1. Introduction

With the development of the mobile internet technology and the coming of 5G mobile networks, the applications such as short video, real-time video in mobile phone make the wireless communication network traffic more vulnerable [1,2]. In order to cope with the serious congestion of wireless network, optical wireless communication (OWC) using infrared spectrum is becoming impressively besides the traditional Wi-Fi and visible light communication (VLC) technology. Because the OWC system can provide ultra-high transmission capacity with almost unlimited bandwidth theoretically and is compatible with current fibre-to-the-home networks [3–6]. A typical OWC system realize point-to-point (PtP) communication by using collimated infrared beam, that means only one user can communicate with the base station at a time. In order to increase system efficiency and practicality, it is necessary for multiple users in different areas to receive the optical signal from the base station at the same time. Thereby, the point-to-multi-point (PtMP) broadcasting scheme will be preferred in an OWC system. The first construction of indoor OWC system with multiple beams based on three-dimensional (3D) beamforming was proposed in [7].

Two different approaches have been proposed to realize PtMP optical wireless broadcasting systems [8]. The first one is based on wavelength control and realized by de-multiplexing the wavelength division multiplexing optical signals. A dispersive device is usually implemented to disperse the input optical wavelengths into different angles of emergence in free-space. Each wavelength corresponds to one user. The users are located in different areas with their own exclusive wavelengths. To improve the total transmission capacity, one can increase the number of wavelengths at the transmitter side. In [9], by using two gratings with matched free spectral range, an optical wireless system with 2D broadcasting function is demonstrated with the total transmission capacity of 297.6 Gb/s (37.2 Gb/s per wavelength × 8 wavelengths). The other one is splitting the optical beam into multiple ones called beam control. This approach shares one input optical wavelength, which means just one laser source is needed. A spatial light modulator based on liquid crystal on silicon (LCoS-SLM) is usually employed to generate the multiple beams by downloading different hologram images into LCoS. This architecture has high flexibility due to its programmable and reconfigurable ability. In [10–12], holographic beam steering optical wireless system with Gerchberg-Saxton (GS) algorithm were presented for two, three and four nomadic users. The total transmission capacities are 50 Gb/s (25 Gb/s per beam × 2 beams), 30 Gb/s (10 Gb/s per beam × 3 beams) and 96 Gb/s (24 Gb/s per beam× 4 beams) by splitting the optical beam into two, three and four, respectively. But the generation of the hologram images by GS algorithm is complex, which slows down the refresh rate of LCoS device and hinders the operability in the actual application. In our previous work [13], we experimentally demonstrated a broadcasting system using LCoS device based on rotated-splitting-SLM (RSS) algorithm, realizing 2D beam steering for four nomadic users with the total capacity of 240 Gb/s (60 Gb/s per beam × 4 beams). However, each splitted beam cannot be manipulated independently. It is still a challenge by using the beam control approach as further increasing the number of nomadic users with random locations.

In this paper, a modified RSS algorithm is proposed to implement a 2D continuous tunable optical wireless broadcasting system for the arbitrarily distributed access points. The angles of emergence for each beam can be controlled independently according to the locations of the randomly distributed users. An angle magnifier (AM) is introduced to expand the field-of-view (FoV) and two matched fiber collimators with calculated beam waists are designed to mitigate the coupling loss and system complexity. With the help of the proposed algorithm and the designed optical subsystem, a 2D programmable broadcasting system with up to 16 access users and ±15° FoV is successfully achieved over 1.2-m indoor free-space link and 1-km standard single mode fiber (SSMF). Our demonstrated system enables to transmit the total capacity of 1.47 Tb/s (92 Gb/s per beam × 16 beams) with the coverage diameter up to 0.64 m at the bit-error-ratio (BER) level of less than 10⁻⁴.

2. Operation principle

Figure 1 illustrates the schematic diagram of the multi-access points optical wireless broadcasting system for communicating with multiple randomly distributed users. In a house or building, one or more beam steering and broadcasting (BSB) modules should be implemented on the ceiling of the room. Each BSB module is arranged with a single wavelength by the optical cross connector which connected to a communication control center. Optical signal coming from the optical fiber access network is then sent to the mobile devices (MDs) via the BSB module. The BSB module consists of three parts: a fiber collimator, an LCoS device and an AM. The collimated beam is divided into several individual ones by the LCoS with modified splitting algorithm, which are steered to arbitrary positions according to the users’ locations. An AM is commonly used in an OWC system to expand the FoV of the transmission system. The angle-expended beams then launch into the indoor free space to the mobile terminals, which are received by the matched fiber collimators for further detection and processing.

 

Fig. 1. Schematic diagram of the multi-access points optical wireless broadcasting system for randomly distributed users. OXC: optical cross connector, CCC: communication control center, BSB: beam steering and broadcasting, SLM: spatial light modulator, AM: angle magnifier, MD: mobile devices.

Download Full Size | PDF

2.1 Optical design

In a typical OWC system, a transmitter fiber collimator is usually utilized to launch optical signal from SSMF into free space, and a receiver fiber collimator is utilized to couple the optical beams into fiber for photodiode (PD) detection. So the coupling loss from free space to SSMF via collimator is the main part of the total power loss in the optical link. The normalized coupling loss due to optical beam waist mismatch between two collimators is represented by [14]:

Figure 2 shows the performance of normalized coupling loss versus beam waist mismatch factor ɛ. It can be seen that when 0.5<ɛ<2, the theoretical coupling loss due to the mismatch of the Gaussian beam waist is less than 5 dB. That is to say, if we want to get a low optical path loss design, employing a pair of matched optical collimator (0.5<ɛ<2) is necessary. In our previous work [13], we do not employ a AM in the optical path design, so we used the same optical collimator at the receiver side as at the transmitter side to make ɛ=1. But the FoV of the system is limited to ±3°. In [15], the system employed an AM to expend the FoV from ±3° to ±30°. In order to control the optical loss under an acceptable level, they used the same AM at the receiver side to ensure ɛ=1. But this scheme increases the system cost and complexity because the used AM is consist of three focusing lens. In our optical path design here, we use focusing lens 1 with focal length of f₁=100 mm and focusing lens 2 with focal length of f₂=10 mm to obtain about 10 times angle magnification. The angle magnification factor γ is defined as γ= f₁ / f₂ [16].

 

Fig. 2. Coupling loss versus beam waist mismatch factor. Inside: the transmitter collimator with beam waist ω_T = 6.5 mm and the receiver collimator with beam waist ω_R = 0.42 mm.

Download Full Size | PDF

After passing through the AM, the optical beam waist ω'_T is described as:

Here we employed a transmitter collimator with beam waist ω_T = 6.5 mm. So after passing through the AM, ω'_T = 6.5/10 = 0.65 mm. If we employ the same optical collimator at the receiver side as at the transmitter side, the ɛ = 6.5/0.65 = 10. As we can see in Fig. 2, the theoretical value of the optical power loss is 32.4 dB when ɛ = 10. It is unacceptable in the optical design. So here we employ a receiver optical collimator with beam waist ω_T = 0.42 mm to couple the optical beam in free space into the fiber. The beam waist mismatch factor becomes ɛ = 0.42/0.65 = 0.65, which is within the scope [0.5, 2], and the theoretical value of the optical power loss is just 1.8 dB. The use of an appropriate receiver collimator keeps the theoretical optical loss of the system at an acceptable level, which makes the system simply and cost effective at the same time. The photograph of the mentioned transmitter collimator and receiver collimator are also shown in Fig. 2. It is noting that, in the actual operation the optical beam is distorted as transmitting in the free space that may introduce extra coupling loss besides the theoretical calculation.

2.2 Algorithm design

Based on the research results in [8], it takes 3.5 ms and 56.2 ms to generate a 1-to-2 broadcasting holograms by using splitting-SLM and GS algorithm with the same computer configuration, respectively. In our previous work [13], we proposed the RSS algorithm to generate the 2D broadcasting holograms. By applying the same phase function in every separated area and rotating the generated whole hologram for an arbitrary angle, the output beams can be steered with calculated angles in two directions. In this work, the LCoS active area is proposed to be divided into N sections and each section can be programmed independently to realize 1-to-N-points continuous tunable broadcasting. Here, we take N=4 for example as shown in Fig. 3. If the grating pitches for each sub-hologram are the same, the generated multiple beams have the same steering angle α. After a free-space transmission link, the beams are distributed on a circle with radius of R. In Fig. 3(a), the linear phase function of the separated areas is set with the same value, which means the value of R is the same for every divided optical beams. So the beams are distributed circularly in the far field as shown in Fig. 3(b). The grating pitches for the N independent holograms are adjusted and the whole N hologram are entirely rotated by an angle θ as shown in Fig. 3(c). Thereby, the generated optical beams are all rotated by the same angle θ and distributed circularly with different radius as shown in Fig. 3(d). The divided optical beams have a fixed angle spacing with the adjacent beams, which requires the relative positions of the nomadic users fixed.

 

Fig. 3. Principle of the modified rotated-splitting-SLM (RSS) approach that supports 2D continuous tunable broadcasting.

Download Full Size | PDF

Here, we improve the RSS algorithm to realize arbitrarily distributed optical beams at the receiver side. By applying the independent areas with different phase function and rotating the independent hologram for arbitrary angle, the value of R and θ for an individual beam are arbitrarily set and have no relationship with the others. As shown in Fig. 3(e), by rotating the generated hologram by an angle θ in the red area, the corresponding optical beam is also rotated by the same angel as shown in Fig. 3(f). By changing the grating pitch of the hologram in the red area in Fig. 3(g), the value of R is going to change accordingly as shown in Fig. 3(h). Thereby, a 1-to-N-points broadcasting system with arbitrary beam distribution can be achieved.

Figures 4(a) to 4(e) show the images for different broadcasting schemes based on the modified RSS algorithm. The number of the divided beams are up to 16. We draw the same line phase distribution in the separated LCoS active areas. The measured images illustrated in Fig. 4 are based on near-field principle by using an infrared CCD (OPHIR Photonics, BM-USB-SP907-1550-OSI). The CCD is placed at the focus of lens 2 so that all the sub-beams are collected and captured. As shown in Figs. 4(f) to 4(j), the corresponding one, two, four, eight, and sixteen optical beams are distributed on a circular. As shown in Figs. 4(k) to 4(n), we fix the phased distribution of the three independent LCoS active areas and rotate the hologram of the fourth active area by 45°, 135°, 225° and 315° compared to the hologram shown in Fig. 4(c). The independent beam spots are also correspondingly rotated by 45°, 135°, 225° and 315° as shown in Figs. 4(p) to 4(s). In Fig. 4(o), by adjusting the grating pitch of the chosen area into about two times value compared to Fig. 4(m), the corresponding independent beam spot shown in Fig. 4(t) is about two times far away from the center as compared to Fig. 4(r).

 

Fig. 4. Principle of the RSS algorithm that supports different broadcasting cases: (a) ∼ (e) Generated different hologram images with same phase function in every independent area; (f) ∼(j) detected beam spots which distributed on a circular with same radius. (k) ∼ (n) Generated different hologram images by rotating one independent area for 45°, 135°, 225° and 315°; (o) Generated hologram image by changing one independent area with double grating pitch; (p) ∼(t) detected beam spots when changing the hologram image of one independent area arbitrarily.

Download Full Size | PDF

An extra beam appears in the center of the beam distribution circular and its intensity is further increased compared with the individual beam as the broadcasting number increasing. This unsteered beam includes two parts: one is caused by the polarization characteristics of the LCoS device as the polarization of the input optical beam is not set perfectly. The other one is caused by limitation of the diffractive device. When LCoS device is employed for beam steering, it acts like a diffractive grating and the zero order light will appear in the illumination direction of the input beam. Based on the above two reasons, a part of the beam is unsteered and still appears in the center of the optical path. Because our system is a broadcasting transmission system, the energy from the unsteered beam will not bring crosstalk among the MDs at receiver side.

To verify the effectiveness of the proposed algorithm, we carried out another experiment to measure the diffraction efficiency. The unsteered beam is launched into free space and detected by a space optical power meter (Coherent, FieldMaxII-TOP). The total transmitted optical powers to free space for broadcasting 1, 2, 4, 8, and 16 beams are kept to 10 mW. The optical powers of the unsteered beams are measured for these five broadcasting schemes, whose values are about 0.9 mW. Thus, as increasing the broadcasting number, the optical powers of the unsteered beam keep almost unchanged. The diffraction efficiency is calculated as (10mW-0.9 mW)/10mW=91%. Thereby, the increasing of the broadcasting number makes very small impact on the desired multi-beams generation. The results show the proposed modified RSS algorithm used in our transmission system can provide flexible and stable operations for different broadcasting schemes.

3. Experiment

The experimental setup of the proposed multi-access-points optical wireless broadcasting system is illustrated in Fig. 5(a). A laser source at operating wavelength of 1550 nm is employed as optical carrier. A bit sequence with length of 2¹⁶ is generated and modulated to digital PAM-4 signal using the MATLAB program. During this processing the digital signal is oversampled by 2 times. An arbitrary waveform generator sampling at 92 GSa/s is employed to produce the 46 Gbaud/s analog PAM-4 signal. The generated analog PAM-4 signals are used to drive the Maher-Zender modulator. After transmitting over a 1-km SSMF, an Erbium Doped Fiber Amplifier (EDFA-1) is used to amplify the optical power to appropriate value for different transmission schemes. The modulated and amplified optical signals are launched into free space via a fiber collimator and then incident onto the LCoS. Since LCoS chip is sensitive to the polarization of input light, a single mode fiber polarization controller is introduced. In our experiment, the LCoS device produced by HOLOEYE Company has 1080×1920 resolution and fill factor of 93%. The maximum launched optical power to free space for each beam is set to 10 dBm for the eye safety limitation. The beam waist of the transmitter fiber collimator is 6.5 mm which matches the LCoS active area. By loading different hologram images generated by modified RSS algorithm mentioned in Section 2, the LCoS can split the input optical beam into multiple ones and steer them with different angles. An AM consisting of a pair of focusing lens is introduced to expand the FoV to ±15°. The angle expended optical beam is transmitted over 1.2 m free space distance to the user side, which results in the coverage are of 0.32 m². Receiver fiber collimators with small beam waist of 0.42 mm are employed to couple the light beams into SSMFs. Another EDFA (EDFA-2) is used to amplify the received optical signal for further PD detection because no electrical amplifier is adopted in our experiment. In our transmission experiment, the optical power at the receiver side for the worst situation is about −20dBm due to the beam aberration at the largest steering angle. Such power level is not enough for performance evaluation of the 92Gbit/s PAM-4 signal. It is commonly to introduce a pre-amplify before the PD detection [5]. The analog PAM-4 signals are then converted and sampled into digital signals by a digital storage oscilloscope (DSO) running at 160 GSa/s for offline signal processing. Figure 5(b) shows the experimental field, where a reflector mirror is employed due to the limited length of the experimental platform.

 

Fig. 5. (a) Experimental setup of the proposed multi-access-points OWC system and (b) the experimental field (the transmitter and receiver are not shown in the picture). AWG: arbitrary waveform generator; MZM: Maher-Zender modulator; SSMF: standard single mode fiber; EDFA: Erbium doped fiber amplifier; PC: polarization controller; PD: photodiode; DSO: digital storage oscilloscope.

Download Full Size | PDF

4. Results and discussions

The total path loss for PtP transmission scheme is firstly investigated and measured at the steering angles of −15°, −10°, −5°, 0°, 5°, 10° and 15°, respectively. As exhibited in Fig. 6(a), the measured corresponding losses are 28.2 dB, 25.1 dB, 21.3 dB, 17.3 dB, 21.3 dB, 24.5 dB and 28.7 dB, respectively. As enlarging the steering angle, the optical power loss is increased due to the aberration introduced by the AM configuration. Because the divergence of optical beam profile at the steering axis becomes more serious as increasing the steering angle, which brings extra optical power loss. The total path loss for the PtMP transmission scheme is then measured. The splitting loss (12 dB for the Pt16P case) is not included in calculation. As exhibited in Fig. 6(a), the measured losses are 29.1 dB, 26.9 dB, 24.3 dB, 25.5 dB, 28 dB and 30.2 dB at the steering angles of −15°, −10°, −5°, 5°, 10° and 15°, respectively. The optical power losses have the similar tendency as the PtP case. The steering angle of 0° is not addressed for the PtMP transmission case, because the measured results are influenced by the unsteered beam as discussed in Section 2. We also measured the total path loss as a function of different broadcasting numbers at the steering angle of 15°. As displayed in Fig. 6(b), the optical power losses are 28.7 dB, 28.4 dB, 28.6 dB, 29.3 dB and 30.2 dB when the broadcasting numbers are set to be 1, 2, 4, 8 and 16, respectively. The largest power fluctuation for the five broadcasting schemes is 1.8 dB, which shows the proposed modified RSS algorithm used in our transmission system can provide flexible and stable operations within the FoV of ±15°.

 

Fig. 6. (a) Measured optical power loss versus different steering angles; (b) Measured optical power loss versus different broadcasting number when steering angle is 15°.

Download Full Size | PDF

Figure 7 shows the variation curve of BER performance of the 92-Gb/s PAM-4 signal with the received optical power for the different broadcasting schemes. The optical power launched to free-space for PtP cases is amplified to 10 dBm by EDFA-1 at the transmitter. The theoretical splitting losses for two, four, eight, and sixteen points are 3 dB, 6 dB, 9 dB and 12 dB, respectively. Taking eye safety requirement into consideration, the launched optical powers are amplified to 13dBm, 16dBm, 19dBm and 22dBm by EDFA-1 for broadcasting for two, four, eight and sixteen users, respectively. Figure 7(a) and 7(b) show the BER performance of the PtP and Pt16P transmission cases versus received optical power at the different steering angles. When the received optical power reaches more than 3 dBm, the achieved BER values are all below the 7% FEC limit at the BER level of 3.8×10⁻³. As increasing the steering angles, the BER performance becomes worse. This is mainly caused by gain nonlinearity of EDFA-1 and EDFA-2 [17]. The different broadcasting schemes have the similar BER performance at the steering angle of 15° as illustrated in Fig. 7(c), which indicates the validity and feasibility of the high access-point optical wireless broadcasting system when serving different number of nomadic users. The demodulated and recovered PAM-4 eye diagrams for broadcasting 16 users at the largest steering angle of 15° are also presented in Fig. 7(c).

 

Fig. 7. BER performance of 92-Gb/s PAM-4 signal versus received optical power (ROP) for different broadcasting schemes. (a) Point to point transmission with different steering angles, (b) point to 16 points transmission with different steering angles, (c) different broadcasting shames when the steering angle is 15 degrees. Inset: recovered eye diagrams when ROP=8dBm, BER=7.44×10⁻⁴; ROP=7dBm, BER=1.14×10⁻³; ROP=6dBm, BER=1.8×10⁻³; ROP=5dBm, BER = 3.5×10⁻³.

Download Full Size | PDF

Table 1 shows the comparison of the three different broadcasting technologies including crossed gratings, LCoS with GS algorithm and LCoS with modified RSS algorithm. By introducing an active SLM for programmable broadcasting, we have high scalability to expand the number of access users and flexible steering ability of 2D arbitrary tuning. And by using the modified RSS algorithm, the steering speed of the optical beams is improved compare to the GS algorithm. In our work, by using the designed optical path with two matched collimators for both transmitter and receiver, the system complexity is significantly reduced.

Table 1. Comparison of multi-beam optical wireless broadcasting technologies.

View Table

5. Conclusions

A programmable 2D optical wireless system with high-capacity and multi-access-points has been investigated by using a commercial LCoS-SLM and modified RSS algorithm. An AM configuration and a pair of designed fiber collimators are introduced to expand the FoV and reduce the system complexity. Based on the above optical design and algorithm, a continuous tunable broadcasting system with ±15° FoV for up to 16 nomadic users has been experimentally demonstrated and successfully achieved over 1-km SSMF and 1.2-m indoor free space link. The optical beams can be generated and distributed according to the users’ locations. In our demonstration, each beam is modulated with 92 Gb/s PAM-4 signal, offering a total wireless capacity of beyond 1.47 Tb/s at the BER value of less than 10⁻⁴. The experimental results reveal that the proposed system is feasible to the future indoor high-speed dynamic networks.