Remote monitoring of sleep disorder using FBG sensors and FSO transmission system enabled smart vest

Optical sensors, particularly fiber Bragg grating (FBG) sensors have achieved a fast ingress into the fields of medical diagnostic and vital signs monitoring. Wearable smart textiles equipped with FBG sensors are catching huge research attention in different applications for measurement and monitoring of physiological parameters. In this paper, we report a simple technique for remote monitoring of sleep disorder using a smart vest implemented with four FBG stress sensors located at different sides of the vest and free space optics (FSO) transmission system. The sleep disorder of the patient is monitored in real time through shifts in the original Bragg wavelengths of sensors by stress loading during random changes in patient’s sleeping postures. The reflected wavelength from a stress loaded sensor at a certain posture is transmitted over 0.5 km long FSO channel towards remote medical center, photodetected, and then can be processed in a PC to record the restlessness in a certain time interval in terms of total number of times sleeping postures are changed, total time spent at a certain posture etc. To correctly detect the stress loaded FBG sensor at the medical center, various parameters of FBG sensors and demultiplexer are carefully adjusted to minimize the power leakages from unloaded sensors that may result into errors in the detection. Maximum dynamic range around 45 dB has been achieved ensuring accurate detection. This study not only provides a cost-efficient and non-intrusive solution for monitoring the sleep disorder of patients but also can be used for real-time monitoring of various other ailments, such as lung, brain, and cardiac diseases in future.


Introduction
Sleeping is considered as an essential function in every stage of human life that allows the body and mind of an individual to remain fresh and alert to perform various physical and mental activities throughout of the day [1,2].Moreover, sleeping also helps the body to remain healthy and prevent it from various diseases.The U.S center for disease control and prevention (CDC) confirms that most of the adults require at least seven hours of night sleep [1,3].However, the CDC's statistics show that every 1 in 3 American adults does not get the recommended amount of sleep [3].Sleep disorders are the medical problems in sleep patterns of individuals affecting the overall health, safety, and quality of life [2].Sleep disorders contain problems with the quality, timing, and amount of sleep, which result in daylight anxiety and impairment in functioning [2].Moreover, some sleep disorders are severe enough to interfere with normal physical, mental, social, and emotional functioning.For example, the CDC's statistics confirm that about 75 million Americans, i.e. every 1 in 3 adults has high blood pressure due to sleep problems which is considered one of the leading risks for heart disease and stroke [3].Sleep disorders can be grouped on the basis of behaviors, issues with natural sleep cycles, breathing issues, difficulty in snoozing or how sleepy an character feels for the duration of the day [4].Some familiar forms of sleep disorders consist of (I) Insomnia, (II) Sleep Apnea, (III) Restless Legs Syndrome (RLS), (IV) Narcolepsy, (V) Shift Work Sleep Disorder (SWSD), (VI) Delayed Sleep Phase Syndrome (DSPS), and (VII) Random Eye Movement (REM) Sleep Disorder [4].
A sleep study is a check that document the various activities of the body amid sleep [5,6].There are 5 fundamental types of sleep studies that use specific methods to check for one-of-a-kind sleep characteristics and disorders.They include (I) Simple Sleep Studies, (II) Polysomnography, (III) Multiple Sleep Latency Tests (MSLTs), (IV) Maintenance of Wakefulness Tests (MWTs), and (V) Home Sleep Tests (HSTs) [5,6].Sleep research had been beneficial in figuring out and ruling out various sleep issues as mentioned above.Moreover, sleep studies have additionally been important to psychology, in which they have got provided awareness into brain activity and other physiological elements of both sleep disorders and normal sleep.
Recently, remote monitoring technologies have gained significant interest in different application areas such as mechanical engineering, process industry, structural integrity monitoring, healthcare systems, medical diagnostics , smart textiles etc [7,8].The main idea behind remote monitoring is the continuous real-time monitoring of a process based on a monitoring station that is stationed many kilometers apart from the sensors [7].The sensors usually transmit signals directly to the remotely located monitoring station enabling the operators to take certain actions immediately.Thus, remote monitoring is a viable remedy to minimize the operation staff, time-consumption, and maintenance costs [7].FBG sensors are sensitive to various environmental disturbances such as temperature, stress, strain, and pressure [7].They can be extensively used as high-accuracy sensors in different areas of civil engineering, automotive industry, aerospace, petro-chemical industry, and medical science to remotely monitor various parameters [7,[9][10][11].Compared to their electrical and mechanical counterparts, they are lightweight, resistant to electromagnetic interferences, sensitive, and flexible [12].These desirable properties enable FBG sensors as an imminent solution for monitoring physiological parameters and are especially appealing for smart textiles.
FBG based wearable systems have been widely researched to monitor various vital physiological parameters of the human body such as heart rate monitoring using smart garment [13,14], probe [15], pad [16], mattress [17][18][19], respiratory rate monitoring using smart garment [13,20], conventional oxygen cannulas [21], resistive strip [22], body temperature monitoring using FBG temperature sensors [23], smart garment [12], body postures monitoring during sleep using smart pillow [24], smart bed [8], silicon embedded FBG pressure sensors [25], plantar pressure monitoring using pressure sensitive soles [26,27], smart shoe [28], joint movement monitoring [29,30], and blood glucose monitoring using specially developed FBGs [31,32].The goal of this paper is twofold: first the experimental characterization of FBG stress sensors is performed to compare with simulated results obtained using OptiSystem software to validate the simulation results.Secondly, the proof of concept of remote monitoring of sleep disorder using a smart vest enabled with four FBG stress sensors and FSO transmission system is presented.The random changes in sleeping postures of the patient due to anxiety that is triggered by sleep disorder is monitored in real time through shifts in the Bragg wavelengths due to applied stress on the FBG stress sensors.The reflected wavelength from an FBG stress sensor that is loaded during a change in sleeping posture is transmited over 0.5 km long FSO channel towards a medical center for further processing and analysis by the medical staff.Based on above discussion, the novel contribution of this work is as.
(i) Proof of concept of FBG stress sensors and FSO transmission system enabled smart vest for remote monitoring of sleep disorder is demonstrated.
(ii) Single broadband low-cost light-emitting diode (LED) and four FBG stress sensors are used to implement the design.
(iii) Simple technique of stress loading during random changes in patient's sleeping postures and resulted shifts in Bragg wavelengths of FBG stress sensors is used for monitoring.
(iv) Maximum dynamic range around 45 dB has been achieved ensuring accurate detection.
(v) Experimental characterization of FBG stress sensors is performed and compared with simulated results using Optisystem software to validate the accuracy of the simulation results.
OptiSystem 21 commercial software from Optiwave System Inc. is used to perform this research work [33].The sequence of the remaining paper is as follows.The theory of FBG sensing is presented in section-2.section-3 discusses the experimental characterization while proposed method, and working principle are presented in section-4.Similarly, the results are presented and discussed in section-5.Finally, section-6 concludes the paper.

Theory
Figure 1 illustrates a pictograph of a Bragg grating written inside the core of an optical fibre and its response to an input broadband optical signal emitted from a broadband light source, such as LED in inset of figure 1.It may be observed that n 1 and n 2 are the refractive indices of core and cladding of the fiber, respectively while n 3 is the effective refractive index.The grating effect is achieved by creating a periodic variation in the refractive index of the optical fiber core.The Bragg diffraction condition must be achieved to enable the grating works as Bragg reflector.The Bragg condition is achieved when the product of effective refractive index (n 3 ) and grating period (Λ) become equal to half the wavelength of the light passing through the fibre [14].At this condition, the grating reflects back a small sliced part of light efficiently centered at the Bragg wavelength (λ B ) and transmits the remaining.Mathematically, the Bragg condition may be written as [14].
The grating period of an FBG sensor can be changed by a physical purturbation, such as stress, strain, temperature etc [14].The change in Bragg wavelength is linearly related to the change in strain (Δε) and temperature (ΔT) which may be defined as [8,14].
Where Δλ B and λ B are wavelength shift and original Bragg wavelength reflected from the FBG sensor towards the remotely located interrogator.Similarly, p e , α, and ξ are the photo-elastic coefficient, thermal-expansion coefficient, and thermo-optic coefficient of the core of the FBG sensor, respectively [8].

Experimental characterization
The models used in the simulation software were experimentally certified in a government-funded project NSERC/Card1 538 408-18.Measurements were performed on five different FBG sensor samples at the Algonquin Optical Research Lab for temperature and strain.The detailed report including experimental setup devised for FBG reflection and transmission measurements can be found at [34].The sensing temperature and strain were varied between 10°C to 100°C and 0 to 0.012 micro-strain, respectively.The comparison between simulation and experimental results is shown in figure 2. It is evident that the simulated results are in close agreement with the experimental results, which validate the simulation models used in the OptiSystem commercial software.Moreover, the root mean square (RMS) error in wavelength shifts for reflection is calculated for the measured data which is around 0.28% and 1.39% for temperature and strain sensing, respectively.This error can be referred to diversion from the actual central wavelength of the commercial laser, the thermo-optic coefficient, and the photo-elastic coefficient employed in simulation.

Working principle and sensing architecture
Figure 3(a) shows the encapsulation of four FBG stress sensors using Silicon-rubber and then glued at different sides of the smart vest i.e.S1, S2, S3, and S4, which are located at back, near the right shoulder, near the left shoulder, and at middle of the chest, respectively such that a particular sensor experiences the maximum stress loading during a sleeping posture as shown in figure 3(b).The thickness of coating of the Silicon-rubber used to encapsulate the FBG sensors is 0.102 in (≈2.6 mm).The advantage of choosing Silicon-rubber as the encapsulation material for FBGs is that various mechanical parameters of the rubber are similar to those of human skin [24].When the Silicon-rubber is loaded, it generates a large range of elastic compression which is evenly transferred to the encapsulated FBG stress sensors.It is evident that only one sensor is loaded at a time when the patient occupies a particular sleeping posture randomly.Therefore, S1, S2, S3, and S4 are loaded on occupying the starfish lying (SL), log right lying (LRL), log left lying (LLL), and free faller lying (FFL) postures, respectively as shown in figure 3(b).The smart vest is assumed to be worn by a male subject having a mass of 125 kg and the area of the each encapsulated FBG stress sensor glued at different sides of the vest is 0.5 in 2 (i.e.≈0.003 ft 2 ).For simplicity, we suppose that equal stress is exerted by the patient over each sensor for each posture.The stress loading over each sensor may be calculated using the expression [35].
Where F is the load in Newton and A is the area in ft 2 .This equation may be re-written as.Where m is the mass of the body and g is the gravitational acceleration.Thus as per equation (4), a weight of 1225 N exerts around 5 MPa pressure onto the sensing surface of an area of 0.5 in 2 .Figure 4 illustrates the sensing architecture of the proposed method.It may be perceived that the subject wearing the smart vest is lying on the bed.The fiber pigtails of FBG stress sensors located at different sides of the smart vest are combined using a power combiner (PC) that is attached at the nearest wall.It may be observed that the broadband light signal from LED centered at 1552 nm whose spectral plot is shown in figure 5, is split into four parts of equal powers using a 1 × 4 power splitter (PS).Each part is then given as input to the sensing  element of FBG stress sensors that are fixed at different sides of the vest.The original Bragg wavelengths of S1, S2, S3, and S4 are 1542 nm, 1547 nm, 1552 nm, and 1557 nm, respectively.The subject having the sleep disorder randomly changes the sleeping postures and loads the respective FBG sensor causing reflection to input optical signal at a shifted wavelength.To realize this scenario, the FBG stress sensors are randomly selected in the software for twenty sweep iterations.
It is important to mention here that the outputs of all four FBG stress sensors are combined together using a 4 × 1 PC and connected with a 20 m long piece of standard single-mode fiber (SSMF).The SSMF feeds the transmitter telescope installed at the roof-top pointing towards receiver telescope that is also installed at the roof-top of medical center, thus establishing a line-of-sight (LOS) FSO link.The reflected light at a shifted wavelength which is around 0.7 nm corresponding to stress of 5 MPa from a loaded FBG stress sensor, is amplified using an optical amplifier (OA) and then transmitted over 0.5 km FSO channel towards the medical center.A point-to-point LOS-FSO channel between the patient's home and health center is considered.We consider that the FSO channel is memoryless, stationary, and ergodic.Hence, the system may be represented by the the following mathematical expression [36].
Where g is intensity gain, Y is received signal, S is the reflected optical signal, η is the conversion efficiency of photodetectors (PDs), I is intensity, and n is additive white Gaussian noise (AWGN).The intensity of reflected signals from FBG sensors transmitted over FSO channel is mainly affected by atmospheric attenuation and turbulence.The atmospheric attenuation is modeled by Beer-Lambert's law as [36].
Where ξ is the scattering coefficient.The scattering coefficient depends on the weather and is a function of the visibility ν, that can be calculated using the expression [36].
The intensity of the optical signal received at the PD randomly varies called turbulence or intensity scintillation [37] which is a primary cause to degrade the performance of FSO systems.Various FSO channel models have been proposed to estimate turbulence, but the Gamma-Gamma channel model is commonly used for all types of turbulence, i.e., weak, medium, and strong turbulence strengths.Therefore, the probability density function (PDF) of the received optical signal's intensity variation at PD is a function of the small and large scale turbulence eddies (α and β) of Gamma-Gamma channel model, is given as [38,39].
Where K(. ) is 2nd order Bessel function and ( ) ( ) 1 is the Gamma function in terms of propagation distance z. α and β are related to variance of the received irradiance by the following equation [38,39].
2.4 1.17After FSO transmission, the received signal is demuliplexed using 1 × 4 wavelength division demultiplexer (DEMUX) with its channel wavelength is centered at one of the FBG Bragg wavelength and its bandwidth covers the possible shift of each grating sensor.The technical specifications of different components related to FSO system are shown in table 1 [40].
The optical signal is photodetected and can be processed by the medical staff.The important simulation parameters used in this work are listed in table 2.

Results and discussion
Figure 6 shows the spectral plots of original and shifted Bragg wavelengths of FBG sensors without and with stress loading for reflection and transmission.It may be observed from figures 6(a)-(d) that the reflected signals from S1, S2, S3, and S4 after stress loading of 5 MPa are at shifted wavelengths of 1542.7 nm, 1547.7 nm, 1552.7 nm, and 1557.7 nm, respectively.Similarly, it is evident from figures 6(e)-(h) that the transmitted light spectra from S1, S2, S3, and S4 experience a sharp dip after stress loading of 5 MPa exactly at shifted wavelengths of 1542.7 nm, 1547.7 nm, 1552.7 nm, and 1557.7 nm, respectively.The proposed system works for different weights of the patients.We have chosen the center wavelength separation of the four FBG sensors to be 5 nm, which allows using an optical bandpass filter with maximum bandwidth of 5 nm.Thus, the grating wavelength could shift within ±2.5 nm from the center due to the applied pressure by the patient.The bandwidth of the used optical bandpass filter in the simulation is 650 GHz, which allows maximum stress of 5.22 MPa.
The performance of the sensors is analyzed under different weather conditions such as light fog, dense fog, and heavy fog considering the Gamma-Gamma channel model.These conditions are simulated in the software by changing the atmospheric attenuation value.Therefore, the atmospheric attenuation values used in the simulation for light haze, heavy haze, and heavy fog are 2 dB/km, 10 dB/km, and 21 dB/km, respectively [36,41]. Figure 7 shows plots depicting the random loading of FBG sensors as the patient changes the sleeping postures for twenty sweep iterations.It may be observed that S1, S2, S3, and S4 are loaded for 3, 7, 9, and 1 times, respectively.In other words, we can say that the patient occupies SL, LRL, LLL, and FFL postures for 3, 7, 9, and 1 times, respectively.
It is clear that the average power of reflected signals is around 16 dBm while the average power leakage from other sensors is around −24 dBm.Therefore, average dynamic range around 40 dB has been achieved ensuring accurate detection of respective sensors.To further elaborate the loading scenario of FBG stress sensors and posture's occupation as discussed above, bar plot of figure 8 shows the number of times a particular FBG sensor is loaded or a sleeping posture is occupied by the patient.Figure 9 shows the plots depicting the random loading of FBG sensors as the patient changes the sleeping postures for twenty sweep iterations.Again, it may be observed that S1, S2, S3, and S4 are loaded for 3, 7, 9, and 1 times, respectively.It is clear that the average power of back reflected signals is now around 7 dBm while the average power leakage from other sensors is around −30 dBm.The decrease in average power of reflected lights and average power leakage is due to the increase in atmospheric attenuation from 2 dB/km to 10 dB/km.Therefore, average dynamic range around 43.5 dB has been achieved ensuring accurate detection of respective sensors.
Similarly, figure 10 shows the plots depicting the random loading of FBG sensors as the patient changes the sleeping postures for twenty sweep iterations.Again, it may be observed that S1, S2, S3, and S4 are loaded for 3, 7, 9, and 1 times, respectively.It is clear that the average power of back reflected signals become around 2 dBm while the average power leakage from other sensors is around -40 dBm.The further decrease in average power of reflected lights and average power leakage is due to the increase in atmospheric attenuation from 10 dB/km to    As previously mentioned, an estimated one in three people report regular sleep problems.So it is no surprise that people are more concerned than ever about getting enough sleep.This increased interest has led to an explosion of contactless sleep trackers that measure how much sleep you get each night.Most sleep trackers are watches worn on the wrist of the subject and monitoring the body movements of the subject during sleep to estimate the time spent in each sleep cycle.Despite the popularity of sleep trackers, the accuracy and reliability of sleep trackers is a topic of great interest.Only few studies have investigated about the accuracy of sleep monitoring devices.Previous research has found that sleep trackers are only 78% accurate at distinguishing between sleep and wakefulness, compared to polysomnography which experts use to diagnose sleep disorders [42,43].Nevertheless, the electrodes attached to the human body in later technique can create discomfort among the subjects, hence affecting the normal sleep behavior [43].Therefore, future researchers are strongly encouraged to conduct research on the accuracy, validity, and reliability of different brands of sleep monitoring devices in different populations, including psychiatric patients, burn patients, rehabilitation patients, intensive care patients, children with autism, patients with dementia, the elderly, and young children.The results can be useful for treating the patients of sleep disorder and various other ailments, such as lung, brain, and cardiac diseases in future.

Conclusion
A simple method for remote monitoring of sleep disorder and proof of concept of FBG stress sensors and FSO transmission enabled smart vest is demonstrated.Four FBG stress sensors located at different sides of a vest are used to detect the random changes in sleeping postures of a patient through monitoring the wavelength shift of the reflected optical signal of the FBG stress sensors.The reflected wavelength from an FBG stress sensor that is loaded during a change in sleeping posture is transmited over 0.5 km long FSO channel towards a medical center for processing and analysis by medical staff.Maximum dynamic range around 45 dB has been achieved for accurate detection by carefully adjusting various parameters of fiber Bragg grating sensors and wavelength division demultiplexer to minimize the power leakages from unloaded sensors that may result into errors in detection.Experimental characterization of sensors is performed and compared with simulated results to validate the accuracy between experimental and simulation results.

Figure 2 (
a) and figure 2(b) show the simulated and experimentally measured wavelength shifts in FBG sensors' reflection and transmission for different values of strain, respectively.Similarly, figure 2(c) and figure 2(d) show the simulated and experimentally measured wavelength shifts in FBG sensors' reflection and transmission for different values of temperature, respectively.

Figure 1 .
Figure 1.Construction and working principle of FBG sensor.

Figure 2 .
Figure 2. (a) Micro-strain versus wavelength shift plots based on experimental and simulation data for FBG sensor's reflection port (b) Micro-strain versus wavelength shift plots based on experimental and simulation data for FBG sensor's transmission port (c) Temperature versus wavelength shift plots based on experimental and simulation data for FBG sensor's reflection port (d) Temperature versus wavelength shift plots based on experimental and simulation data for FBG sensor's transmission port.

Figure 4 .
Figure 4. Proposed sensing architecture, LED: Light-emitting diode, PS: 1 × 4 Power splitter, SSMF: Standard single-mode fiber, OSA: Optical spectrum analyzer, S: FBG stress sensor, PC: Power combiner, OA: Optical amplifier, FSO: Free space optical communication link, DEMUX: Demultiplexer, PD: Photodetector.OSAs are used only for analysis in the software and not needed in realization.

Figure 5 . 6
Figure 5. Spectral plot of input optical signal from LED.

Figure 6 .
Figure 6.Original and shifted wavelength plots of FBG sensors without and with stress loading (a) and (e) S1 reflection and transmission (b) and (f) S2 reflection and transmission (c) and (g) S3 reflection and transmission (d) and (h) S4 reflection and transmission.These plots are taken before FSO transmission.

Figure 8 .
Figure 8. Sequence of (a) Loading of different FBG sensors (b) Occupation of different sleeping postures.

Table 2 .
Main simulation parameter and their values.