1. Introduction

The upcoming beyond fifth-generation/sixth-generation (B5G/6 G) communication systems are expected to deal with enormous advances such as virtual reality (VR) and augmented reality (AR) applications, video streaming services, high-performance computing applications, and other technologies, which require higher data rates, lower latency, and more reliable connections. Aside from this, the Internet of Things, comprising a large amount of end-user devices and sensors, also raises a higher demand for the performance of communication systems [1–3]. Traditional radio-frequency (RF) communication is approaching its limit due to its limitations in bandwidth and range, susceptibility to interference, and security issues. By comparison, optical wireless communication (OWC) offers a number of advantages, including high security, strong anti-jamming ability, high bandwidth, energy efficiency, abundantly available spectrum, and high bit rates that can reach up to gigabits and even terabits, which makes it a promising candidate for serving the requirements of B5G/6 G communication systems [4–6]. Visible light communication (VLC) and infrared light communication (ILC) are the two primary methods for OWC. VLC systems usually use low-cost and low-power LEDs to generate optical signals and illuminate their surroundings. They have been studied in different indoor/outdoor/underwater applications [7–10]. However, they are constrained by limited modulation bandwidth, reach and capacity, and the requirement for illumination to be switched on, which may not be desirable in all situations. In contrast, ILC provides several benefits, including ultra-high capacity, strong privacy, eye safety, and excellent power efficiency, leading to high data rates and long reach. Additionally, ILC is capable of direct access to the current mature optic fiber network infrastructures. Yet, resulting from the narrow radiation angle of the infrared light beam, ILC suffers from the issue of limited coverage area and difficulty in establishing a point-to-point connection where line-of-sight (LoS) is needed [4]. Recently, researchers have implemented many beam-steering devices in transceivers to expand the coverage area [11–14] and developed numerous kinds of acquisition tracking pointing (ATP) systems to precisely steer the beams to the terminal [15–21]. But the terminal is still facing challenges in achieving a wide field-of-view (FOV) to reduce the difficulty in alignment, high bandwidth, small size, weight, and power (SWaP) design, and high modulate rate for upstream signals to realize high-speed bidirectional all-optical communication. In this paper, FOV refers to the angular area that can receive the signal, sharing a similar meaning with angle-of-arrive.

For the receiver part of the terminal, there is a trade-off between the photodetector (PD)'s light collection area (LCA) and its bandwidth. The small LCA could help reduce the PD’s conjunction capacitance, thus reaping high bandwidth. On the contrary, the large LCA contributes to expanding the FOV of the receiver, which benefits obtaining more optical power and mitigating the difficulty of setting up LoS link between the transceiver and the terminal. In 2020, a fluorescent-fiber-based receiver was proposed in [22] that possesses a FOV of 240°. However, due to the long fluorescence lifetime, this type of receiver has limited bandwidth and low power efficiency. A wide FOV of 30° at a data rate of about 120 Mbps was yielded by an angle diversity receiver deploying multiple PDs in [23]. Also, many researchers leverage the matrix of PD to upgrade the FOV of the receiver [24–26]. In 2020, Koonen et al. experimentally demonstrated 1 Gbps on-off keying (OOK) transmission at a FOV of 10° [24]. In 2022, Soltani et al. designed a receiver that achieved data rates of about 23 Gbps with a FOV of 15° and 8 Gbps with a FOV of 20° by simulation [25]. In 2022, Umezawa et al. carried out an 8.5 GHz frequency response with a FOV of 6° [26]. However, the combination of PDs would be troubled by high cost, complex electronics, and great noise. Besides, in 2022, Pham et al. reported a receiver with a FOV of 1.2° at the data rate of 1 Gbps by means of utilizing the motorized actuator to control the orientation of the receiver [27]. In addition, the wide-FOV optics structure has attracted lots of attention from many researchers [28–30]. In 1993, the holographic lens was designed to image incoming laser radiation at 850 nm onto a solid-state detector array over a FOV of 45°×32° (53.4° in the diagonal direction) by Tai et al [28]. In 2010, Hahn et al. proposed a receiver design employing a microlens array to couple the incident optical signal into an individual fiber in a bundle routed to remote optical detectors [29]. However, no communication experiments were conducted for the above two works. In 2021, Alkhazragi et al. introduced a detector with a FOV of 65° at the data rate of 1 Gbps through the usage of a fused fiber-optic taper [30]. Table 1 summarizes the recent research progress of receivers in terms of transmission distance, FOV, and data rate. Obviously, the existing receiver structure is still having issues of either limited FOV, low bandwidth, or violating the eye-safety standard (IEC60825 and US-oriented variant ANZIZ136). And also, deploying these receiver structures, the terminal is required to be equipped with an extra light source and massive ATP systems to generate and send upstream optical signals to the transceiver [21]. The optics-single-mode-fiber (optics-SMF) antenna is a promising method that could solve these problems effectively and is expected to be used in mobile terminals. By coupling the optical signal into the SMF through the wide-FOV optics, the FOV of the terminal will be greatly improved. Also, high-speed detectors could be adopted to improve the bandwidth of the receiver. Moreover, it is possible for high-speed, small SWaP, and energy-efficient transmitter part, such as a fiber-type modulating retro-reflector (MRR) system, to be employed in the terminal, which can greatly upgrade the data rate of upstream signals and eliminate the light source and ATP system in the terminal by sending downstream light back to the transceiver in the co-propagation path after modulation, thus reducing size, weight and power consumption. However, the present fiber-type retroreflective systems always use a collimator with narrow FOV and place them with full alignment [31–34].

Table 1. Comparison of wide-FOV and high-speed receiver in recent works

View Table

In this paper, we proposed and experimentally demonstrated a wide-FOV auto-coupling optical antenna system for high-speed bidirectional optical communication. Employing a wide-FOV telecentric lens, fiber array, optical switch, 3D motorized linear stage, and angle sensor, the proposed antenna is capable of automatically coupling the downstream optical signals into a SMF and transmitting the upstream optical signals from it in the sharing propagation path within a wide FOV of 100°. For both downstream and upstream optical signals, the maximum coupling loss of the antenna is simulated down to 8.1 dB, manually optimized down to 9.3 dB, and automatically measured down to 10.6 dB between -50° and +50°. The MRR system is used in the terminal of the bidirectional OWC system, where the upstream signal is modulated on the reflected optical carrier at the terminal. Eye diagrams and bit-error-rates (BERs) are measured for both downstream and upstream signals within the whole FOV, confirming the validation of the presented auto-coupling antenna system. The experimental results show that the proposed auto-coupling antenna system supports 10-Gbps downstream OOK signals and 10-Gbps upstream OOK signals both with FOV of up to 100° and has a comparable coupling capacity in C band. To the best of our knowledge, this is the first reported auto-coupling optical antenna structure, which has the largest FOV for ILC automatic alignment works in terminals.

2. Principle

The diagram of the wide-FOV auto-coupling optical antenna system is shown in Fig. 1(a). The downstream beams incident onto the Rx optics at different angles and converge to a spot on the focal plane. Meanwhile, the angle sensor attached to the Rx optics detects the vertical and horizontal components of the incident angles ${\theta _V}$ and ${\theta _H}$ and sends the angle information to a controller. The $N \times N$ two-dimensional (2D) fiber array with adjacent channels distance ${D_{CH}}$ is placed on a three-dimensional (3D) motorized linear stage and the fiber enfaces are placed on the focal plane of the lens. A ${N^2} \times 1$ optical switch is connected with this fiber array. After that, a photodetector PD1 is connected, which works as a feedback power monitor. The controller actuates the 3D motorized linear stage to couple the focal spot into its nearest fiber channel of the fiber array and opens up the corresponding channel of the ${N^2} \times 1$ optical switch according to the information of incident angle and coupling power intensity acquired from the angle sensor and PD1 respectively.

 

Fig. 1. (a) The diagram of the proposed optical auto-coupling antenna system; (b) The configuration of the Rx optics (i.e., telecentric lens), the ray tracing of incident beams and corresponding spot diagrams on the image surface at incident angles of 0°, 25°, 50°, and 60°; (c) The lateral position of focal points for different incident angles (${\theta _H}$) on the focal plane of the telecentric lens; (d) Auto-coupling process for the proposed optical auto-coupling antenna system.

Download Full Size | PDF

To reduce the complexity of the auto-coupling optical antenna system and improve its coupling efficiency and FOV, we choose the wide-FOV telecentric lens with a numerical aperture (NA) matching SMFs as Rx optics of the antenna. The telecentric lens is designed partially based on a specific collimator utilized in our transceiver by using the ray tracing software Zemax. The beam diameter of this collimator will diverge to about 2 mm at a distance of 3 m from collimator housing, matching the entrance pupil diameter of the lens. Finally, the designed lens employing a four-piece combination and fabricated from a material denoted as H-ZF52A has a back focal length of 7.55 mm and an entrance pupil diameter of 2 mm. The detailed design of the telecentric lens is presented in our previous work [35]. It is obvious that, after beams at different incident angles pass through the telecentric lens, the main rays of the image are all parallel to the optical axis. And the SMFs of the 2D fiber array are evenly and vertically distributed on the focal plane, thus also parallel to the optical axis. In this case, the angular deviation of all fiber channels can be corrected simultaneously after assembling this system, and we only need to consider the axial and radial deviations during the auto-coupling procedure. The configuration of the telecentric lens we used in this paper is shown in Fig. 1(b), as well as the ray tracing and spot diagrams at incident angles of 0°, 25°, 50°, and 60°. In spot diagrams, high-ratio spots are distributed within the Airy disk at incident angles of 0°, 25°, and 50°, indicating great focusing capacity. And at 60°, the ratio decreases a little, and minor coma is introduced. To acquire the fitted curve of X-axis coordinates $X({{\theta_V}} )$ and Y-axis coordinates $Y({{\theta_H}} )$ of focal spots, we build a coupling system shown in Fig. 2(b). In the experiment, we rotate the rotatable platform and move the motorized linear stage to the position where the maximum coupling ratio is achieved. In this way, the angle and position are recorded sequentially. As is shown in Fig. 1(c), the simulation and experiment share identical results of the lateral distance of these spots, which are within 16 mm between -60° and +60°. This good coincidence also verifies the accuracy of the optical assembly. The combination of the fiber array and optical switch is designed to shorten the radial distance the 3D motorized linear stage needs to move, that is the distance between the focal spot and corresponding coupling fiber, thus accelerating the coupling rate and reducing the size of the antenna. And the actuation of the Z axis could help compensate for the axial error caused by optical aberration. Furthermore, the design of the whole antenna structure guarantees that upstream beams could be transmitted to the transceiver while sharing the propagation path with downstream beams, and therefore prevents their energy from being dispersed into other fiber channels.

 

Fig. 2. (a) The schematic of the bidirectional OWC system using auto-coupling optical antenna system; (b) The experimental setup and free space optical path figure of the OWC system; (c) The figure of the auto-coupling optical antenna system. LD: laser diode; PRBS: pseudo-random binary sequence; MZM: Mach-Zehnder modulator; EDFA: erbium-doped-fiber amplifier; OC: optical coupler; VOA: variable optical attenuator; BPF: bandpass filter; PD: photodetector; BERT: bit error rate tester; OSC: oscilloscope; SOA: semiconductor optical amplifier.

Download Full Size | PDF

The auto-coupling process is illustrated by a flowchart shown in Fig. 1(d). When communication is lost, that is PD1 cannot detect any light intensity, the controller starts to run the auto-coupling program. In the first step, the angle information ${\theta _V}$ and ${\theta _H}$ is obtained by the controller from the angle sensor. Then, the controller will calculate the position $({X({{\theta_V}} ),\textrm{}Y({{\theta_H}} )} )$ of the focal spot based on the pre-calibrated results (Fig. 1(c)) on the focal plane, the channel coordinate $({{N_X},{N_Y}} )$ of its nearest fiber, and the required x- and y-directional shift $({{D_X},{D_Y}} )$ of the fiber array, which can be obtained by the equation:

After that, if the calculated shift ${D_X}$ and ${D_Y}$ does not exceed the limits of the linear stage, the controller will open up the calculated fiber channel of the optical switch, actuate the motorized linear stage (the calculated fiber channel) to scan within an area around the focal spot position $({X({{\theta_V}} ),\textrm{}Y({{\theta_H}} )} )$ with coarse step sizes. Typically, the scanning range is not greater than the fiber channel spacing. Once arriving at nearby the focal spot, the controller starts to scan around the position with fine step size and acquire coupling power intensity through PD1 after every step. Eventually, the program will end up finding the maximum coupling power intensity through the gradient descent algorithms.

3. Experimental setup

Figure 2(a) shows the schematic of the bidirectional OWC system using an auto-coupling optical antenna system. The light source is provided by a continuous-wave (CW) laser with optical power of 13.23 dBm at 1550 nm. For downlink communication, it is modulated by 10-Gbps OOK sequence generated by a pseudo-random binary sequence (PRBS) generator via a Mach-Zehnder modulator (MZM1) at the transmitter. The modulated optical signals are then amplified by means of an erbium-doped-fiber amplifier (EDFA) and launched into free space through circulator1 and a fiber collimator used for beam narrowing. Note that as an indoor OWC system, the optical power at point A is 9.5 dBm and a free-space transmission distance of 3 m is conducted in this paper. After 3-m transmission, the incident beam impinges on the antenna at an angle of θ. Due to the limited length of the optical table, the transmission distance is prolonged to 3 m by two high-reflectivity mirrors (M1 and M2) in our experiment (Fig. 2(b)). Also, a customized telecentric lens is positioned at the center of a rotatable platform to simulate different incident angle θ. It is noted that, to simplify the experimental setup, we simulate various incident angles by rotating the antenna system in the YZ plane, so only ${\theta _H}$ has been changed in our experiment. Considering the symmetry of the lens, it is believed that our setup also works for various ${\theta _V}$.

Then the auto-coupling optical antenna system collects the incident beams using the telecentric lens and searches for the status with maximum coupling efficiency. This is done by opening up the calculated fiber channel of the optical switch and actuating the fiber array placed on the motorized linear stage to the corresponding position based on the incident angle information ${\theta _H}$ and the feedback of the coupling power obtained by PD1. In this experiment, we utilize an attitude sensor to obtain the angle ${\theta _H}$, acting as an angle sensor. A 1*15-channel fiber array with an adjacent spacing of 1 mm to couple these signals. The details of the auto-coupling optical antenna system setup are shown in Fig. 2(c). After that, the downstream signal is coupled to the terminal.

At the terminal, 10% power of the downstream optical OOK signals is pre-amplified, filtered, and sent into a photodetector (PD2), where the eye diagram is observed using an oscilloscope (OSC) and the BER is measured using a BER tester (BERT). The other 90% power of the downstream optical signals is sent into another MZM (MZM2) and modulated by a 10-Gbps OOK sequence as the upstream signals. The upstream OOK signal is then amplified to about 12 dBm by a semiconductor optical amplifier (SOA) and launched into free space through circulator2 and the antenna system. As the measured loss of the optical switch is about 3 dB, the power of upstream signals in free space remains within the eye safety limits. And the retro-reflected optical signals (upstream OOK optical signal) transmit back to the transceiver through the same free-space optical propagation path and are pre-amplified, filtered, and detected by another photodetector (PD3). The eye diagram and BER of the upstream signals are also measured. The upstream and downstream communication operate in a time-division multiplexing, half-duplex mode. Two bandpass filters (BPFs) are placed before PD2 and PD3 to reduce the noise, and two pre-EDFA are utilized to amplify the optical signals to about 0 dBm in order to reduce the impact of power loss on the quality of the detected signal. Additionally, the optical power of the received signals can be tuned by the variable optical attenuator (VOA) and then we measure the received power after VOA and corresponding BER via PD2 and PD3 to evaluate the performance of the OWC system.

4. Experimental results

To experimentally measure the manually optimized, and automatically coupling loss of both downstream and upstream signals at different incident angles, we place a light source at point A(B), and measure the optical power at point B(A) (see Fig. 2(a)). The coupling loss is defined as the ratio of power measured at point B(A) to the power measured at point A(B).

Figure 3(a) shows the simulated and experimentally measured coupling loss of the downstream signals. The simulated coupling loss conducted by the FICL function in Zemax is down to 8.1 dB between -50° and +50°, 9.4 dB between -60° and +60°, and 14.6 dB between -65° and +65°. The manually optimized coupling loss is measured down to 7.8 dB between -25° and +25°, 9.3 dB between -50° and +50°, and 20 dB between -60° and +60°, slightly larger than the simulated results. The auto-coupling loss is measured down to 8.0 dB between -25° and +25°, and 10.6 dB between -50° and +50°, with the maximum difference being 1.3 dB in comparison with manual optimized results. The coupling loss of the upstream signals is depicted in Fig. 3(b), which is comparatively the same as downstream signals shown in Fig. 3(b), and beyond ±50°, the energy attenuation soars for both cases.

 

Fig. 3. The simulated, manually optimized, and automatically coupling efficiency of the proposed optical antenna system versus different incident angles for (a) downstream signal and (b) upstream signal; (c) The voltage of the PD1 versus different coupling efficiency.

Download Full Size | PDF

For simulated results, the efficiency at positive and negative incident angles are identical, while for experimental results, they are not as symmetrical as the simulated ones. In general, the experimental results and simulation results fit well. The deviation could result from the assembly error during lens fabrication, misalignment between the incident beam and the pupil of Rx optics, the divergence of the beam caused by the adjustment error of the collimator, and distortion of the wavefront after two reflections.

For automatic results, the coupling loss is slightly larger than the manually optimized results. Besides, beyond ±50°, the auto-coupling system cannot identify the optimal status, thus making it difficult to establish communication between the transceiver and the terminal. Several reasons are responsible for these unpleasant circumstances. For one thing, due to the open-loop design, the linear stages of the X-axis and Z-axis have about 20% error between the input and actual step size. In this case, the linear stages are unable to arrive at the calculated position in every loop of the gradient descent algorithms, and cumulative error will make the result worse after several loops. For another, the intensity value acquired by the analog-to-digital converters (ADC) also has some errors. While executing the gradient descent algorithms to find the position with maximum coupling efficiency, a large light intensity difference between two adjacent moving steps could help reduce the impact of these errors. The output voltage of the PD1 at each different coupling efficiency is also measured. As is shown in Fig. 3(c), when the coupling efficiency is larger than -10 dB, the voltage changes more drastically than the one changes at the coupling efficiency less than -10 dB, leading to the excellent performance of the auto-coupling optical antenna system.

Figure 4(a2)-(a11) present the manually and automatically measured eye diagrams of the 10-Gbps downstream signals at different incident angles. One can see that the eye-opening is almost the same not only in the manual and automatic results at the same angle but also at different angles between +50° and -50°. As a comparison, direct communication results through the optical fiber back-to-back (B2B) are presented in Fig. 4(a1). In general, the noise introduced by free space is extremely low and signals show high quality.

 

Fig. 4. (a) Eye diagrams of the 10-Gbps downstream signals in B2B link and manually and automatically coupling at different incident angles (16.4ps/div); (b) BER results of 10-Gbps downstream signals as received optical power varies; (c) BER results of 10-Gbps manually and automatically coupling downstream signals as incident angle varies with a received optical power of -28 dBm.

Download Full Size | PDF

The BER performance of the manually and automatically coupling downstream optical signal transmitting at different incident angles and the B2B results for comparison are shown in Fig. 4(b), where the least required received optical powers for the manually and automatically coupling downstream signal between +50° and -50° are -33.00 dBm, -33.44 dBm respectively, and for B2B is -34.20 dBm at the 7%-overhead hard-decision forward error correction (HD-FEC) limit of BER @ 3.8 × 10⁻³. Also, between +50° and -50°, the same performance for manual and automatic results at each incident angle is achieved in the system, as well as their results at different incident angles, which are all similar to the B2B performance. We could draw the same conclusion as in the eye diagram results that our proposed optical antenna whether in manually or automatically coupling conditions bring little noise to the downstream signals.

We also compare the BER performance of the manually and automatically coupling downstream optical signals at a data rate of 10 Gbps at different incident angles with a fixed received optical power of -28 dBm. As is shown in Fig. 4(c), for angles between +60° and -60°, BERs are between 1 × 10⁻⁸ and 1 × 10⁻⁶ for manually coupling results. And for angles between +50° and -50°, the BER performances are almost the same in manually and automatically coupling cases which are below 1 × 10⁻⁶. The subtle difference in BER performance at different angles could be attributed to the polarization change caused by the fiber channel change and the drift of the MZM1’s working point.

Next, we explore the uplink performance of the proposed auto-coupling optical antenna system. Figure 5(a1)-(a11) present the eye diagrams of the 10-Gbps manually and automatically coupling upstream optical signals at different incident angles and B2B result for comparison. It is seen that a few jitters are introduced in the free-space link as compared to B2B link due to the noises introduced in the optical paths such as the reflected lights from the surface of the fiber array. Figure 5(a2)-(a11) show that the difference between manually and automatically coupling results at each angle are relatively minor, indicating that coupling errors brought by the automatic antenna have little impact on the quality of upstream communication. Besides, jitters at some angles are more drastic than others, which owes to the instability of our setup and that the reflected lights at certain angles from the fiber end-face are more intense than that at other angles, deteriorating the signal-to-noise ratio (SNR) of the retro-reflected signal.

 

Fig. 5. (a) Eye diagrams of the 10-Gbps upstream signals in B2B link and manually and automatically coupling at different incident angles (16.4ps/div); (b) BER results of 10-Gbps upstream signals as received optical power varies; (c) BER results of 10-Gbps manually and automatically coupling downstream signals as incident angle varies with a received optical power of -28 dBm.

Download Full Size | PDF

Figure 5(b) shows the BER performance of the manually and automatically coupling upstream optical signals at a data rate of 10 Gbps at different incident angles, and the B2B link is also provided as a reference. We can see that the least required optical received powers for the manually and automatically coupling upstream signal at angles between +50° and -50° are -32.69 dBm and -30.97 dBm respectively, and for B2B is -35.19 dBm at the BER of 3.8 × 10⁻³. As is the same as the eye diagrams results, the performance of manually and automatically coupling results are similar at each angle, indicating the low noise of our proposed auto-coupling antenna. In addition, the BER performance varies at different angles resulting from different jitter conditions and intensity of reflected light power from the fiber end-face, which can strongly disturb the upstream optical signals and decrease the SNR of the upstream signals.

The BERs of manually and automatically upstream optical signals at data rates of 10 Gbps are also measured with a received optical power of -28 dBm as the incident angle varies. As is shown in Fig. 5(c), the BERs of the manually coupling upstream signals measured are below 1 × 10⁻⁵ for angles between +60° and -60°, and below 1 × 10⁻⁶ between +50° and -50°. Clearly, the performance degrades as angles become larger outside +50° and -50°. Besides, for angles between +50° and -50°, on the whole, the BERs of automatically coupling results are nearly the same as the manually coupling results, slightly worse in general. This circumstance is due to the drop in coupling efficiency caused by larger incident angles and automatically coupling errors, leading to a decrease of SNR.

Overall, Fig. 4(c) and Fig. 5(c) manifest a wide FOV of up to 100° for our proposed auto-coupling system and 120° for our telecentric lens in it within which the 10 Gbps bidirectional OWC can be realized in high quality with a fixed received optical power of -28 dBm.

In addition, we conduct an extensive evaluation of the BER and coupling efficiency performance for both downstream and upstream signals in the whole C band. Our evaluation comprises the free space links with our proposed optical auto-coupling antenna system within the entire auto-coupling FOV of ±50° and B2B link scenarios, with a fixed received optical power of -28 dBm. Figure 6(a)-(e) shows that at each incident angle, manually and automatically coupling results share comparable performance for both downstream and upstream signals over the entire C band. A good performance appears in the wavelength range of around 1543 nm to 1557 nm, within which the performance of longer wavelengths is slightly better than the shorter ones. Moreover, for the wavelength outside the range of 1543 nm to 1557 nm, the BER performance somewhat declines. To further explore the reason for the deterioration of BER performance outside the wavelength range of around 1543 nm to 1557 nm, we measure the manually optimized and automatically coupling efficiency at different angles versus different wavelengths in C band for downstream signal and upstream signal. It can be obviously seen in Fig. 6(f) and (g) that, within the angles between -50° and +50°, a similar coupling performance of them is achieved in C band for both downstream and upstream signals, suggesting that the degradation of the BER performance is unrelated to our proposed optical auto-coupling antenna system. The degradation may be due to the limited working band of amplifiers in our experimental setup, including EDFA, pre-EDFA1, pre-EDFA2, and SOA. The quality of the signals is poorly amplified outside the wavelength range of around 1543 nm to 1557 nm in C band. One can solve this problem by replacing the amplifier which shares the uniform amplification performance in C band, which points the way to further optimization for our OWC system working in the high-capacity WDM system.

 

Fig. 6. BER performance of manually and automatically coupling downstream and upstream signals at incident angles of (a) 0°, (b) 25°, (c) -25°, (d) 50°, and (e) -50°, and B2B results for reference at different wavelengths in C band. The manually optimized, and automatically coupling efficiency of the proposed auto-coupling optical antenna system at different angles versus different wavelengths in C band for (f) downstream signal and (g) upstream signal.

Download Full Size | PDF

5. Discussion

In the future, to further reduce the SWaP of this design, the 3D motorized linear stage can be replaced with miniaturized motors or microelectromechanical actuators. Also, the 2D fiber and the optical switch could be integrated on the same chip with planar lightwave circuits. Furthermore, the larger FOV and higher coupling efficiency could be realized by means of utilizing higher-precision 3D motorized linear stage, PD with better linear sensitivity, and ADC with improved precision. Additionally, to further increase the system capacity, employing advanced modulation formats like quadrature phase-shift keying (QPSK) signals [36] and multiple-dimensional multiplexing of orbital angular momentum [37,38], polarization [39], and wavelength [40] could be considered.

Moreover, as far as the convergence time of the algorithm, several constraints have impacted our ability to elaborate on this aspect, like the large open-loop step size error and the slow actuating speed of the used piezo-actuators for the X-axis and Z-axis, and the angle error acquired by the angle sensor. Under this circumstance, the convergence time is within several seconds and fluctuates at different operations. Moving forward, the convergence time would be improved to millisecond-level if we employ a 3D motorized linear stage with higher bidirectional repeatability and faster speeds like centimeters per second. Besides, an enhancement in angle sensor precision would result in a smaller scanning area with fewer scanning steps. In this case, the convergence time can be significantly decreased. In the experiment, we simplify the situation and use an angle sensor to capture the attitude of the terminal. For practical scenarios of angle detection, accelerometer, gyroscope and LiDAR [19] need be considered. The information collected from these sensors can be exchanged between the transceiver and the terminals through traditional RF communication, like Bluetooth, Wi-Fi, and 3 G/4 G/5 G Module for real-time beam steering adjustment of the base station and the adjustment of the terminal device. Actually, six/nine-axis accelerometer and gyroscope sensors have been widely employed in mobile phones and other portable devices and the present prototype is compatible with these sensors and devices for future usage.

In this paper, we have specialized the system for indoor communication applications and hence explored the communication performance within the distance of 3 m. Regarding longer transmission distances, the coupling efficiency might drop due to enhanced atmospheric loss, the greater divergence of beams, and the pointing error. In this case, a larger telecentric lens with a similar design rule can be considered. Concurrently, several approaches could be adopted to mitigate the effects caused by the atmosphere, including beam focusing and directional techniques, adaptive optics, and increasing the gain of optical amplifiers.

6. Conclusion

In this paper, we have proposed a novel wide-FOV auto-coupling optical antenna system and established a high-speed bidirectional OWC system based on this antenna. The proposed auto-coupling optical antenna system is built with a wide-FOV telecentric lens, fiber array, optical switch, 3D motorized linear stage, feedback PD and angle sensor, and is able to automatically couple the infrared beams into SMF within the wide FOV of 100°with coupling loss less than 10.6 dB in C band. The experimental transmission results have shown that we realize both 10-Gbps OOK downstream and upstream transmissions in C band. To the best of the authors’ knowledge, this brand-new auto-coupling optical antenna has the largest FOV to all ILC automatic alignment works in terminals. The demonstrated auto-coupling optical antenna presents a bright future in mitigating the difficulty in establishing a point-to-point connection for high-speed bidirectional OWC systems and is expected to be widely utilized in future indoor access networks.