Transmitters and Receivers for High Capacity Indoor Optical Wireless Communication

Abstract

In this paper, we present recent advancements in transmitter and receiver technologies for Optical Wireless Communication (OWC). OWC offers very wide license-free optical spectrum which enables very high capacity transmission. Additionally, beam-steered OWC is more power-efficient and more secure due to low divergence of light. One of the main challenges of OWC is wide angle transmission and reception because law of conservation of etendue restricts maximization of both aperture and field of view (FoV). On the transmitter side, we use Micro Electro-Mechanical System cantilevers activated by piezoelectric actuators together with silicon micro-lenses for narrow laser beam steering. Such design allowed us to experimentally demonstrate at least 10 Gbps transmission over 100° full angle FoV. On the receiver side, we show the use of photodiode array, and Indium-Phosphide Membrane on Silicon (IMOS) Photonic Integrated Circuit (PIC) with surface grating coupler (SGC) and array of SGC. We demonstrate FoV greater than 32° and 16 Gbps reception with photodiode array. PIC receiver allowed to receive 100 Gbps WDM with single SGC, and 10 Gbps with an array of SGC which had 8° FoV in the vertical angle and full FoV in the horizontal angle. Our results suggest that solutions presented here are scalable in throughputs and can be adopted for future indoor high-capacity OWC systems.

1. Introduction

Optical wireless communication (OWC) is a very promising candidate for the successor of existing radio frequency (RF) wireless technologies, for the sixth-generation (6G) telecommunication network [1]. The optical spectrum offers huge bandwidth, which enables high-capacity transmission. Very high data rates are required in indoor, terrestrial, and non-terrestrial wireless communication. All-optical networks offer low latency due to the avoidance of the expensive conversion of signals between the electrical and optical domains. Data transmission in a beam-steered OWC is highly secure and private since the signal is only sent when and where it is needed, unlike other existing technologies such as wireless radio communication or LiFi. Additionally, optical frequencies do not penetrate walls and other objects which minimizes chances of eavesdropping and allow denser frequency reuse.

Wavelengths used in OWC are less divergent than radio frequencies which makes them more energy efficient because less power is wasted during propagation in free space. Infrared light offers a unique set of advantages to the 6G network. Unlike visible light, infrared light (IR) can operate at a higher power level while being eye-safe, as it is predominantly absorbed by the cornea and does not reach the retina. In the case for the C-Band (1550 nm) wavelengths we use, this allows approximately 10 dBm of free-space optical power [2], significantly higher than the allowed for visible light. Moreover, IR’s not visible nature provides another layer of privacy and not disturbance as the users cannot see the beams. The narrow beam-steering OWC system matches the requirements of 6G with respect to high capacity wireless transmission, due to the very wide available bandwidth and the high density of users, due to dense frequency reuse. However, optical communication requires a line-of-sight between the transmitter and receiver, which can hinder transmission depending on the environment [3]. In such a case where there is no line of sight between the transmitter and the receiver, a hybrid system with OWC and RF would be useful. Users would still have a connection, albeit at a lower data rate.

In this work, we are targeting indoor beam-steered OWC which is characterized by a stable environment, because of which we assume no turbulence effects on the beam, and short ranges in the order of a few meters, as illusttrated on Figure 1. The main challenge of indoor OWC is the wide angle of transmission and reception. The transmitter must have a wide bandwidth for high-capacity data transmission and a wide and precise beam-steering range for alignment to the receiver. The receiver, on the other hand, must have a large aperture to capture as much light as possible, a wide field of view to receive light from various directions, and high bandwidth for high-speed data transmission [2].

This experimental research explores the innovative application of established optical communication theories to develop novel approaches in OWC. Focusing on the experimental validation and practical implementation of advanced OWC concepts, demonstrating how scientific principles can be transformed into new telecommunication technologies.

The paper is organized as follows. Section 2 discusses the transmitter for OWC with piezoelectric actuators, Section 3 presents the receivers for OWC such as the photodiode array and the photonic integrated circuit-based receiver, and finally Section 4 concludes the paper.

2. Beam-Steering for Optical Wireless Communications Transmitter

2.1. Introduction

One of the main challenges in OWC systems is achieving precise beam steering while maintaining high data rates and signal integrity. Established beam steering approaches show limited throughput, steering speeds, optical bandwidth, reliability, and miniaturization potential. In this section, we present the implementation of C-Band beam steering for an indoor OWC system using silicon micro-lenses integrated with piezoelectric Micro Electro-Mechanical systems (MEMs) actuators to achieve bidirectional beam steering with 51.3° horizontal and 22.4° vertical angular ranges, while maintaining an error-free transmission rate of 10 Gbps.

Although primarily designed for indoor environments such as homes and offices, the architecture can be adapted for terrestrial applications by adjusting the beam divergence and receiver specifications. Another advantage is that the data rate of the system is not dependent on the steering mechanism, being primarily limited by the bandwidth of transmitter components (modulators, drivers) and receiver characteristics (photodiode bandwidth, aperture size and field-of-view (FoV)).

2.2. State of the Art for Narrow Beam-Steering

Several techniques have been investigated for optical beam steering, each with distinctive characteristics and limitations, such as: Polarization gratings employ liquid-crystal imprinted with gratings to deflect circular polarized beams [4]. The gratings create a diffraction pattern that is switchable depending on the state of the liquid-crystal. Each grating has a fixed steering angle, but can be stacked to give a greater FoV. Although these systems benefit from low power consumption, their functionality is limited by fixed steering angles determined by the number of stacked gratings; they are also constrained by the circular polarized light for side lobe suppression.

Arrayed waveguide grating routers implement a passive steering mechanism by selecting outputs from a fiber array positioned in a lens’s focal plane [2]. This approach enables efficient multi-beam handling but faces limitations due to discrete steering angles and reliance on tunable laser sources.

Meta-surfaces use sub-wavelength resonators to manipulate a reflected beam wavefront [5]; these nanoscale structures offer compact form factors and rapid steering capabilities. However, their application is restricted by narrow optical bandwidth operation, one-dimensional steering constraints, and beam squint effects, as the resonators are optimized for specific wavelengths and directions.

Optical phased arrays achieve beam steering through coordinated phase manipulation in multiple emitting elements [6]. By adjusting the phase shifts across the coupling grating array, the overall wavefront can be shaped, allowing for beam steering through constructive and destructive interferences. While this approach provides an integrated solution, it presents significant fabrication and control complexities, particularly for two-dimensional steering applications that require wavelength tuning.

Spatial light modulators enable beam steering by modulating the refractive index of pixels in a two-dimensional matrix [7]. These devices offer flexibility in wavefront shaping for multiple beams, through careful design of the screen hologram, but are inherently limited by pixel size, pitch, and refresh rate constraints.

The proposed beam steering approach employs MEMs cantilevers with piezoelectric transducers (PZTs) [8], as illustrated in Figure 2, offering a simpler solution for optical wireless communications. By manipulating the light source within a lens’s focal plane, the system achieves precise beam steering with remarkable flexibility. A key innovation lies in the integration of silicon micro-lenses with piezoelectric benders, which provides superior optical characteristics through silicon’s high refractive index and transparency at C-Band wavelengths.

The method’s unique design ensures that optical properties remain nearly unchanged, as only the fiber tip is moved. Silicon micro-lenses magnify this movement, increasing achievable steering angles to effectively cover a large wireless cell. Experimental implementations have demonstrated impressive steering capabilities, including angles of 51.3° horizontally and 22.4° vertically, with wavelength independence across the infrared spectrum and continuous beam movement [9].

Despite its potential, the approach presents challenges such as asymmetric steering ranges and mechanical coupling between actuators. Each user beam requires individual piezoelectric transducers and control circuits, as seen in Figure 2b. This system offers a promising solution for indoor optical wireless communications, particularly where dynamic user tracking and flexible beam management are critical.

As shown in

Table 1. Comparison between the state of the art on IR beam-steering.

Metric	This Paper	Metasurfaces [5]	OPA [10]	SLM [7]	AWGR [2]
Field of View	51.3° × 22.4°	54.2°	160°	60°	17° × 17°
Throughput	10 Gbps	NA	NA	224 Gbps	112 Gbps
Steering Type	Continuous	Discrete	Discrete	Discrete	Discrete
Multi-Beams	No	No	No	Yes	Yes
Wavelength	1550 nm	950 nm	1550 nm	1550 nm–1600 nm	1550 nm

, while the current implementation doesn’t support multi-beam operation, it demonstrates a robust 10 Gbps OOk throughput. Although the throughput is lower than the Spatial Light Modulator (224 Gbps) and Arrayed Waveguide Grating (112 Gbps) solutions, the system offers unique advantages that compensate for this limitation. Unlike the discrete steering approaches of Metasurfaces, Optical Phased Arrays, and other alternatives, this approach offers continuous movement capability, enabling more precise and fluid beam positioning. And wavelength independence ensuring compatibility with any optical telecommunications infrastructure, while the simple implementation using commercial components like benders and tapered fibers reduces system complexity and potential manufacturing costs.

The continuous movement and wide Field of View (51.3° × 22.4°) make this approach particularly promising for dynamic indoor optical wireless communication environments. While current multi-beam functionality is limited, the fundamental design suggests a potential for future scalability and performance improvements. The use of silicon micro-lenses with high refractive properties and transparency at IR wavelengths further enhances the system’s optical performance, positioning it as an innovative solution for beam steering technologies.

2.3. Transmitter Experimental Setup

Our experimental setup implements an Anritsu pulse pattern generator to create an OOK (On-Off Keying) pseudorandom binary sequence (PRBS) data stream. This signal drives a high-bandwidth intensity modulator, which is powered by a C-band laser operating at 10 dBm optical power. The topology of the experiment is illustrated on Figure 3a, and its implementation on Figure 3b. To achieve sufficient signal strength, the modulator’s output is amplified to 13 dBm using an Amonics erbium-doped fiber amplifier (EDFA). The amplified beam maintains eye safety standards, resulting in free-space power of 9 dBm after accounting for the system losses. This losses are due to the 1 dB launching loss at the fiber tip and a 3 dB reflection loss from the silicon micro-lenses, as this elements do not have anti-reflection (AR) coatings. While our experimental setup could compensate for these reflection losses through increased amplification, systems with strict power constraints would benefit significantly from AR coatings on the fiber and micro-lenses to preserve transmission power.

The beam steering mechanism consists of a pair of commercial bimorph piezoelectric actuators that manipulate the EDFA output fiber along the X and Y axes. For optimal beam formation, the fiber tip is precisely positioned in the focal plane of the fixed micro-lens, with actuator control provided by variable power supplies operating in the range of 0 to +25 V. To achieve precise beam characteristics, we employ a tapered lensed fiber that forms a 2 µm waist spot at the lens’s focal plane, resulting in a well-collimated beam with a divergence half-angle of 3.2°.

The silicon lenses were design to achieve a 30° with a 20 µm displacement from optical axis. This parameters were selected based on the limitations of the given actuators [8]. We used the paraxial matrix equation [11] to determine the appropriate focal length. The Equation (1) allows for estimate calculation of ray propagation through the optical system.

$$\left[\begin{array}{c}{x}_1\\ {\theta }_1\end{array}\right]=\left[\begin{array}{cc}1& 0\\ -{\displaystyle \frac{1}{f}}& 1\end{array}\right]\left[\begin{array}{c}{x}_0\\ {\theta }_0\end{array}\right]$$

(1)

where $x_0$ and $x_1$ represent the initial and final ray heights, ${\theta }_0$ and ${\theta }_1$ the initial and final ray angles, and f is the focal length of the lens. This Equation (1) depends on the paraxial approximation where $sin\left(\theta \right)\approx \theta$ and $cos\left(\theta \right)\approx 1$. Consequently, it holds up to 10°, after having an increasing error. Assuming that the center ray has a initial incidence angle of ${\theta }_0=0$, and ignoring the translation of the output beam as it does not influence the steering angle we get the following simplified Equation (2):

$${\theta }_1=-{\displaystyle \frac{x_0}{f}}$$

(2)

Through our initial calculations, we determined the lens radius to be 94.8 µm. To validate and refine this design, we utilized Ansys Zemax Opticstudio 2023, for ray-tracing. The iterative process of ray tracing allowed us to further optimize the lens size, adjusting the radius to 100 µm and a focal length of 40.2 µm. These silicon lenses are fabricated in our clean room using a combination of photolithography, dry etching, and thermal reflow. Resulting in silicon lenses have a radius of curvature of 102.6 µm, Figure 4. This increased size provides a larger area for beam movement, enabling greater steering angles. Due to silicon’s high refractive index of 3.48 [12], the silicon lenses have a shorter focal length of 41.3 µm.

This dimensional profile was precisely characterized using a Dektak XT SurfaceProfiler, enabling accurate measurement of the lens curvature and surface quality. The profilometer high-resolution vertical scanning capability allowed us to verify the lens specifications and manufacturing consistency across the wafer, shown in Figure 4, compared with the ideal case of 102.6 µm.

The mechanical assembly comprises actuators mounted on a 3-axis MicroBlock for precise initial lens alignment, with the lens secured in a custom 3D printed housing. For signal detection, the narrow beam is coupled to a multimode fiber using a Thorlabs adjustable fiber collimator positioned at 37.5 cm. The collimator system features a 20.5 mm maximum aperture and a 40 mm effective focal length. The beam is collimated to a 50 µm multi-mode fiber (MMF), to ensure the best coupling with this setup. However, the collimator has a very narrow FoV demanding to be precisely aligned to the incoming beam. Meaning that for every measured point the beam was set at the desired angle, and then the collimator was also set at that angle, normal to the beam, and aligned at the peak power.

While this receiving configuration introduces a 7 dB loss due to coupling inefficiencies, need from precise alignment, and beam vignetting, it offers a critical advantage in bandwidth performance. Traditional wide FoV receivers, which typically employ large-area photodetectors, are fundamentally limited by the inherent trade-off between their active area and bandwidth capability. The increased capacitance of larger photodetectors substantially reduces their frequency response. In contrast, our fiber-coupled approach, which utilizes a high-speed MMF photodetector, that has a bandwidth of 25 GHz. This significant bandwidth advantage enables our system to support high-speed data transmission at 10 Gbps, making the additional coupling losses an acceptable trade-off for enhanced communication performance. This compromises were acceptable as the best way to characterize the transmitter side, while assuming an ideal receiver.

System performance monitoring includes bit error rate (BER) measurements using an Anritsu error detector, signal-to-noise ratio (SNR) measurements, and eye diagram analysis using a high-bandwidth Agilent oscilloscope. Precise beam characterization is achieved through a rotation stage-mounted collimator for steering angle measurements, complemented by comprehensive system analysis including eye opening measurements, SNR evaluation, and voltage-steering angle relationship assessment.

2.4. Experimental Results

To characterize the beam steering capabilities of our system, we first evaluated the performance of the piezoelectric actuator throughout its full voltage range. The actuator was initially centered at its median voltage of 12.5 V and precisely aligned. We then performed comprehensive voltage sweep tests from 0 V to 25 V in both forward and reverse directions to assess movement linearity and quantify the hysteresis effect, defined as the positional difference between forward and return trajectories of the piezoelectric benders.

The measurements revealed that along the horizontal X-axis, the actuator demonstrated a highly linear angular displacement of −24.4° to 26.9°, achieving a total field of view (FoV) of 51.3°, as shown in Figure 5. The FoV is defined here as the angular range over which the system maintains a bit error rate (BER) below the forward error correction (FEC) threshold of 10⁻³. By initiating operation at the median 12.5 V, we successfully minimized the hysteresis of the PZT transducer to only 0.5°.

For the vertical Y-axis characterization, we employed an identical testing methodology. Starting from the 12.5 V center position, the voltage was swept through its full range. The Y-axis exhibited linear angular displacement from −9.3° to 13.1°, resulting in an operational FoV of 22.4° with a comparable hysteresis of 0.7°. This asymmetry in the FoV between the axes is attributed to the cascaded mounting arrangement, where the additional mass of the X-axis actuator impacts the Y-axis performance. However, these results represent a significant improvement over previous implementations using photoresist lenses, which achieved a FoV of only 12.8° vertically and 27.9° horizontally.

To fully characterize the steering capabilities of the system, we analyzed the relationship between beam profile and steering angle by adjusting the MicroBlock actuator mount. The signal strength at the receiver, which directly impacts the BER, is determined by both the beam shape preservation and the geometric path loss that increases with steering angle. Using an idealized receiver setup with a perfectly aligned fiber collimator, we measured a maximum FoV of 100° for the silicon lens and lensed fiber combination, depending on the distance from the receiver, determined by the Error-Free area on Figure 6 (BER = ${10}^{-9}$). The BER limit was calculated by the theoretical BER with OOK modulation [13] to 15.56 dB, by the Equation (3), assuming only additive white Gaussian noise (AWGN).

$$BER={\displaystyle \frac{1}{2}}erfc\left(\sqrt{{\displaystyle \frac{SNR}{2}}}\right)$$

(3)

where erfc is the complementary error funtion and SNR is the power signal to noise ration given by $SNR=10log\left({\displaystyle \frac{P_{Signal}}{P_{Noisel}}}\right)$.

By removing the limitation of an open receiver, the system meets the requirements for 10 Gbps transmission. However, we observed a noticeable sharp drop in performance after reaching 30°, the value for which the lenses were designed, making it difficult to sustain the required high performance. For beams wider than 38.5°, the system experiences an increasing loss, as the center of the beam is vignetted by the lens edge, while angles beyond this threshold are achieved by the marginal rays; they present a significantly degraded performance. The observed deviations from theoretical predictions can be attributed to various practical factors, including manufacturing tolerances, alignment precision, and minor imperfections in the experimental setup.

Eye diagram measurements conducted across steering angles from 0° to 50° at 10 Gbps NRZ OOK modulation consistently showed clear openings, as seen in Figure 7, although with progressive reduction in amplitude at larger angles. Even at the critical 50° steering angle, while exhibiting reduced amplitude compared to normal incidence, the eye diagram maintained a sufficient opening for reliable communication.

3. Receiver for Optical Wireless Communication

3.1. Introduction

For the receiver, the challenge lies in optimizing the aperture, field of view, and bandwidth. A large aperture is required to capture as much light as possible given the limited transmission powers, a wide field-of-view is necessary for receiving optical beams from various directions, and a high bandwidth is vital for high-speed data transmission. However, one cannot improve one parameter without tampering with the other. There are two compromises here. The bandwidth of top-illuminated photodiodes, which are commonly used in OWC is inversely proportional to their aperture. This is because with an increase in the area of a photodiode, its capacitance increases, which negatively impacts the bandwidth. The second compromise is between the aperture and the field-of-view, which is caused by the law of etendue. The larger the aperture, the narrower the field-of-view becomes.

3.2. State-of-the-Art

There are several approaches to solving the problem of bandwidth-aperture and aperture-FoV optimization. Photodiode arrays have been proposed in [14,15]. Their advantage is that by arranging the photodiodes in a square matrix, the aperture and the generated photocurrent are increased but the bandwidth remains the same as of a single photodiode. A 8 × 8 photodiode array presented in [14] achieved 3.3 Gbps with a field of view greater than 40° over the distance of 1 m, and an 8 × 8 array presented in [15] showed transmission of 25 Gbps over 10 m. The performance of PD arrays can be improved by placing microlenses on top to expand the field of view and increase tolerance to misalignment [16]. In [17], 12 photodiodes are arranged in an array, however, each is processed in the electrical domain separately, which allows dynamic tracking and localization with which 40 Gbps was achieved. This concept is called an Angle Diversity Receiver (ADR), and in our previous work [18], we presented a quad photodiode array with which we achieved 20 Gbps and a field of view greater than 32°. The field of view of the 2 × 2 ADR PD array can also be improved to 60°with mechanical structures such as compound parabolic concentrators [19]. A nature-inspired angle diversity receiver, shown in [20], is based on an animal eye. The receiver consists of a photodiode array with microlenses placed on a hemisphere for a wide field of view. It is not a receiver for OWC but a camera; nonetheless, it is mentioned for its potential adaptation. Another type of OWC receiver that has been proposed is a large-area avalanche photodiode shown in [21]. In that work, 1 Gbps over 3.3 m with a field-of-view of 68° was achieved. The sensitivity of OWC receivers can be reduced to as low as one photon per bit as shown in [22], while still maintaining a high data rate of 10.5 Gbps. However, this solution must be cooled down to a few degrees Kelvin, deeming it impractical for widespread indoor OWC systems. Photoluminescent detectors have been presented in [23,24]. They can be omnidirectional receivers with very large aperture areas, which is promising for the indoor OWC, however the bandwidth of these devices is relatively low. Metasurfaces can be considered a likely candidate for future research, as shown in [25,26,27]. These materials absorb light coming from a wide range of angles and reemit collimated light at longer wavelengths, effectively reducing the etendue of the beam. However, etendue-reducing metasurfaces have low efficiency, at the time of writing, making them impractical for indoor OWC. An aperture of OWC receivers can be made larger with mechanical elements such as non-imaging optics [28,29]. A compound parabolic concentrator is a device which can make the aperture substantially larger at the cost of a lower field-of-view.

However, the proposed existing receivers for optical wireless communication listed above hinder the scale up of transmission networks. The solution to overcome compromises between the aperture, bandwidth, and field of view can be decoupling light collection from light detection. In this way, both can be independently optimized. Additionally, by decoupling light detection, the receiver can facilitate additional functionality such as WDM demultiplexing, or coherent detection. A device which allows that is a photonic integrated circuit. In [30], a surface grating coupler was connected to a waveguide photodiode with a bandwidth of more than 110 GHz, which could allow very high data rate transmission. The surface grating couplers can also be arranged in an array as shown in [31]. There, the array in combination with Mach-Zehnder Modulators is used to decouple SDM beams. Surface grating couplers can also be used for WDM signals. Refs. [32,33] show grating couplers with an integrated wavelength demultiplexing function. However, these couplers can decouple only two and three wavelengths, respectively, which limits the scalability. In our previous work, we showed a 100 Gbps WDM receiver with a single surface grating coupler [34], and a wide field-of-view “windmill” array of six surface grating coupler array [35]. In this paper, we combine our previous work. In [34], the focus lies on achieving high-capacity wireless data transmission while collecting light from free space with a surface grating coupler, and in [35], the focus is on increasing the field-of-view of the receiver so that light can be collected from free space from wider range of angles.

Indium-Phosphide Membrane on Silicon

The platform which we used for the fabrication of the Photonic Integrated Circuits (PIC) is Indium-Phosphide Membrane on Silicon (IMOS) [36]. IMOS wafers consist of InP and Silicon wafers bonded together with BCB which creates benefits over other commonly used platforms for the fabrication of PICs. Because of the high refractive index ratio between the InP and the Silicon, IMOS allows small feature sizes of the devices on-chip and small bend radii of waveguides, hence one can achieve the same level of functionality on a much smaller area compared to a conventional InP platform. Furthermore, unlike silicon photonics, IMOS allows for light generation and amplification on-chip, and enables the fabrication of monolithically integrated receivers for indoor OWC.

3.3. Quad Photodiode Array

3.3.1. Description

The 2 × 2 photodiode array presented in [18] was designed and manufactured by VTEC Lasers & Sensors. The array has a diameter of 100 µm. The individual photodiodes in the array have the shape of a 90° circular sector so that the whole array takes the shape of a circle. Each photodiode is connected to a separate Trans-Impedance Amplifier (TIA). This concept allows for an increase in the active area while maintaining a high bandwidth. Additionally, having four independently processed photodiodes enables the detection of the incoming beam direction. This photodiode array could be used simultaneously for receiving data and localizing and tracking of users.

3.3.2. Experiment

The diagram of the experimental setup is shown in Figure 8. In the experiment, the tunable laser source emitted light at 1530 nm with 12 dBm output power. Then, the light polarization was aligned to the TE mode for optimization of the transmission power of the modulator. The BERT modulated the signal with OOK NRZ PRBS 31. The light was then amplified with EDFA to compensate for the losses in the system, and later VOA was used for optical power sweep. The output fibre was placed very closely above the photodiode array as shown in Figure 8. The fibre was mounted to a six-degree-of-freedom platform such that the position and the angle of the fibre were controlled. The output of the photodiode was fed back to the BERT via a DC block to remove the bias voltage.

In the experiment, two parameters of the photodiode array were verified: the field-of-view and the effect of the received power on the BER. The field of view was measured by changing the angle of the fibre with respect to the photodiode array. The maximum angle supported by the mounting is 16° relative to the normal incidence. This measurement was done by transmitting 1 Gbps at different angles and measuring BER. In the second measurement, the VOA was used to sweep the power for various bit rates, which was controlled by the BERT.

3.3.3. Results

Figure 9a shows BER curves in relation to the received power. The BER curves at 1 Gbps overlap for both incidence angles of 0° and 16°. Therefore, the photodiode array has the same response in the tested range of incidence angles. Figure 9b shows the BER relationship with the optical received power for data rates between 10 Gbps and 23 Gbps. The eye diagrams for 16 Gbps and 20 Gbps are shown in Figure 10. Error-free transmission was achieved at 16 Gbps. The transmission of 20 Gbps could still be done with hard decision FEC, and 23 Gbps with soft decision FEC.

3.4. WDM OWC Receiver with a Single Surface Grating Coupler

3.4.1. Introduction

The photodiodes used for OWC are limited in their ability to receive only the signal amplitude. In order to overcome this, light collection needs to be decoupled from light detection, as previously mentioned. PICs can facilitate this concept. OWC PIC-based receivers can use surface grating couplers as light collection and waveguide photodiodes for light detection. Additionally, the functionality of PIC-based receivers can be expanded with e.g., WDM demultiplexing, coherent detection, or separation of SDM beams. In this section, we present the combination of the IMOS surface grating coupler with AWGR to receive a WDM signal, which was previously shown in [34]. Wavelength Division Multiplexing can significantly increase the throughput of optical links. The AWGR allows the splitting of wavelengths and processing them separately. OWC WDM can be used in applications which require very high-capacity transmission, e.g., in outdoor OWC between buildings or in top-of-rack switches in data centers.

In this part of the work, a PIC with surface grating couplers is used together with commercially available AWGR to create a WDM receiver for OWC. The grating couplers used here were designed for fibre-to-chip coupling. Their dimensions are 15 µm × 15 µm. This is a proof-of-concept demonstration. Further optimization and integration are necessary.

3.4.2. Experiment

In Figure 11 the diagram of the experimental setup used in the experiment is shown. Four tunable lasers emitted light in the C-band with wavelength spacing of 200 GHz (approx. 1.6 nm) on ITU grid. All wavelengths were modulated with 25 Gbps NRZ PRBS15, amplified with EDFA, and combined with three stages of 3-dB combiners. Light is coupled into the free space with a collimator. Then, a one-inch diameter lens and 50 mm focal length focused light on the surface grating coupler. The distance between the collimator and the lens is 5 cm, and between the lens and the PIC is another 5 cm. Therefore, the total length of the wireless path is 10 cm. The collimator and lens were tilted 9° with respect to the surface of the PIC. The light was then coupled out of the PIC and amplified with the EDFA to compensate for the losses. Then, four wavelengths were demultiplexed with a commercial AWGR with 200 GHz channel spacing. Data transmission performance was analyzed using an Optical Spectrum Analyzer (OSA), a Digital Communication Analyzer (DCA), and a Bit Error Rate Tester (BERT).

3.4.3. Results

Firstly, the losses of the surface grating coupler were characterized by probing on both sides of the structure with a cleaved SMF. The relation of the SGC loss with the angle of incidence and the optical frequency in C-band is shown in Figure 12a. The measured sample has an increasing loss for increasing frequency in the C-band. The approximate FoV of a surface grating coupler is around 10°, however here we show only four discrete angles due to the limitation in the setup. For the angles between 8° and 11° the loss can vary up to 3 dB. The frequencies used in the WDM transmission experienced an 8 dB loss. The addition of the 10 cm wireless path, collimator and lens increased the optical path loss by 3 dB. Figure 12b shows the measured spectra of the modulated WDM signal after splitting by AWGR and amplification by EDFA. The adjecent channel isolation of the AWGR is >30 dB and non-adjacent channel isolation >40 dB. The power of each wavelength was equal at the beginning of the transmission link. However, channels experienced slightly different losses because of the AWGR and the wavelength-dependent loss of the SGC. The peak powers are within 2 dB margin and OSNR levels are 17 dB for all channels. Figure 12c shows BER curves of each channel and back-to-back for comparison. An error-free 25 Gbps NRZ OOK per channel is achievable for power levels above -9.5 dBm, which is 1 dB higher than back-to-back. The eye diagrams of the error-free signals are shown as inset in Figure 12c.

3.5. OWC Receiver with an Array of Surface Grating Couplers

3.5.1. Description

In previous sections, the proposal to decouple light collection and light detection was discussed and we have also demonstrated that SGC can be used to collect light from free space efficiently. This section proposes an improvement for light collection in PICs, as already published in [35]. Here, we present an array of six surface grating couplers arranged in a “windmill” shape for an increased field of view compared to an individual surface grating coupler, as shown in Figure 13. The PIC is fabricated on the IMOS platform, which was introduced in the previous section. The grating couplers are standard focusing grating couplers used for fibre-to-chip coupling. The active area has a length of 16 µm and a width that decreases from 18 µm at its end to 10 µm at the input to the taper. The output waveguides of the SGCs are combined with Multi-Mode Interferometers (MMI). The simulation results are shown in Figure 14. For certain combinations of wavelength and incidence angle, the coupling efficiency is the highest.

3.5.2. Experiment

Figure 15 shows the experimental setup in which wireless data transmission has been demonstrated with the array of SGC used as a light collection device. Firstly, a tunable laser emitted light at 1550 nm, and then the light was modulated with NRZ OOK PRBS15 at 1 or 10 Gbps and amplified with an EDFA to compensate for the losses in the system. The light is coupled in an into free space with a collimator. A lens with 1-inch diameter and 75-mm focal length is used to focus the light on the surface grating coupler. The total wireless path is 15 cm long. The collimator and lens were tilted 10° with respect to the surface of the PIC. The light was coupled out of the PIC with a cleaved fibre and passed to an optical receiver consisting of a photodiode and a trans-impedance amplifier. Transmission performance was analyzed with a BERT. A test structure consisting of two focusing surface grating couplers connected by a waveguide was used to characterize individual grating couplers. In this case, two cleaved fibres were used to couple light in and out of the PIC, and the optical power loss was measured.

3.5.3. Results

Figure 16a shows the coupling loss of the individual grating couplers for different angles at wavelengths in the C-band. The angle with the highest coupling efficiency is 10° for which the loss is about 7.5 dB at 1550 nm and all values are within 1 dB over the C-band. The coupling loss for angles below 10° is decreasing for increasing wavelengths. The opposite holds for angles greater than 10°. In the middle of the C-band at 1550 nm, the 2-dB full angle field-of-view is 8°, currently limited by the maximum adjustable angle in our setup. It can be seen that the lines in Figure 16a are not smooth. That is caused by the fact that the grating couplers were connected by a 1 mm long waveguide. Cleaved fibres together with a finite optical path length of the waveguide create a cavity, which results in power variation for neighbouring wavelengths.

Focusing light with a lens on the grating coupler array and coupling light with a cleaved fibre resulted in a loss of 32 dB. Therefore, the loss of the array is 25 dB. Figure 16b shows the BER curves for 1 Gbps and 10 Gbps. The received power of −14 dBm is sufficient to achieve error-free 1 Gbps data transmission. Data transmission below the 10⁻³ Forward Error Correction (FEC) limit is achievable at −16.5 dBm and −14 dBm at 1 Gbps and 10 Gbps, respectively.

4. Conclusions

We experimentally demonstrated several transmitter and receiver technologies for potential applications in indoor OWC systems for 6G networks. Due to the very high available bandwidth, the data rates in OWC exceed the limits of radio telecommunication. Furthermore, the properties of light enable the transmission to be more energy efficient and secure.

For the indoor OWC transmitter, we propose a piezoelectric cantilever pair with a silicon micro-lens. This system allows for quick and direct optical beam steering towards mobile wireless users. This approach is attractive because of its minimal loss, compact size, and relatively simple implementation using commercially available resources. The use of silicon microlens in conjunction with tapered and lensed fiber, moved by the pair of piezoelectric actuators, resulted in large steering angles of 51.3° horizontally and 22.4° vertically, and a beam with 3.2° half-angle divergence. The observed asymmetry, between the axes, is attributed to the additional weight and fixtures associated with the horizontal actuator on top of the vertical one. Despite showing hysteresis in the actuators, the transmission angle maintains a predictable relationship with the applied voltage. By evaluating the lens and fiber combination, an error-free error transmission of 10 Gbps can be achieved up to ±50° at 37.5 cm distance.

The results of this prototype establish a foundation for advancing toward indoor optical wireless networks and expand it to outdoor terrestrial use. Future developments will focus on implementing comprehensive user tracking capabilities designed for typical indoor movement walking speeds. Additionally, we plan to expand the system’s capacity through parallel actuator arrays, enabling simultaneous high-speed connections to multiple users. While they can be used to make a full-duplex system, the actuator is too large to integrate into common use electronics; for this a small two-dimensional VCSEL array using focal plane steering can be used, as they are compact and integrable on its implementation.

We have also shown the variety of potential receivers which could be used in optical wireless communication. We demonstrated that a quad photodiode array could potentially be used for indoor OWC. It has a full-angle FoV greater than 32° and allows reliable error-free transmission of 16 Gbps. Because of the separation of the individual photodiodes in the array, this device could allow the localization and tracking of users or automated alignment. The transmission capacity can be increased by using more spectrally efficient modulation formats, e.g., PAM4 or DMT. We have also shown WDM transmission of 100 Gbps (four channels of 25 Gbps) received with a single surface grating coupler, and an array of surface grating couplers for increased field of view for OWC receivers. Separation of light collection and light detection, enabled by photonic integrated circuits, can enable scalability, which future receivers in the OWC will need. PICs allow light to be collected with surface grating couplers and light to be detected with high-speed photodiodes. Additionally, this separation allows processing of light in the optical domain, e.g., wavelength demultiplexing, coherent detection, or spatial beam demultiplexing. Although grating couplers may be sensitive to only one polarization state, there are many ways to increase the capacity of data transmission. Large-area and wide field-of-view grating couplers can be combined with existing fibre-pigtailed detectors as receivers for OWC, making use of the existing infrastructure.

Despite promising high bandwidth, low latency and privacy, an OWC system needs to consider the use of hybrid optical/RF approach for critical systems. Due to it necessity of direct line of sight to the receiver. In critical applications, a complementary radio frequency communication path ensures continuous connectivity and reliability when optical transmission becomes obstructed. This approach allows for seamless fail-over between optical and RF communication modes, adapting to environmental challenges. This redundancy mechanism not only enhances the overall system resilience but also provides a comprehensive communication strategy that strengths the unique qualities of optical wireless while having the reliability of radio as a back-up.