems [2]. As a disadvantage, gratings inscribed in those fibers generally present multipeak spectra, which inhibit their application in sensors arrays and some sensors applications [20]. For this reason, different techniques are needed for the grating inscription in multimode fiber, such as the one presented in [20]. Different sensing techniques with POFs have been proposed throughout the years, such techniques are based on interferometers [21], long period gratings [22], fiber Bragg gratings (FBGs) [18], evanescent wave [23], intensity variation [24] and nonlinear effects [25]. Among those techniques, intensity variation based sensors are the ones that present the lowest cost, highest simplicity in fabrication, since it does not need specialized equipment such as the gratings-based sensors and also have simplicity in the signal processing [26]. Aiming at a low-cost sensor system with the advantages of POFs for sensor applications, intensity variation-based sensors with multimode POFs have been used in movement analysis [27], robots instrumentation [28], industrial applications [29] and structural health monitoring [30]. However, intensity variation-based sensors have the disadvantage of errors due to the light source power deviation, which can be reduced by the use of self-referencing techniques [31,32]. Another important disadvantage is their lack of multiplexing capabilities, which inhibit the application in multipoint [33] or multiparameter [34] measurements. In order to use more than one sensor, it is necessary another fiber, reducing the system compactness and making the system less flexible, which can influence the natural pattern of movement in movement analysis applications [35]. In addition, a less flexible system harms its application in soft robotics, which is an emerging technology for wearable systems and actuators [36]. Regarding the techniques for distributed sensors, optical time-domain reflectometry (OTDR) and optical frequency-domain reflectometry (OFDR) are commonly employed [37]. However, such techniques have issues related to their spatial resolution of some meters. Although a resolution of centimeter level can be achieved with distributed sensors based on Brillouin analysis, these interrogation techniques generally present much higher cost. In addition, OFDR generally requires bulk hardware that includes components like sweptlaser interferometer and microwave photonic circuit. Furthermore, quasi-distributed sensors system based on FBG arrays can have higher resolution, since it is only limited by the physical separation between FBGs, these arrays still have aforementioned disadvantages of high interrogation cost and, for FBGs, there is also the necessity of specialized equipment for the grating inscription [18]. A cost-effective sensor quasi-distributed sensor system for liquid leakage detection is presented in [38], where a flexible lamp belt with light emitting diodes (LEDs) is side-coupled to POF with lateral sections. Each end facet of the fiber is connected to a power meter and the forward and backward optical power are compared. However, the system was applied only on the discrete detection of water, i.e., it only shows if there is water or air on each of the detection points [38]. In order to tackle the limitations of the previously proposed distributed and quasi-distributed techniques for optical fiber sensors, this paper presents a multiplexing technique for intensity variation-based sensors. The technique is based on the side-coupling of the light source

2. Operation principle and experimental setup The schematic representation of the proposed technique is presented in Fig. 1. Two photodetectors are positioned on each end facet of the fiber. In addition, each lateral section on the fiber presented in Fig. 1 is one sensor, where each sensor has a side-coupled LED. A microcontroller is responsible for the activation of the LED array, which are activated with a predefined sequence and frequency in a way that each LED is activated at a time, i.e., there is not the simultaneous activation of two or more light sources. Furthermore, a microcontroller is responsible for the signal acquisition of both photodetectors when each LED is activated. Thus, a matrix is obtained in which each column represents the POF power acquired by one photodetector when a predefined LED is activated. This matrix is presented in Fig. 1, where the number of columns is 2n (two photodetectors with n LEDs) and the number of lines is the number of signal acquisitions. By analyzing and comparing each vector of the matrix, it is possible to obtain the response of n sensors without the influence of light source deviations. One limitation of the technique is the maximum number of sensors that can be multiplexed. Since there is a power attenuation at each sensor’s lateral section, if the fiber presents a high number of sensors, the output power may be too low to be detected. In addition, the sensors sensitivities also lead to a power reduction, which need to be

Fig. 1. Schematic representation of the proposed multiplexing technique for intensity variation-based sensors. 82

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

considered in the design of the sensor arrays. However, the initial power attenuation due to lateral section and sensor sensitivity depends on the lateral section parameters (length and depth) of each optical fiber sensor [39,40]. In order to verify the maximum number of sensors that can be multiplexed, an analytical approach based on the model presented in [41] is performed. Since the sensors work with the power attenuation principle, the sensors sensitivities and dynamic range also are considered in the POF output power estimation. Taking these parameters into account, the power attenuation with respect to all aforementioned effects is estimated as follows:

P0 (S −NS0) = c −αL−Σsensi ri Pin Sc

(1)

Referring to Eq. (1), N is the number of sensors, α is the fiber attenuation coefficient, sensi and ri are the sensitivity and dynamic range of each sensor, respectively. After the investigation of the number of sensors and their sensitivities, the next analysis is made to estimate each sensor response of the array. In this case, we consider a fiber with 3 sensors in two scenarios: (i) sensors with the same sensitivity and dynamic range; (ii) sensors with different sensitivities and dynamic ranges. Since the fiber attenuation coefficient and the loss due to the lateral section only lead to an offset in the sensor response, only the last term on the right-hand side of Eq. (1) is considered. Thus, the equation used in the sensors simulation is:

⎛ P0 ⎞ = ⎝ Pin ⎠i, j ⎜

⎟

n

∑ sensi ri i=1

(2)

where i is the sensor (i = 1, 2, 3 in this case) and j is the photodetector (j = 1, 2). Eq. (2) shows that the response of each sensor is comprised of the sum of the other sensors, i.e., the power of Sensor 3 is the sum of the responses of Sensors 1, 2 and 3, whereas Sensor 2 is the sum of 1 and 2. Therefore, in order to obtain the response of a single sensor without the influence of other sensors, it is necessary to compensate the response of the other sensors. In this case, Eq. (2) is rewritten as:

ri =

Fig. 2. Experimental setup for (a) multiplexing technique validation with the array of 3 curvature sensors. (b) Validation for a multi-parameter application.

sensor performance for multi-parameter sensing. Thus, the fiber with three lateral sections is positioned on the setup of Fig. 2(b), where there is the angle variation through a DC motor with a closed loop position control. The angular range for the angle characterization is 0 to 90° with constant angular velocity of 0.1 rad/s. Furthermore, there is a setup for transverse force assessment in which different calibrated weights are applied on the fiber lateral section, resulting in transverse forces of 0 to 50 N. Finally, the last lateral section of the fiber is positioned in a thermoelectric Peltier plate (TEC1-12706, Hebei IT) with a temperature controller (TED 200C, Thorlabs). The temperature in this characterization was varied from 25 °C to 55 °C in 10 °C steps. After the characterization, the multi-parameter sensing capabilities of the proposed multiplexing technique is evaluated through the simultaneous variation of all three parameters within the range each sensor was characterized. It is worth to mention that Fig. 2(a) and (b) are only a schematic representation of the employed setup. The fiber used in the tests has about 50 cm length. Moreover, the distances between the photodetector 1 and the Sensors 1, 2 and 3 (LP1→S1, LP1→S2, and LP1→S3) are 1.6 cm, 7.6 cm and 12.7 cm, respectively. It is worth to mention that all the performed tests were made at controlled humidity conditions of 64% ± 2%, since the humidity can affect the sensor response [42].

( ) −( ) P0 Pin i, j

P0 Pin i − 1, j

sensi r

(3)

In this way, Eq. (3) shows the necessity of characterizing each sensor individually, prior to their applications for simultaneous measurements. In order to validate the proposed technique for intensity variationbased sensors multiplexing, the same two scenarios are applied. Thus, three lateral sections are made in a multimode POF made of PMMA (HFBR-EUS100Z, Broadcom Limited). This fiber has a core diameter of 980 µm, a cladding with 10 µm thickness and a polyethylene coating. Each lateral section has about 14 mm length and 0.1 mm depth, which are made through abrasive removal of material. In addition, a flexible lamp belt with an array of 3 LEDs each is fixed at each lateral section for the light coupling. The LEDs of the flexible lamp belt emit white light and have about 3 mm diameter each. The POF power is acquired by two phototransistors IF-D92 (Industrial Fiber Optics, USA). The activation of the LED array as well as the control of the signal acquisition are made with a FRDM-KL25Z board (Freescale, Austin, TX, USA), where both the LED activation and signal acquisition frequencies are 10 Hz. The first set of tests is made with the experimental setup presented in Fig. 2(a), where there are three rotation units with a goniometer embedded. First, each sensor is individually characterized with relation to the other sensors at 0°, where the maximum range of the sensor characterization is between 0° to 80° in 20° steps. Then, tests with simultaneous movements of all three sensors are performed to show the capability of the system of tracking the response of each sensor. Thereafter, we employ another experimental setup to evaluate the

3. Numerical analysis In the analysis of the tradeoff between number of sensors, sensors sensitivity and dynamic range, we consider the sensors with same sensitivity, lateral section parameters and dynamic range. Two 83

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

Table I Sensitivities and dynamic ranges for the sensors used in the simulation. Scenario 1

Scenario 2

Sensor

Sensitivity

Range

Sensitivity

Range

1 2 3

0.0010 0.0010 0.0010

0–50° 0–50° 0–50°

0.0020 0.0089 0.0014

0–60° 0–90 N 25–55 °C

ranges) are employed. However, in order to obtain the compensated response, the sensitivity of each sensor must be known a priori, which can be easily accomplished by characterizing each sensor without the influence of the others. Then, Eq. (3) must be applied, where the power of each sensor is subtracted by the one of its neighbors and divided by the sensitivity of each sensor. In this way, it is possible to distinguish the response of each sensor as shown in Fig. 4(c), where the blue curve shows the normalized power of each sensor and the red curve presents the sensor response for each parameter (angle, force and temperature). By analyzing the curves of Fig. 4, one can verify that it was possible to estimate the range of each sensor at each scenario. Nevertheless, it is worth to mention that factors as nonlinearities and cross-sensitivity between sensors can lead to errors in the measurement. In the next section, the feasibility of the proposed multiplexing technique will be experimentally evaluated in the same scenarios and the sensors’ responses will be compared with the ones of the reference sensor for each parameter.

Fig. 3. Analytical results for the relation between POF output power, number of sensor, sensors sensitivities and dynamic range.

approaches are performed, considering a curvature sensor: sensor sensitivity and dynamic range of 0.0001 and 50°, respectively, with the number of sensors varying from 1 to 15; and fixed number of sensors (N = 10), dynamic range of 50° and sensitivity between 0.0001 and 0.002 in steps of 0.0001. The sensor sensitivity considered in the simulations are the ones obtained in previous works [39]. In addition, the fiber length is 10 m, attenuation coefficient is 12 dB/km, fiber crosssection area (Sc) is 0.78 mm2 and area of removed material is 0.04 mm2. Fig. 3 presents the normalized power attenuation as a function of the number of sensors and the sensor normalized response at full range, defined here as sens × r for 2 and 10 sensors. The results presented in Fig. 3 show that there is a relation between the output power and the number of sensors as well as the sensors sensitivities and dynamic range. Therefore, the presented analytical approach provides guidelines for the sensors array design, where one can reduce the sensor sensitivity or dynamic range to accomplish an array with more sensors. Nevertheless, if the objective is an array with highly sensitive sensors, the maximum number of sensors can be obtained by following the presented approach. It is worth to mention that the sensor sensitivity in intensity variation-based sensors with a lateral section can be controlled by the lateral section parameters as thoroughly discussed in [39]. It is also interesting to note that the analysis presented in Fig. 3 is made with respect to only one photodetector (P1). Considering another photodetector (P2) in the configuration presented in Fig. 1, it may be possible to double the number of sensors, since one half (closer to P1) is interrogated with P1 and the other half, with P2. Then, the simulations of the sensor response under different conditions were made under two scenarios. In these simulations, we consider curvature sensors in the first scenario (equal sensitivity and range) with a sensitivity of 0.001 and range of 0 to 50° in 10° steps. The second scenario applies sensors with different sensitivity and dynamic range, where those sensitivities were obtained in previous works for angle [39], temperature [4] and transverse force analysis [43], which are summarized in Table I. Fig. 4 shows the simulated response for the sensors with the two aforementioned scenarios, where Fig. 4(a) shows the response for Scenario 1 and Fig. 4(b) for Scenario 2. The blue curves in Fig. 4(a) and (b) are the ones obtained through Eq. (2), whereas the red curves/ markers are the compensated responses of the sensors applying Eq. (3). The simulation results prove the feasibility of the proposed approach, where the response of each sensor can be compensated by applying Eq. (3) on the response of each sensor. In addition, the technique also is feasible when different sensors (with different sensitivities and

4. Results and discussion 4.1. Angle measurement with 3-DOF As the first step in the multiplexing technique application, all three sensors are individually characterized, i.e., when one sensor is under curvature the others are at straight position (without curvature). Fig. 5(a) shows the signal acquired by photodetectors 1 and 2 during a test, where each sensor is individually activated. From the results of Fig. 5(a) it is possible to see the interesting feature of the proposed technique, when one sensor is active, the highest response in the photodetector is the one acquired when the corresponding sensor is illuminated by its respective LED, i.e., when Sensor 1 is active, the highest signal variation occurs in vectors RLED1, P1 and RLED1, P2. Similarly, the highest variation for the Sensor 2 was acquired in photodetector 1 (RLED2, P1), which is the one closest to the Sensor 2, where similar feature also happens with Sensor 3. In addition, the characterization of each sensor in different intervals is depicted in Fig. 5(b), where a linear regression was made for each sensor to obtain the sensitivities and initial offset in the curves. Furthermore, Sensors 1, 2 and 3 are characterized with the angular ranges of 60°, 40° and 80°, respectively due to experimental limitations of the setup presented in Fig. 2(a). In this case, the movement of one sensor is limited, especially for Sensor 2, when the other sensors are kept at the straight position. Moreover, Sensor 3 is the one with more freedom of movement when the other sensors are positioned without bending. Regarding Fig. 5(a), it is worth to mention that, besides photodetector 1 for Sensor 1, the sensors responses are acquired by both photodetectors. The reason for the low signal variation in photodetector 1 is related to the distance between Sensor 1 and photodetector 1; and the coupling between the lateral section of this sensor and its respective LED. Nevertheless, Fig. 5(b) shows almost the same sensitivity (3.41 × 10−4 ± 4.24 × 10−5 (°)−1) for all three sensors. The next step is the application of the sensor array in angle measurement with 3-DOF. In this case, the sensors are placed on the desired positions and the signals are acquired by the photodetectors. The test is made by placing the sensors in 10 different positions as presented in Fig. 6 (as the reference signals for Sensors 1, 2 and 3) with the 84

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

Fig. 5. POF sensor array characterization for 3-DOF angle measurement (a) signal acquired by the photodetectors when each LED is active and (b) linear regression for the angle characterization with each sensor.

Fig. 4. Simulated responses for the POF sensors in two different scenarios (a) normalized power for 3 sensors with the same sensitivity and dynamic range, (b) estimated angle for 3 sensors with same sensitivity and dynamic range and (c) normalized power and estimated parameters for 3 sensors with different sensitivities and dynamic range.

Fig. 6. Sensor response in the 3-DOF validation test.

85

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

comparison between each sensor position estimation and the actual position obtained from the goniometer position. The results presented in Fig. 6 show the capability of each sensor of tracking the angle of each rotation unit (see Fig. 2(a)). Therefore, the sensor array with the proposed multiplexing technique is an interesting alternative to measure multi-DOF angles, which is a condition commonly observed in lower limb exoskeletons, where there is 3-DOFs, one for the hip joint and the others for knee and ankle joints [28]. Regarding the estimated errors, the root mean squared error (RMSE) between Sensor 1 and the reference is 2.76°, whereas the RMSE for sensors 2 and 3 are 3.22° and 2.44°, respectively. Comparing to the characterization condition, where the mean RMSE is 2.01° for the three sensors, the difference between the RMSE in the simultaneous measurement condition is similar to the one when each sensor is activated without the influence of the others. This results also show the feasibility of the proposed technique. The errors obtained may be related to the low signal amplitude obtained from photodiode 1 when Sensor 1 is active and due to influence of modal distribution in the fiber under curvature in different points. Nevertheless, these effects do not lead to high errors in the measurement, since the mean RMSE difference between the 1-DOF (condition used in the characterization) and 3-DOF is about 0.80°. 4.2. Multi-parameter application with 3 POF sensors In order to validate the sensor array for multi-parameter applications, the fiber is positioned in the experimental setup of Fig. 2(b) and the analysis is made in two steps: sensor characterization and simultaneous measurement of angle, transverse force and temperature. Thus, the first set of tests is performed with the variation of a single parameter, while the others are kept constant. The characterization of each sensor is presented in Fig. 7, where the sensitivities for angle, force and temperature are, respectively, 2.08 × 10−4 (°)−1, 6.41 × 10−5N−1 and − 1.745 × 10−5 (°C)−1. After characterizing the three sensors, simultaneous variations of angle, transverse force and temperature are performed, where the signals acquired for Sensors 1, 2 and 3 with the photodetector 1 are presented in Fig. 8(a). As also observed in the response of Fig. 5(a), the signal acquired when each LED is illuminating its respective sensor has the response of the active sensor. Thus, the response of Sensor 1 (employed for angle measurements) presents similar pattern that the one applied by the DC motor, whereas Sensor 2 shows the signal variation

Fig. 8. (a) Responses of the three sensors acquired by the photodetector 1. (b) Angle, force and temperature measurements using the proposed multiplexing technique acquired for 210 s.

when the transverse force is applied. Similarly, the temperature sensor (Sensor 3) shows the signal variation according to the applied temperature variation. However, since its sensitivity is lower than the other sensors, the power variation of the Sensor 3 is lower than the one observed to the other sensors. Fig. 8(b) shows the estimated angle, force and temperature with Sensors 1, 2 and 3, respectively. The responses are obtained with the linear regression coefficients of the sensors characterization (see Fig. 7) for the estimation of each parameter. The presented results show that the proposed multiplexing technique for the POF sensor array is capable of tracking simultaneous variation of different parameters (angle, force and temperature) in sensors with different sensitivities. Regarding the errors in the multiparameter application, the RMSE for the angle measurement is about 2.87°, which results in a relative error of 3.18% when the whole angular range is considered (about 100°). In the force measurement, the RMSE is 3.47 N with a relative error of 5.78% if the whole force range is considered, which is relatively higher than the one for the angle measurement. The reason for this behavior is the nonlinearities of the transverse force measurement, which presents the lowest linearity among the analyzed parameters. Such nonlinearities can be related to the nonlinear behavior of the stress-strain curves in polymers. Thus, the stress transversely applied in the fiber can result in a nonlinear variation of the fiber curvature. In addition, these nonlinearities lead to

Fig. 7. Experimental characterization of each sensor for multi-parameter application. 86

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

higher errors when a linear regression is made to obtain the sensor response. Regarding the errors in the temperature measurement, a RMSE of 0.75 °C was obtained in the temperature range of 55 °C to about 25 °C, which results in a relative error of 2.50%. This relative error is the lowest among the tested parameters. Such lower error can be related to the fact that the fiber fixed in both ends of the Peltier plate, keeping the fiber always in the straight position and without movement, which reduces the influence of modal distribution in the sensor response. In addition, temperature characterization has shown the highest linearity among the ones tested. It also contributes to lower errors in the linear regression of the sensor response for temperature estimation. In summary, the results presented in this section validate the multiplexing technique in different experimental conditions. In this way, a quasi-distributed sensor system is obtained, where the sensor array presents much lower cost than quasi-distributed sensor arrays based on FBGs [3]. Moreover, the sensor spatial resolution is only limited by the LEDs irradiance, since each lateral section needs to be illuminated by only one LED. For this reason, one can assume that the spatial resolution, i.e., the minimum distance between two sensors can be similar to the ones of sensor arrays based on FBGs.

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

5. Conclusions [15]

This paper has presented a multiplexing technique for intensity variation-based sensors in which the light source is side-coupled to the lateral section of the fiber. Since each lateral section is an intensity variation-based sensor, each sensor has its own light source. Then, two photodetectors are positioned at each end facet of the fiber and a microcontroller is responsible for the LED activation frequency and the signal acquisition when each LED is active. The presented analytical approach gives guidelines for the design of the sensor array, whereas the experimental results show the feasibility of the proposed technique for 3-DOF angle measurement and multi-parameter applications. Therefore, it is possible to obtain a cost-effective quasi-distributed sensor array to measure different parameters with relative errors up to 5%. In addition, by analyzing the tradeoff between the number of sensors and their sensitivity (controlled by the sensors’ lateral section length and depth) and dynamic range for each given application, it may be possible to scale the sensor array to large sensors networks, which will be further explored in future works. The investigation of the proposed technique in exoskeleton instrumentation, wearable sensors and other multi-parameter applications will be performed in future works.

[16]

[17] [18] [19]

[20]

[21] [22]

[23] [24]

Acknowledgments [25]

This research is financed by CAPES (88887.095626/2015-01), FAPES (72982608 and 80709036), CNPq (304192/2016-3 and 310310/2015-6). C. Marques acknowledges the financial support from FCT through the fellowship SFRH/BPD/109458/2015, program UID/ EEA/50008/2013 by the National Funds through the Fundação para a Ciência e a Tecnologia/Ministério da Educação e Ciência, and the European Regional Development Fund, Portugal under the PT2020 Partnership Agreement.

[26] [27]

[28]

[29]

References

[30]

[1] O. Ziemann, J. Krauser, P.E. Zamzow, W. Daum, POF Handbook, Springer, Berlin Heidelberg, Berlin, Heidelberg, 2008. [2] K. Peters, Polymer optical fiber sensors—a review, Smart Mater. Struct. 20 (1) (2010) 13002. [3] C.A.F. Marques, G.-D. Peng, D.J. Webb, Highly sensitive liquid level monitoring system utilizing polymer fiber Bragg gratings, Opt. Express 23 (5) (2015) 6058–6072. [4] A. Leal-Junior, A. Frizera-Neto, C. Marques, M. Pontes, A polymer optical fiber temperature sensor based on material features, Sensors 18 (2) (Jan. 2018) 301. [5] A. Leal Junior, A. Frizera Neto, M.J. Pontes, T.R. Botelho, Hysteresis compensation

[31]

[32]

[33]

87

technique applied to polymer optical fiber curvature sensor for lower limb exoskeletons, Meas. Sci. Technol. (2017). G. Rajan, Y.M. Noor, B. Liu, E. Ambikairaja, D.J. Webb, G.D. Peng, A fast response intrinsic humidity sensor based on an etched singlemode polymer fiber Bragg grating, Sens. Actuat., A Phys. 203 (2013) 107–111. N. Zhong, M. Zhao, L. Zhong, Q. Liao, X. Zhu, B. Luo, Y. Li, A high-sensitivity fiberoptic evanescent wave sensor with a three-layer structure composed of Canada balsam doped with GeO2, Biosens. Bioelectron. 85 (2016) 876–882. R. Ishikawa, H. Lee, A. Lacraz, A. Theodosiou, K. Kalli, Y. Mizuno, K. Nakamura, Pressure dependence of fiber Bragg grating inscribed in perfluorinated polymer fiber, IEEE Photonics Technol. Lett. 29 (24) (Dec. 2017) 2167–2170. A. Stefani, S. Andresen, W. Yuan, N. Herholdt-Rasmussen, O. Bang, High sensitivity polymer optical fiber-Bragg-grating-based accelerometer, IEEE Photon. Technol. Lett. 24 (9) (2012) 763–765. A.R. Prado, A.G. Leal-Junior, C. Marques, S. Leite, G.L. de Sena, L.C. Machado, A. Frizera, M.R.N. Ribeiro, M.J. Pontes, Polymethyl methacrylate (PMMA) recycling for the production of optical fiber sensor systems, Opt. Express 25 (24) (2017) 30051. A. Leal-Junior, A. Frizera, H. Lee, Y. Mizuno, K. Nakamura, T. Paixão, C. Leitão, M.F. Domingues, N. Alberto, P. Antunes, P. André, C. Marques, M.J. Pontes, Strain, temperature, moisture, and transverse force sensing using fused polymer optical fibers, Opt. Express 26 (10) (2018) 12939. C. Markos, A. Stefani, K. Nielsen, H.K. Rasmussen, W. Yuan, O. Bang, High-Tg TOPAS microstructured polymer optical fiber for fiber Bragg grating strain sensing at 110 degrees, Opt. Express 21 (4) (2013) 4758–4765. G. Woyessa, A. Fasano, C. Markos, A. Stefani, H.K. Rasmussen, O. Bang, Zeonex microstructured polymer optical fiber: fabrication friendly fibers for high temperature and humidity insensitive Bragg grating sensing, Opt. Mater. Express 7 (1) (2017) 286. A. Theodosiou, X. Hu, C. Caucheteur, K. Kalli, Bragg gratings and Fabry-Perot cavities in low-loss multimode CYTOP polymer fibre, IEEE Photonics Technol. Lett. 30 (9) (2018) 857–860. A. Fasano, G. Woyessa, P. Stajanca, C. Markos, A. Stefani, K. Nielsen, H.K. Rasmussen, K. Krebber, O. Bang, Fabrication and characterization of polycarbonate microstructured polymer optical fibers for high-temperature-resistant fiber Bragg grating strain sensors, Opt. Mater. Express 6 (2) (2016) 649. A.G. Leal-Junior, C. Marques, A. Frizera, M.J. Pontes, Dynamic mechanical analysis on a PolyMethyl Methacrylate (PMMA) polymer optical fiber, IEEE Sens. J. 18 (6) (2018) 2353–2361. M.C.J. Large, L. Poladian, G.W. Barton, M.A. van Eijkelenborg, Microstructured Polymer Optical Fibres, Springer, US, Boston, MA, 2008. Y. Luo, B. Yan, Q. Zhang, G.-D. Peng, J. Wen, J. Zhang, Fabrication of Polymer Optical Fibre (POF) gratings, Sensors 17 (3) (2017) 511. K. Cook, J. Canning, S. Leon-Saval, Z. Reid, M.A. Hossain, J.-E. Comatti, Y. Luo, G.D. Peng, Air-structured optical fiber drawn from a 3D-printed preform, Opt. Lett. 40 (17) (2015) 3966. A. Theodosiou, A. Lacraz, A. Stassis, C. Koutsides, M. Komodromos, K. Kalli, Planeby-plane femtosecond laser inscription method for single-peak Bragg gratings in multimode CYTOP polymer optical fiber, J. Light. Technol. 35 (24) (Dec. 2017) 5404–5410. A. Pospori, D.J. Webb, Stress sensitivity analysis of optical fiber Bragg grating-based Fabry-Pérot interferometric sensors, J. Light. Technol. 35 (13) (2017) 2654–2659. M.P. Hiscocks, M.a. van Eijkelenborg, a. Argyros, M.C.J. Large, Stable imprinting of long-period gratings in microstructured polymer optical fibre, Opt. Express 14 (11) (2006) 4644–4649. N. Zhong, Q. Liao, X. Zhu, M. Zhao, Y. Huang, R. Chen, Temperature-independent polymer optical fiber evanescent wave sensor, Sci. Rep. 5 (2015) 1–10. D.Z. Stupar, J.S. Bajic, L.M. Manojlovic, M.P. Slankamenac, A.V. Joza, M.B. Zivanov, Wearable low-cost system for human joint movements monitoring based on fiber-optic curvature sensor, IEEE Sens. J. 12 (12) (2012) 3424–3431. A. Minardo, R. Bernini, L. Zeni, Distributed temperature sensing in polymer optical fiber by BOFDA, IEEE Photonics Technol. Lett. 26 (4) (2014) 387–390. D. Sartiano, S. Sales, Low cost plastic optical fiber pressure sensor embedded in mattress for vital signal monitoring, Sensors (Switzerland) 17 (12) (2017). M. Donno, E. Palange, F. Di Nicola, S. Member, G. Bucci, F. Ciancetta, A new flexible optical fiber goniometer for dynamic angular measurements: application to human joint movement monitoring, Instrumentation 57 (8) (2008) 1614–1620. A.G. Leal-Junior, A. Frizera, L. Vargas-Valencia, W.M. dos Santos, A.P.L. Bo, A.A.G. Siqueira, M.J. Pontes, Polymer optical fiber sensors in wearable devices: toward novel instrumentation approaches for gait assistance devices, IEEE Sens. J. 18 (17) (2018) 7085–7092. Y. Hou, W.-Y. Liu, S. Su, H.-X. Zhang, J. Zhang, J. Liu, J. Xiong, Polymer optical fiber twisted macro-bend coupling system for liquid level detection, Opt. Express 22 (19) (2014) 23231. D.Z. Stupar, J.S. Bajić, B.M. Dakić, M.P. Slankamenac, M.B. Živanov, The possibility of using a plastic optical fibre as a sensing element in civil structural health monitoring, Phys. Scr. 2013 (T157) (2013) 14031. A. Vallan, M.L. Casalicchio, M. Olivero, G. Perrone, Assessment of a dual-wavelength compensation technique for displacement sensors using plastic optical fibers, Instrum. Meas. IEEE Trans. 61 (5) (2012) 1377–1383. M.A. Zawawi, S. O’Keeffe, E. Lewis, Plastic optical fibre sensor for spine bending monitoring with power fluctuation compensation, Sensors (Basel) 13 (11) (2013) 14466–14483. A. Lacraz, A. Theodosiou, K. Kalli, Femtosecond laser inscribed Bragg grating arrays in long lengths of polymer optical fibres; a route to practical sensing with POF, Electron. Lett. 52 (19) (2016) 1626–1627.

Optics and Laser Technology 111 (2019) 81–88

A.G. Leal-Junior et al.

[39] A.G. Leal-Junior, A. Frizera, M. José Pontes, Sensitive zone parameters and curvature radius evaluation for polymer optical fiber curvature sensors, Opt. Laser Technol. 100 (2018) 272–281. [40] Y. Fu, H. Di, R. Liu, Light intensity modulation fiber-optic sensor for curvature measurement, Opt. Laser Technol. 42 (4) (2010) 594–599. [41] A.G. Leal-Junior, A. Frizera, M.J. Pontes, Analytical model for a polymer optical fiber under dynamic bending, Opt. Laser Technol. 93 (2017) 92–98. [42] W. Zhang, D.J. Webb, Humidity responsivity of poly(methyl methacrylate)-based optical fiber Bragg grating sensors, Opt. Lett. 39 (10) (2014) 3026. [43] A.G. Leal-Junior, A. Frizera, L.M. Avellar, C. Marques, M.J. Pontes, Polymer optical fiber for in-shoe monitoring of ground reaction forces during the gait, IEEE Sens. J. 18 (6) (2018) 2362–2368.

[34] W. Qiu, X. Cheng, Y. Luo, Q. Zhang, B. Zhu, Simultaneous measurement of temperature and strain using a single Bragg grating in a few-mode polymer optical fiber, J. Light. Technol. 31 (14) (2013) 2419–2425. [35] A. Nag, S.C. Mukhopadhyay, J. Kosel, Wearable flexible sensors: a review, IEEE Sens. J. 17 (13) (2017) 3949–3960. [36] M. Manti, V. Cacucciolo, M. Cianchetti, Stiffening in soft robotics: a review of the state of the art, IEEE Robot. Autom. Mag. 23 (3) (2016) 93–106. [37] Z. Ding, C. Wang, K. Liu, J. Jiang, D. Yang, G. Pan, Z. Pu, T. Liu, Distributed optical fiber sensors based on optical frequency domain reflectometry: a review, Sensors 18 (4) (2018) 1072. [38] Y. Zhang, Y. Hou, Y. Zhang, Y. Hu, L. Zhang, X. Gao, H. Zhang, W. Liu, A costeffective quasi-distributed liquid leakage sensor based on the polymer optical fiber and flexible lamp belt with LEDs, Opt. Express 26 (8) (2018) 10152.

88