Utilization of an OLED-Based VLC System in Office, Corridor, and Semi-Open Corridor Environments

Organic light emitting diodes (OLEDs) have recently received growing interest for their merits as soft light and large panels at a low cost for the use in public places such as airports, shopping centers, offices, and train or bus stations. Moreover, the flexible substrate-based OLEDs provide an attractive feature of having curved or rolled lighting sources for the use in wearable devices and display panels. This technology can be implemented in visible light communications (VLC) for several applications such as visual display, data communications, and indoor localization. This article aims to investigate the use of flexible OLED-based VLC in indoor environments (i.e., office, corridor and semi-open corridor in shopping malls). We derive a two-term power series model to be match with the root-mean-square delay spread and optical path loss (OPL). We show that, for OLED positioned on outer-wall of shops, the channel gain is enhanced in contrast to them being positioned on the inner-wall. Moreover, the channel gain in empty environments is higher compare with the furnished rooms. We show that, the OPL for a 10 m link span are lower by 4.4 and 6.1 dB for the empty and semi-open corridors compared with the furnished rooms, when OLED is positioned on outer-wall of shops. Moreover, the channel gain in the corridor is higher compared with the semi-open corridor. We also show that, in furnished and semi-open corridors the OPL values are 55.6 and 57.2 dB at the center of corridor increasing to 87.6 and 90.7 dB at 20 m, respectively, when OLED is positioned on outer-wall of shops.

For further improvement, the field of view angles of these PDs can be optimized to increase the received signal-to-noise-ratio (SNR) as reported in [12,13]. Optical cameras have been recently introduced to capture data [14,15], resulting in interesting VLC systems applications. It is also possible to utilize the organic PDs with more synthetic flexibility [16], however the limited spectral responsivity range is the drawback. VLC relies on the use of light-emitting diodes (LEDs), organic LEDs (OLEDs) as well as white laser diodes (LDs) as the light source [17]. Owing to OLEDs attractive advantages, including transparent displays, rich color, low power consumption and large active areas [18,19], there has been a growing interest in using OLEDs for soft lighting and display applications in public places [20]. The main differences between OLEDs and the LEDs are (i) the modulation bandwidth of OLEDs, which increases linearly with the drive current, is lower than silicone LEDs (i.e., kHz compared to MHz); and (ii) OLEDs have wider radiation patterns compared with non-organic LEDs, which influences the optical path loss (OPL). This work emphasizes on utilizing OLEDs for VLC systems, where is the potential of flexible light sources could be used in office and public indoor environments.
Numerous efforts have been made in modeling VLC channel in order to determine the channel impulse response (CIR) and its characteristics in terms of the average OPL and the root-mean-square (RMS) delay spread. The achievable SNR for a given transmit power can be calculated by OPL obtained from CIR [21]. In addition, the RMS delay spread provides a good estimate of how susceptible the channel is to inter-symbol interference (ISI), thus leading to transmission data rate R b restriction [18,21]. That is why quantifying channel characteristics is vital; hence, a number of studies have been done so far. For instance, in [22], the CIR of an empty room was evaluated using Monte Carlo (MC) ray tracing at the visible wavelength range where the surface materials reflectance were not wavelength-dependent. However, in [23] the VLC channel was investigated including wavelength-dependent reflectance of materials. In [24], the modified MC ray tracing approach was used for analyzing the CIR as a function of the wavelength using a simplified matrix model. In [25], a three-dimensional (3D) model based on MC algorithm using a CAD software was presented for the VLC system. A simulation of VLC channel by the use of OpticStudio ® simulator produced by Zemax [26] was reported in [27,28], which was endorsed by the IEEE 802. 15.7r1 Task Group. In addition, the use of OpticStudio ® for validation of the channel modelling was reported in [29]. Recently, utilizing OLEDs in VLC systems has captured attention. In [30], it is claimed that the use of curved OLED in VLC system for an empty room offers lower RMS delay spread and the average OPL values of 8.8% and 3 dB, respectively compared with Lambertian source. The impact of reflections using flat and half-circular OLEDs in a furnished office was investigated [31]. The recorded results in [31] reveals the ability of OLED based VLC system to achieve R b of 4 Mb/s with a bit-error-rate (BER) below the forward error correction BER limit. In [32], investigating of a flexible OLED-based VLC link in a shopping mall was reported, in both empty and furnished rooms using both full and half-circular OLEDs. The results indicated that, the OPL in an empty room is about 5 dB less than the furnished room.
Currently OLED panels are more costly than LEDs; however, with advances made in fabrication and manufacturing as well as the wider use of OLED-based lights the cost will be reduced as was the case with the non-organic LEDs a few years ago. Since, OLEDs come in different shapes and size, we have decided to investigate their characteristics when used as a transmitter (Tx) in VLC systems. This work emphasizes on the evaluation of an attractive feature of OLEDs, which is the mechanically flexible potential for utilizing in VLC system. The simulation was carried out to determine the impact of the symmetrical beam pattern of curved OLEDs, which is wider than Lambertian, on the VLC channel. In this work, we consider a VLC system in a typical office, corridor, and semi-open corridor environments with and without furniture. In the office environment, the user (i.e., the Rx), is moving along a circular path while holding a mobile phone. In corridor and semi-open corridors, the user is then moving on a straight path along the corridor. We investigate the proposed system optical features and show a new numerical model for the RMS delay-spread and OPL for the channel. We provide statistics for the BER performance and compared it for curved and flat OLED-based VLC systems. The rest of the paper is organized as follows. In Section 2, the features of simulation and scenarios are described. Section 3 discusses the results. Finally, conclusions are given in Section 4.

Simulation Features
To determine the detected optical power and path lengths from the Tx to the Rx, non-sequential ray-tracing approach was used in the 3D environment. It evolves the specification and location of the Tx and the Rx, features of the CAD models of objects, wavelength-dependent reflectance of surfaces (wall, ceiling, floor, and objects), and transmission/reflection coefficient of glass windows. Next, the captured output data of the OpticStudio ® is processed in MATLAB to obtain the CIR expressed as given by [27,32]: where P i and τ i are the power and the propagation time of the ith ray, respectively. δ is Dirac delta function and N is the number of rays received at the Rx. Note, a number of reflections from the floor, ceiling, walls, and other objects are considered until the normalized intensity of rays after intercepting an object drops to 10 −3 .
The spatial intensity distribution of light emitted from the light source is determined by the optical radiation pattern profile. The luminous intensity defined in terms of the angle of irradiance φ is given as [1]: where I(0) is the center luminous intensity of the OLED and m L is Lambertian order, which is defined in terms of the Tx semi-angle φ 1/2 as [1]: As inputs of the simulator, the measured characteristics of a flexible OLED from UNISAGA with a size of 200 × 50 mm 2 , see Figure 1a, were used. The measured beam pattern of the flexible OLED for flat and a half-circular configuration is depicted in Figure 1b, showing symmetry but not fitting with Lambertian radiation pattern (the solid line for m L = 1). A close match between the simulated and the measured beam patterns can be seen in Figure 1b. The measured spectrum profile of the flexible OLED is presented in Figure 1c, showing the red, green, and two blue components at 620, 553 and 454 and 480 nm, respectively.

Simulation Features
To determine the detected optical power and path lengths from the Tx to the Rx, non-sequential ray-tracing approach was used in the 3D environment. It evolves the specification and location of the Tx and the Rx, features of the CAD models of objects, wavelength-dependent reflectance of surfaces (wall, ceiling, floor, and objects), and transmission/reflection coefficient of glass windows. Next, the captured output data of the OpticStudio ® is processed in MATLAB to obtain the CIR expressed as given by [27,32]: where Pi and are the power and the propagation time of the i th ray, respectively. is Dirac delta function and N is the number of rays received at the Rx. Note, a number of reflections from the floor, ceiling, walls, and other objects are considered until the normalized intensity of rays after intercepting an object drops to 10 −3 .
The spatial intensity distribution of light emitted from the light source is determined by the optical radiation pattern profile. The luminous intensity defined in terms of the angle of irradiance is given as [1]: where I(0) is the center luminous intensity of the OLED and mL is Lambertian order, which is defined in terms of the Tx semi-angle 1/2 as [1]: As inputs of the simulator, the measured characteristics of a flexible OLED from UNISAGA with a size of 200 × 50 mm 2 , see Figure 1a, were used. The measured beam pattern of the flexible OLED for flat and a half-circular configuration is depicted in Figure 1b, showing symmetry but not fitting with Lambertian radiation pattern (the solid line for mL = 1). A close match between the simulated and the measured beam patterns can be seen in Figure 1b. The measured spectrum profile of the flexible OLED is presented in Figure 1c, showing the red, green, and two blue components at 620, 553 and 454 and 480 nm, respectively.   Figure 2 shows the analyzed example of an office environment designed with the size of 10 × 10 × 3 m 3 and a number of objects within. Here, a curved OLED with the size 1 × 0.5 m 2 is mounted on the wall with a curvature radius of 32 cm. In the office environment, the scenario is to move the Rx over a semi-circular path, where the radius d is 2 m. An angle of radiation θ with respect to the normal from the center point of OLED (i.e., −90 • < θ < 90 • ) is given, see Figure 3. The Rx height is assumed to be 1 m above the floor to represent people holding mobile phones while sitting at their desks. In simulation, we have not considered the synchronization. However, in real time systems synchronization protocols defined by the standards will be adopted, which does not affect the transmission characteristics of the proposed system.  Figure 2 shows the analyzed example of an office environment designed with the size of 10 × 10 × 3 m 3 and a number of objects within. Here, a curved OLED with the size 1 × 0.5 m 2 is mounted on the wall with a curvature radius of 32 cm. In the office environment, the scenario is to move the Rx over a semi-circular path, where the radius d is 2 m. An angle of radiation θ with respect to the normal from the center point of OLED (i.e., −90° < θ < 90°) is given, see Figure 3. The Rx height is assumed to be 1 m above the floor to represent people holding mobile phones while sitting at their desks. In simulation, we have not considered the synchronization. However, in real time systems synchronization protocols defined by the standards will be adopted, which does not affect the transmission characteristics of the proposed system.     Table 1.

Comparison of Flat and Curved OLED Based System Performance
For intensity modulation/direct detection (IM/DD) optical transmission systems, the electrical SNR is defined as where γ is the photodetector's responsivity in (A/W), P E and P R are the emitted and received optical power, respectively, and N 0 /2 is double-sided power spectral density. Considering a link with non-return-to-zero (NRZ) on-off keying (OOK), the BER is given as [33]: Figure 6 shows the plots of the BER for flat and curved OLEDs at R b of 4 and 6 Mb/s along with the 7% forward error correction (FEC) BER limit of 3.8 × 10 −3 [1] for an office. Note, for 4 Mb/s the BER is below the FEC limit for curved OLED. As illustrated, the BER plot displays a symmetry about the origin (i.e., at θ of 0 • ) because of the same achievable SNR that is maintained across the entire face of OLED. It is obvious that, for the curved OLED the BER is improved over a wider θ compared with the flat OLED. Note, for the flat OLED with −30 • < θ < 30 • the BER values are <10 −6 . At R b of Sensors 2020, 20, 6869 8 of 15 4 Mb/s, the BER remains below the FEC limit for θ within the range of ±90 • and ±53 • for the curved and flat OLEDs, respectively. However, for R b of 6 Mb/s, θ drops by 15 • and 4 • for the curved and flat OLEDs, respectively.
Sensors 2020, 20, x FOR PEER REVIEW 9 of 16 origin (i.e., at θ of 0°) because of the same achievable SNR that is maintained across the entire face of OLED. It is obvious that, for the curved OLED the BER is improved over a wider θ compared with the flat OLED. Note, for the flat OLED with −30° < θ < 30° the BER values are <10 −6 . At Rb of 4 Mb/s, the BER remains below the FEC limit for θ within the range of ±90° and ±53° for the curved and flat OLEDs, respectively. However, for Rb of 6 Mb/s, θ drops by 15° and 4° for the curved and flat OLEDs, respectively.

Channel Charactristics
The channel gain H(0) defines the achievable SNR for a given incident power. To quantify the data rate, H(0) and the optical signal attenuation OPL = −10log 10 (H(0)) caused by reflections and transmission in the free space are obtained [21,34]. The RMS delay spread is commonly used to define Sensors 2020, 20, 6869 9 of 15 the time dispersion along the propagation path. The channel mean excess delay τ and the RMS delay spread τ RMS are given as [27,31]. Figure 7 depicts the τ RMS plot for the flat and curved OLEDs in an office. The angle θ is shown in Figure 7 to identify the Rx's location on the semi-circular path with the radius d. τ RMS increases with θ reaching the maximum value of 5 and 10.7 ns at θ of 90 • for the curved and flat OLEDs, respectively. It is obvious that, for the curved OLED, there is a slight change in τ RMS by about 0.8 ns with respect to θ. However, τ RMS has changed about 7.3 ns for the flat OLED. Note, there is a significant increase in τ RMS for θ > 40 • for the flat OLED.
the time dispersion along the propagation path. The channel mean excess delay τ and the RMS delay spread τRMS are given as [27,31].  Figure 7 to identify the Rx's location on the semi-circular path with the radius d. τRMS increases with θ reaching the maximum value of 5 and 10.7 ns at θ of 90° for the curved and flat OLEDs, respectively.
It is obvious that, for the curved OLED, there is a slight change in τRMS by about 0.8 ns with respect to θ. However, τRMS has changed about 7.3 ns for the flat OLED. Note, there is a significant increase in τRMS for θ > 40° for the flat OLED.
Using a non-linear approximation algorithm for both cases, a two-term power series model can be derived from simulations for RMS as a function of θ given by where p1, p2 and p3 are summarized in Table 2. Note, the empirical parameters can vary based on the number of objects in the room and the size of the specified confined space.   Using a non-linear approximation algorithm for both cases, a two-term power series model can be derived from simulations for τ RMS as a function of θ given by where p 1 , p 2 and p 3 are summarized in Table 2. Note, the empirical parameters can vary based on the number of objects in the room and the size of the specified confined space.  Figure 8 shows the azimuthal dependence of the OPL distributions for flat and curved OLEDs for the proposed scenario in an office. OPL increases with θ reaching a maximum of 60.4 and 70.2 dB at θ of 90 • for curved and flat OLEDs, respectively. Note, up to~14 dB drop in the channel gain can be seen for the flat OLED when θ changes from 0 • to 90 • , which is considerably higher compared with the reduced~2 dB channel gain for curved OLED. It can be seen that, for the flat OLED and θ < 30 • there is an improvement in OPL by~1.8 dB compared with the curved OLED. However, for θ > 45 • , there is high received power enhancement for the curved OLED compared with the flat OLED, e.g., OPL penalties for flat OLED are 5 and 10 dB for θ of 75 • and 90 • , respectively. In addition, for both cases OPLs can be determined as the 2-term power series models as where the derived parameters a 1 , a 2 and a 3 are shown in Table 3.
Sensors 2020, 20, x FOR PEER REVIEW 11 of 16 Figure 8 shows the azimuthal dependence of the OPL distributions for flat and curved OLEDs for the proposed scenario in an office. OPL increases with θ reaching a maximum of 60.4 and 70.2 dB at θ of 90° for curved and flat OLEDs, respectively. Note, up to ~14 dB drop in the channel gain can be seen for the flat OLED when θ changes from 0° to 90°, which is considerably higher compared with the reduced ~2 dB channel gain for curved OLED. It can be seen that, for the flat OLED and θ < 30° there is an improvement in OPL by ~1.8 dB compared with the curved OLED. However, for θ > 45°, there is high received power enhancement for the curved OLED compared with the flat OLED, e.g., OPL penalties for flat OLED are 5 and 10 dB for θ of 75° and 90°, respectively. In addition, for both cases OPLs can be determined as the 2-term power series models as where the derived parameters a1, a2 and a3 are shown in Table 3.        In furnished environments, τ RMS increases with the distance reaching the maximum value of 40 and 60 ns at a distance of 16 m for the corridor and semi-open corridor, respectively. In all environments, τ RMS drop significantly for case2 compared with case1. Note, in both furnished environments, there is a huge drop in τ RMS for case2 compared with case1 for d y up to 10 m; however, for d y > 10 m τ RMS for case2 reaches the corresponding value of case1. E.g., in a furnished corridor, for case2, τ RMS values are lower than case1 by 13, 6 and 2 ns at d y = 0, 8 and 14 m, respectively. However, in an empty corridor τ RMS for case2 drop by 11, 5 and 5 ns at d y = 0, 8 and 14 m, respectively.
Note, positioning OLEDs behind the window will result in decreased received optical power compare with when located on the outer-wall of shops inside the corridor; i.e., OPL for case2 is lower than case1 in all empty and furnished indoor environments. For instance, in a furnished semi-open corridor for case1, OPL penalties of 6.5, 4.8, 3.4, and 1.6 dB at 5, 10, 15, and 20 m, respectively, can be seen in comparison to case2. Additionally, for both cases, OPL in the corridor is lower compared with the semi-open corridor. e.g., in furnished environments and case2, OPL in the corridor reaches the maximum of 87.6 dB, which is lower than the value corresponding to the semi-open corridor (i.e., 90.7 dB). Note, the channel gain for both cases in furnished environments for d y < 4 m remains the same; however, for case1 and In addition, the channel gain enhancement in an empty corridor in contrary with furnished one is 5 dB at 5 m reaching 7 dB at 15 m for case1. However, for case2 it remains around 4 dB for d y > 4 m. In addition, the channel gain enhancement in an empty semi-open corridor in contrary with furnished one are 7.2 and 5 dB at 5 m increasing to 9.1 and 7 dB at 15 m for case1 and case2, respectively.
Using a non-linear approximation algorithm for both cases in all environment, a two-term power series model have been derived from simulations for τ RMS and OPL as a function of d y given by where the derived values of r 1 , r 2 , r 3 , l 1 , l 2 , and l 3 are summarized in Tables 4 and 5.

Conclusions
In this paper, we investigated the performance of OLED-based VLC system and the channel characteristics in office, corridor, and semi-open environments. The measured beam pattern profile of the curved OLED was closely matched with the simulation result. We showed that, when a flat OLED was used in an office, τ RMS increased significantly by 7.3 ns compared with 1 ns for the curved OLED. In the office, contrary to the flat OLED, the curved OLED showed improved BER performance over a wider range of θ. A data rate of 4 Mb/s was achieved using both the curved and flat OLEDs for θ within the range of ±90 • and ±53 • , respectively. A two-term power series model was found to match τ RMS and OPL as a function of θ and d y for the office and corridors, respectively and models' parameters for all three environments with and without furniture were derived. We showed that, when OLED is positioned on the outer wall of shops inside the corridor, the channel gain enhanced in contrast to them being located on inner shop wall, e.g., the channel gain enhanced by 5.