Survey on Optical Wireless Communication with Intelligent Reflecting Surfaces

Abstract

Optical Wireless Communication (OWC) technology has gained significant attention in recent years due to its potential for providing high-data-rate wireless connections through the large license-free bandwidth available. A key challenge in OWC systems, similar to high-frequency Radiofrequency (RF) systems, is the presence of dead zones caused by obstacles like buildings, trees, and moving individuals, which can degrade signal quality or disrupt data transmission. Traditionally, relays have been used to mitigate these issues. Intelligent Reflecting Surfaces (IRSs) have recently emerged as a promising solution, enhancing system performance and flexibility by providing reconfigurable communication channels. This paper presents an overview of the application of IRSs in OWC systems. Specifically, we categorize IRSs into two main types: mirror array-based IRSs and metasurface-based IRSs. Furthermore, we delve into modeling approaches of mirror array-based IRSs in OWC and analyze recent advances in IRS control, which are classified into system power or gain optimization-oriented, system link reliability optimization-oriented, system data rate optimization-oriented, system security optimization-oriented, and system energy optimization-oriented approaches. Moreover, we present the principles of metasurface-based IRSs from a physical mechanism perspective, highlighting their application in OWC systems through the distinct roles of light signal refraction and reflection. Finally, we discuss the key challenges and potential future directions for integrating IRS with OWC systems, providing insights for further research in this promising field.

1. Introduction

1.1. B5G with IRSs

With the rapid development of science and technology, the demand for high-speed wireless communications has surged, particularly driven by bandwidth-intensive applications such as high-definition video-on-demand, virtual reality, and augmented reality [1]. However, the congested microwave spectrum has caused conventional Radiofrequency (RF) wireless communication systems to struggle to meet these bandwidth and speed demands [2,3]. To overcome these limitations, Beyond Fifth-Generation (B5G) has attracted significant attention for its potential to provide high-speed wireless connections through large, license-free bandwidths. B5G is envisioned to rely on a variety of wireless technologies, including millimeter-wave (mmWave), terahertz (THz) communications, and Optical Wireless Communication (OWC) [4], as summarized in

Table 1. Comparison of B5G wireless communication technologies.

	mmWave Systems	THz Systems	OWC System	FSO System
Distance	Long (several kilometers in ideal conditions, tens to hundreds of meters in urban environments)	Short (a few meters to several hundreds of meters in ideal conditions)	Limited (tens to hundreds of meters indoors)	Several kilometers (can be much longer in space-based systems (lunar-to-Earth or deep space))
Carrier frequency	Low (30 GHz to 300 GHz)	Moderate (0.1 THz to 10 THz)	High (hundreds of THz)	High (hundreds of THz)
Bandwidth	Several GHz	Tens to hundreds of GHz	Broad (in the order of THz)	Broad (in the order of THz)
Data rate	Tens of Gbps	Tens to hundreds of Gbps	Tbps	Tbps

.

Millimeter-wave communication has been explored for short-to-moderate-distance data transmission, typically ranging from tens to hundreds of meters. However, mmWave suffers from a high path loss and sensitivity to blockages, which limit its effective transmission distance and require advanced techniques like beamforming and Multiple-Input–Multiple-Output (MIMO) to maintain reliable connections [5]. Similarly, THz communication offers even larger bandwidths and higher data rates, potentially reaching tens to hundreds of gigabits per Second (Gbps) [6]. However, THz waves face even higher attenuation and are highly sensitive to atmospheric conditions, limiting their transmission distance from a few meters to several hundreds of meters under ideal conditions [7].

In comparison, OWC has been proposed and widely studied for its potential to achieve much higher data rates, reaching up to terabits per second (Tbps), thanks to the vast license-free bandwidth available in the optical spectrum. Additionally, OWC is more cost-effective compared to THz communications, utilizing low-cost transceivers such as Light-Emitting Diodes (LEDs) and Photodiodes (PDs) [8,9,10]. Free-Space Optics (FSO), a type of OWC, extends these benefits over longer distances, making it suitable for applications like space-based communications. Despite these advantages, OWC, like mmWave and THz communications, faces significant challenges related to the vulnerability of signals to obstructions, which can cause considerable signal loss and degradation, or even complete communication interruption [11,12,13,14]. The short wavelength of optical signals leads to higher penetration loss and increased vulnerability to shadowing and blockages, which can create coverage gaps and negatively impact the system’s performance and Quality of Service (QOS) [4].

To mitigate these challenges, innovative solutions are required to ensure robust and reliable wireless systems. Techniques such as adaptive optics [15], diversity schemes [16,17], and the use of Intelligent Reflecting Surfaces (IRSs) have been proposed to enhance signal propagation and overcome the limitations posed by obstacles [18]. Among these, IRSs are highly promising due to the unique capability to dynamically manipulate the wireless propagation channels. IRSs are planar arrays of resonant sub-wavelength elements designed to manipulate the phase of incident beams across their surface to achieve desired directionality, such as anomalous reflection, following the generalized law of reflection. However, for optical applications, conventional mirrors, which primarily reflect light without phase manipulation, do not fall into this category [19,20]. In IRS-based RF systems, IRSs belong to the broader category of metasurfaces, altering various properties of electromagnetic waves such as amplitude, phase, dispersion, momentum, and polarization [18]. In OWC systems, IRSs are typically categorized into two types: mirror array-based IRSs and metasurface-based IRSs. A mirror array-based IRS controls the optical signal path by adjusting the orientation of mirrors within the IRS matrix to intelligently vary the signal reflection angle and guide the signal towards the receiver, enhancing system performance [21,22]. A metasurface-based IRS, on the other hand, uses nanotechnology and materials science to design reflective surfaces that perform beamforming, reducing interference and beam wastage, and increasing the system gain [23,24].

Compared to their application in RF systems, IRSs in OWC systems exhibit distinct differences and advantages. In RF communication systems, IRSs primarily enhance signal propagation and extend coverage by optimizing reflection phases. However, in OWC systems, IRSs must optimize both communication performance and illumination when the optical communication systems serve dual roles of lighting and communication, such as in Visible-Light Communication (VLC). This optimization is not necessarily required for near-infrared systems, which are typically used solely for communication purposes. The unique characteristics of optical signals require IRS designs to consider beam shaping and the receiver’s Field of View (FOV) to ensure adequate illumination and communication link quality [4]. Additionally, while IRS deployment in RF systems typically occurs between the transmitter and the receiver, IRSs in OWC systems can be placed at the transmitter, receiver, or along the transmission path, depending on functional and performance optimization needs [22,25,26,27,28]. This flexibility allows IRSs in optical communication systems to more precisely control signal propagation, effectively overcoming the challenges posed by complex environments, such as urban areas with dense buildings, indoor settings with numerous obstacles, and outdoor environments with varying weather conditions.

1.2. Recent Surveys on IRSs

In recent years, a number of surveys and summary papers on IRSs have been published, which are summarized in

Table 2. Recent surveys and the comparison with this paper.

Reference	Year	Area of Focus
RF Systems
Gong et al. [29]	2020	• IRS applications and design in future wireless networks • Concepts and reconfigurability • Joint optimization • Performance improvements
Hassouna et al. [30,31]	2023	• IRSs in enhancing wireless networks • Hardware, control, and channel models • Performance metrics • Challenges
Zheng et al. [32]	2022	• IRS-aided wireless communications • Design issues • New architectures • Applications
mmWave/THz Systems
Chen et al. [33]	2021	• IRSs in THz communications for 6G • Application scenarios • Enabling technologies • Challenges
Raza et al. [34]	2022	• IRSs in THz communication for B5G/6G • Performance analysis • Enhancing reliability and data rates
Zhu et al. [35]	2022	• IRSs in 6G vehicular communications • Signal strength and security • Positioning accuracy
Mohsan et al. [36]	2022	• IRS and UAV integration • Literature and technologies • Application scenarios
OWC Systems
Abumarshoud et al. [25]	2021	• IRS-assisted LiFi systems • IRS architecture • Operational elements • Major challenges
Sylvester et al. [4]	2023	• IRSs in OWC systems • Indoor applications • Key challenges
This survey	2024	• IRSs in OWC systems • Mirror array models • Recent developments • Major challenges and future scope

. In [29], a brief review of the recent applications and design aspects of IRSs in future wireless networks is presented. The paper introduces the basic concepts and reconfigurability of IRSs, discusses the joint optimization of IRS phase control with transceiver transmission control, and analyzes performance improvements in various network design problems. In [30,31], the principles, performance analysis, and challenges of using IRS to enhance wireless network coverage and data rates are reviewed. These papers discuss IRS hardware, control mechanisms, and channel models, presenting performance analyses using various parameters and metrics. They also address the challenges of integrating IRSs into wireless networks, focusing on channel estimation and deployment comparisons for single- and multi-user scenarios, and propose future research directions for IRS-assisted wireless communication systems. In [32], the up-to-date research on IRS-aided wireless communications is further comprehensively reviewed, focusing on solutions to practical design issues such as channel estimation and passive beamforming, and discussing new and emerging IRS architectures and applications. All of the above papers focus on IRSs in RF systems, highlighting the significant research interest and advancements in this area.

While IRSs have predominantly been applied in traditional RF systems, recent survey papers have also reviewed THz wireless communication systems incorporating IRSs. Ref. [33] reviews the integration of IRSs in THz communications for 6G networks, discussing application scenarios such as mobile communications, secure communications, Unmanned Aerial Vehicles (UAV), Mobile Edge Computing (MEC), and THz localization. It also examines enabling technologies like hardware design, channel estimation, and beam control, while highlighting challenges and open problems in IRS-empowered THz communications. Ref. [34] focuses on performance analysis and future promising applications of IRS-enabled THz communications, highlighting the role of IRSs in enhancing link reliability and data rates to support emerging intelligent applications.

Furthermore, IRSs have also been reviewed from an application-oriented perspective, such as 6G vehicular communications and UAV communications. Ref. [35] surveys current research on the application of IRSs in 6G vehicular communications, both ground-based and aerial, highlighting IRSs’ role in enhancing signal strength, security, and positioning accuracy. Moreover, Ref. [36] surveys the integration of IRSs and UAVs in communications, discussing existing literature, emerging technologies, application scenarios, and future research directions to enhance spectrum and energy efficiencies.

1.3. Contributions of This Survey

While IRS technology has shown potential in RF and THz systems, its application in OWC is promising yet underexplored. OWC offers vast unlicensed bandwidth and high data rates but faces challenges like physical obstructions and sensitivity to alignment. Integrating IRSs can enhance OWC by improving signal propagation and link reliability. However, only a limited number of papers have reviewed the application of IRS in OWC systems. In [4], a comprehensive tutorial on using IRSs in OWC systems is provided, focusing on indoor applications. It covers the basics of OWC and IRSs, the differences between optical IRSs and RF IRSs, key challenges in IRS-assisted OWC systems, and future research directions integrating IRSs with emerging technologies. In [25], a comprehensive overview of IRS-assisted LiFi systems is presented, exploring the underlying IRS architecture from a physics perspective and outlining potential operational elements supported by IRS-enabled transceivers and environments. It also highlights major challenges and promising future research directions. Although the previous survey papers included the IRS aspect, they failed to review recent achievements in IRS techniques in OWC systems. For instance, Ref. [4] is a tutorial focusing on fundamental principles rather than recent progress. On the other hand, Ref. [25] focuses on a particular LiFi system with IRSs, providing insights specific to their system rather than an overview of the entire research field. Therefore, this survey provides a comprehensive review of the recent developments in IRS applications in OWC systems, which include the following:

• A brief introduction and summary of the mirror array-based IRS model in OWC systems.
• A detailed review of the recent development of mirror array-based and metasurface-based IRS in OWC systems.

The rest of this survey is organized as follows: in Section 2, we introduce the general architecture and principles of mirror array-based IRSs in OWC systems and summarize recent applications in UOWC systems. In Section 3, we provide a detailed review of metasurface-based IRSs applied in recent UOWC progress. Finally, we conclude the survey and discuss the key challenges and potential future directions for integrating IRSs with OWC systems in Section 4, offering insights for further research in this promising field.

2. Mirror Array-Based IRSs in OWC Systems

An IRS represents a cutting-edge technology in OWC systems. IRSs can be broadly categorized into two main types: mirror array-based IRSs and metasurface-based IRSs. Both types aim to reconfigure incident signals and manipulate them in intelligent ways to improve communication performance. This section focuses on the mirror array-based IRS, and the metasurface-based type is reviewed in the next section.

A mirror array-based IRS utilizes adjustable mirror arrays to change the propagation of optical signals in OWC systems. Each mirror element in the array can be independently adjusted in real time, allowing precise control over the spatial phase and direction of the light path. This technology enables the control of reflected signal directions so that all waves can converge at a single point (anomalous reflection) rather than reflecting in a uniform manner as with a flat reflector [4]. In the case of a curved reflector, the reflections can be directed towards different focal points depending on the curvature and design.

The control mechanism for a mirror array-based IRS involves the use of ultra-fast switching components such as varactors, Positive–Intrinsic–Negative (PIN) diodes, or Micro-Electromechanical Systems (MEMS) switches, which communicate with a central controller [4]. Unlike traditional fixed metasurfaces, many modern tunable metasurfaces and IRS controllers can autonomously detect environmental changes and utilize intelligent algorithms to adapt in real time [37,38,39]. In RF communication systems, IRS controllers can dynamically adjust the phase and amplitude of reflected signals to optimize performance. In OWC, adaptive techniques are employed to manage signal direction and phase, though the technology is still evolving to match the versatility found in RF systems.

This capability supports the dense deployment of IRSs in wireless communication networks, enabling efficient manipulation of transmitted and reflected waves to optimize the communication channel and improve signal propagation in both RF and OWC systems. The real-time adaptability of IRS controllers helps address challenges such as obstacles, user movement, and varying environmental conditions, enhancing overall network performance.

Understanding the fundamental principles and benefits of mirror array-based IRSs in OWC systems lays the groundwork for effective implementation. To fully leverage these advantages, it is crucial to develop accurate and comprehensive models of OWC systems with mirror array-based IRSs. These models are essential for predicting system performance, optimizing signal propagation, and addressing potential challenges in real-world deployments. In the following section, we review the methodologies and techniques for modeling OWC systems with mirror array-based IRSs.

2.1. OWC System Model with Mirror Array-Based IRSs

Recent research has extensively explored the application of IRSs in indoor OWC [21,22,40,41] and FSO communication systems [18,42,43,44]. In indoor OWC systems, transmitters are typically LEDs with wide beam divergence, which allows them to cover a large communication area. Conversely, in FSO communication systems, transmitters are usually Laser Diodes (LDs), which provide higher power and more focused beams to counteract the losses associated with long-distance communication.

Given the growing demand for high-speed and reliable wireless communication in indoor environments, modeling mirror array-based IRSs in indoor OWC systems becomes particularly important. These environments often face unique challenges such as multipath reflections and obstructions caused by furniture and other users. In this section, we focus on modeling mirror array-based IRSs specifically for indoor OWC systems. As shown in Figure 1, the general indoor OWC system considered includes a mirror array-based IRS based on an LED transmitter and a user’s PD receiver. The channel gain between the transmitter and receiver can be expressed by [4]:

$$G=I{G}_{\mathrm{LOS}}+G_{\mathrm{NLOS}},$$

(1)

and in this review, it is assumed that the variable $I\in \{0,1\}$ represents whether the Line-of-Sight (LOS) link is available or blocked. $G_{\mathrm{LOS}}$ and $G_{\mathrm{NLOS}}$ denote the channel gain of the LOS path and the Non-Line-of-Sight (NLOS) path, respectively.

The channel gain of the LOS path $G_{\mathrm{LOS}}$ can be expressed by:

$$G_{\mathrm{LOS}}=\left\{\begin{array}{cc}{\displaystyle \frac{(m+1)A_{\mathrm{PD}}}{2\pi d^2}}{cos}^m\left(\mathsf{\Phi }\right)T\left(\xi \right)G\left(\xi \right)cos\left(\xi \right),\hfill & 0\le \xi \le {\xi }_{\mathrm{FoV}}\hfill \\ 0,\hfill & \xi >{\xi }_{\mathrm{FoV}},\hfill \end{array}\right.$$

(2)

where $m=-{\displaystyle \frac{ln2}{ln\left(cos{\mathsf{\Phi }}_{1/2}\right)}}$ is the Lambertian order with ${\mathsf{\Phi }}_{1/2}$ representing the semi-angle at half the emitted optical power, $A_{\mathrm{PD}}$ is the capture area of the PD, d denotes the Euclidean distance between the LED transmitter and the receiver, $\mathsf{\Phi }$ denotes the optical irradiance angle of the LED, $\xi$ is the incident angle of the signal light to the receiver, $T\left(\xi \right)$ represents the gain of the receiver optical filter, and $G\left(\xi \right)$ denotes the gain of a concentrator, which is given by $G\left(\xi \right)={\displaystyle \frac{f^2}{{sin}^2{\xi }_{\mathrm{FOV}}}}$, $0\le \xi \le {\xi }_{\mathrm{FOV}}$, where ${\xi }_{\mathrm{FOV}}$ is the FOV of the PD and f denotes the refraction index of the optical concentrator. It is worth mentioning that $\xi$ depends on the receiver’s orientation.

The NLOS channel can be considered in two parts: (i) from the LED transmitter to the IRS, with the IRS serving as the receiver; (ii) from the IRS to the PD, with the IRS considered as the source that re-emits the optical signal [4]. Due to the accuracy requirements and the diminishing strength of higher-order reflections, only the first-order reflections are considered in the analysis of IRSs in OWC systems. The mirror-array IRS is divided into $\mathcal{K}$ squared surfaces with the kth surface having an area $d{A}_k$ [24]. Similar to previous work [4,24,45], it is assumed that the incident ray from the transmitter strikes the center of the reflective surfaces.

The channel gain of the NLOS path in Stage (i) with the kth IRS serving as the receiver can be expressed as:

$$G_{\mathrm{NLOS}}^{\mathrm{Stage}\mathrm{i}}={\rho }_{\mathrm{IRS}}{\displaystyle \frac{(m+1)d{A}_k}{2{\pi }^2{\left(d_k^a\right)}^2}}{cos}^m\left({\mathsf{\Phi }}_k^a\right)cos\left({\xi }_k^a\right),$$

(3)

where ${\rho }_{\mathrm{IRS}}$ is the reflection coefficient of the IRS element, $d_k^a$ represent the distance between the transmitter with the kth reflective surface, ${\mathsf{\Phi }}_k^a$ is the irradiance angle from transmitter to the kth reflective surface, and ${\xi }_k^a$ is the signal incidence angle to the kth reflective surface.

The channel gain of the NLOS path in Stage (ii) with the kth IRS serving as the source can be expressed as:

$$G_{\mathrm{NLOS}}^{\mathrm{Stage}\mathrm{ii}}=\left\{\begin{array}{cc}{\displaystyle \frac{A_{\mathrm{PD}}}{{\left(d_k^u\right)}^2}}cos\left({\mathsf{\Phi }}_u^k\right)cos\left({\xi }_u^k\right)T\left(\xi \right)G\left(\xi \right),\hfill & 0\le {\xi }_u^k\le {\xi }_{\mathrm{FOV}}\hfill \\ 0,\hfill & {\xi }_u^k>{\xi }_{\mathrm{FOV}},\hfill \end{array}\right.$$

(4)

where $d_k^u$ represents the distance between the kth reflective surface and the receiver, ${\mathsf{\Phi }}_u^k$ is the irradiance angle from the kth reflective surface to the receiver, and ${\xi }_u^k$ is the signal incidence angle to the receiver from the kth reflective surface. It needs to be mentioned that the cosine of the angle of irradiance (which is specified by the roll angle $\omega$ and yaw angle $\gamma$ of the mirror array) is [4,24]:

$$cos\left({\mathsf{\Phi }}_u^k\right)={\displaystyle \frac{(x_k-x_u)}{d_k^u}}cos\left(\omega \right)+{\displaystyle \frac{(y_k-y_u)}{d_k^u}}sin\left(\gamma \right)cos\left(\omega \right)+{\displaystyle \frac{(z_k-z_u)}{d_k^u}}sin\left(\omega \right),$$

(5)

where $(x_k,y_k,z_k)$ represent the coordinates of the kth IRS, and $(x_u,y_u,z_u)$ represent the coordinates of the receivers.

Combining Equations (3) and (4), the channel gain of the NLOS path with a mirror array-based IRS system from the kth surface can be expressed as [4,24]:

$$G_{\mathrm{NLOS}}^{{\mathrm{IRS}}_k}(\gamma ,\omega )=G_{\mathrm{NLOS}}^{\mathrm{Stage}\mathrm{i}}\times G_{\mathrm{NLOS}}^{\mathrm{Stage}\mathrm{ii}}=\left\{\begin{array}{c}{\rho }_{\mathrm{IRS}}{\displaystyle \frac{(m+1)A_{\mathrm{PD}}}{2{\pi }^2{\left(d_k^a\right)}^2{\left(d_k^u\right)}^2}}d{A}_k{cos}^m\left({\mathsf{\Phi }}_k^a\right)cos\left({\xi }_k^a\right)\hfill \\ \times cos\left({\mathsf{\Phi }}_u^k\right)cos\left({\xi }_u^k\right)T\left(\xi \right)G\left(\xi \right),\hfill & 0\le {\xi }_u^k\le {\xi }_{\mathrm{FOV}}\hfill \\ 0,\hfill & {\xi }_u^k>{\xi }_{\mathrm{FOV}},\hfill \end{array}\right.$$

(6)

The total channel gain of the NLOS path combining all $\mathcal{K}$ squared surfaces of IRS is given by:

$$G_{\mathrm{NLOS}}=\sum _{k=1}^{\mathcal{K}}{G}_{\mathrm{NLOS}}^{{\mathrm{IRS}}_k}(\gamma ,\omega )$$

(7)

To better understand the characteristics of our model, it is important to compare it with other common modeling approaches. Our model is a ray tracing-based model that relies on simulating the physical paths that light rays take as they interact with the environment, including reflection, refraction, and scattering. This method provides high precision and detailed insight into the physical properties of the environment, making it ideal for scenarios where the optical paths can be clearly defined. However, it is computationally intensive and requires detailed knowledge of the environmental geometry.

In contrast, the diffusion-based model abstracts the environment into a statistical framework, often utilizing machine learning techniques to infer the relationship between the transmitter and receiver based on empirical data. This model is particularly useful in complex or dynamic environments where the exact optical paths are difficult to determine. While it may not provide the same level of physical detail as ray tracing, the diffusion-based model is more adaptable to real-world conditions and can efficiently handle environments with unpredictable or variable characteristics.

2.2. OWC Systems with Mirror Array-Based IRSs

OWC with mirror array-based IRSs has advanced rapidly in recent years. This surge in research interest and technological advancements highlights the potential of mirror array-based IRSs for significantly enhancing OWC systems. Table 3 summarizes the recent studies and key findings in this exciting field. In this section, we review the recent advancements in the application of mirror array-based IRSs in OWC systems. Our review is categorized into four main areas, each representing a critical aspect of OWC system optimization: channel gain optimization, link reliability optimization, data rate optimization, security optimization, and energy optimization. This categorization helps to systematically explore the multifaceted benefits of mirror array-based IRSs in enhancing the performance and reliability of OWC systems.

2.2.1. Channel Gain Optimization

Recent studies on the application of mirror array-based IRSs in OWC systems can be broadly categorized into four main areas based on the major optimization target. The first major area focuses on optimizing the system’s channel gain, to improve the overall performance and efficiency of OWC systems. In [46], it explores the enhancement of energy efficiency in a downlink IRS-assisted OWC system by simultaneously optimizing time allocation and power control. Unique to OWC systems, constraints such as non-negative transmission signals and limited amplitude are introduced. The authors reformulate the problem by reducing the number of variables involved and propose an iterative algorithm to optimize the time allocation and power control. This study demonstrates through simulations that the proposed method outperforms the conventional interior-point method, achieving an energy efficiency gain of up to 0.127 dB. In addition, channel gain optimization with IRSs in an OWC system with Non-Orthogonal Multiple Access (NOMA) has been investigated in [45]. Unlike conventional NOMA-based OWC systems where users’ decoding order and power allocation are based on the corresponding LOS channel gain, IRS-assisted systems allow the total channel gain at the receiver to be manipulated by tuning the IRS. The authors propose a framework for jointly designing the NOMA decoding order, power allocation, and IRS reflection coefficients to enhance the Bit Error Rate (BER) performance. They show that this multi-dimensional optimization problem is NP-hard and propose an adaptive-restart Genetic Algorithm (GA) to solve it efficiently. The IRS setup includes 100 IRS elements mounted on one of the walls in a 5 × 5 × 3 m³ room. The results demonstrate a significant BER reduction, achieving more than 7 dB gain in the transmitted signal’s Signal-to-Noise Ratio (SNR) for a BER of ${10}^{-2}$, compared to scenarios without IRS or with an IRS using fixed maximum reflection coefficients (i.e., all IRS reflection coefficients are set to one).

Table 3. Research progress in OWC systems based on mirror-array IRSs.

Year	Optical Source	Channel Type	Receiver	Optimization Objective	Optimization Algorithm	Comment	Ref.
2020	LD	Free Space	PD	Outage probability	N/A	Dynamic channel	[47,48]
2020	LED	Indoor	PD	Channel gain	Author’s algorithm	Energy efficient	[46]
2021	LED	Indoor	PD	Data rate	Sine–cosine algorithm	N/A	[22]
2021	LED	Indoor	PD	Data rate	Relaxing greedy algorithm	N/A	[49]
2021	LD	Free Space	PD	Outage probability	N/A	Atmospheric turbulence	[18,42]
2021	LED	Indoor	PD	Secure	PSO algorithm	N/A	[21]
2022	LD	Underwater	N/A	Outage probability	N/A	Underwater turbulence	[50]
2022	LED	Indoor	PD	Channel gain	Genetic Algorithm	NOMA	[45]
2022	LED	Indoor	PD	Channel gain	MM algorithm	SE maximization	[41]
2022	LED	Indoor	PD	Channel gain	Cyclic search algorithm	MISO	[51]
2022	LED	Outdoor	PD	Channel gain	BSCI algorithm	V2V	[52]
2023	LD	Underwater	N/A	Outage probability	N/A	Underwater turbulence	[53]
2023	LED	Indoor	PD	Channel gain	N/A	FD channel	[54]
2023	LED	Indoor	PD	Energy	ANN	SLIPT	[55]
2024	LED	Indoor	PD	Channel Gain	N-step localization algorithm	VLP	[40]
2024	LED	Indoor	PD	Secure	PSO algorithm	Multiple APs	[56]
2024	LED	Indoor	PD	Data rate	ANN-based algorithm	Blockage Avoidance	[57]

Abbreviations: LD—Laser Diode; PD—Photodiode; LED—Light-Emitting Diode; PSO—Particle Swarm Optimization; NOMA—Non-orthogonal Multiple Access; MM—Minorization–Maximization; SE—Spectral Efficiency; MISO—Multiple Input, Single Output; BSCI—Binary Search Convergence Iterative; V2V—Vehicle to Vehicle; FD—Frequency Domain; ANN—Artificial Neural Network; SLIPT—Simultaneous Lightwave Information and Power Transfer; VLP—Visible-Light Positioning; AP—Access Point.

Table 3. Research progress in OWC systems based on mirror-array IRSs.

Year	Optical Source	Channel Type	Receiver	Optimization Objective	Optimization Algorithm	Comment	Ref.
2020	LD	Free Space	PD	Outage probability	N/A	Dynamic channel	[47,48]
2020	LED	Indoor	PD	Channel gain	Author’s algorithm	Energy efficient	[46]
2021	LED	Indoor	PD	Data rate	Sine–cosine algorithm	N/A	[22]
2021	LED	Indoor	PD	Data rate	Relaxing greedy algorithm	N/A	[49]
2021	LD	Free Space	PD	Outage probability	N/A	Atmospheric turbulence	[18,42]
2021	LED	Indoor	PD	Secure	PSO algorithm	N/A	[21]
2022	LD	Underwater	N/A	Outage probability	N/A	Underwater turbulence	[50]
2022	LED	Indoor	PD	Channel gain	Genetic Algorithm	NOMA	[45]
2022	LED	Indoor	PD	Channel gain	MM algorithm	SE maximization	[41]
2022	LED	Indoor	PD	Channel gain	Cyclic search algorithm	MISO	[51]
2022	LED	Outdoor	PD	Channel gain	BSCI algorithm	V2V	[52]
2023	LD	Underwater	N/A	Outage probability	N/A	Underwater turbulence	[53]
2023	LED	Indoor	PD	Channel gain	N/A	FD channel	[54]
2023	LED	Indoor	PD	Energy	ANN	SLIPT	[55]
2024	LED	Indoor	PD	Channel Gain	N-step localization algorithm	VLP	[40]
2024	LED	Indoor	PD	Secure	PSO algorithm	Multiple APs	[56]
2024	LED	Indoor	PD	Data rate	ANN-based algorithm	Blockage Avoidance	[57]

Abbreviations: LD—Laser Diode; PD—Photodiode; LED—Light-Emitting Diode; PSO—Particle Swarm Optimization; NOMA—Non-orthogonal Multiple Access; MM—Minorization–Maximization; SE—Spectral Efficiency; MISO—Multiple Input, Single Output; BSCI—Binary Search Convergence Iterative; V2V—Vehicle to Vehicle; FD—Frequency Domain; ANN—Artificial Neural Network; SLIPT—Simultaneous Lightwave Information and Power Transfer; VLP—Visible-Light Positioning; AP—Access Point.

Moreover, Ref. [41] investigates the joint optimization of IRS configuration, power allocation, and user association in a Time-Division Multiple Access (TDMA)-based OWC system under the point source assumption. It provides detailed channel models for both LOS and NLOS links and derives the instantaneous signal expression to maximize the Spectral Efficiency (SE). The IRS configuration consists of 300 elements, each measuring 10 cm × 10 cm, uniformly distributed across a 6 m × 1 m rectangular area with a spacing of approximately 2.5 cm. A binary IRS coefficient matrix is introduced to simplify the configuration process, an alternating optimization algorithm is proposed to maximize SE, and the variable frozen algorithm and Minorization–Maximization (MM) algorithm are used to address subproblems.

The MM algorithm is used in the reviewed studies to optimize the IRS. The detailed steps of the MM algorithm in [41] for IRS optimization are shown in Algorithm 1. This algorithm constructs a surrogate function that approximates the original objective function and iteratively improves the solution. Specifically, the MM algorithm is applied to solve the power allocation problem with fixed parameters and Channel State Information (CSI) matrices.

Algorithm 1 MM algorithm to solve the IRS optimization problem [41]

Require:  Fixed parameters F, G, and ${ϵ}_2$, and CSI matrices $H^{\left(1\right)}$ and $H^{\left(2\right)}$. Ensure:  Suboptimal power allocation matrix P. 1: Init: Set iteration rounds $t\leftarrow 0$ and initialize $x_1^{*\left(0\right)}$, $\eta \left(x_1^{*\left(0\right)}\right)$, $\xi \left(x_1^{*\left(0\right)}\right)$, and $x_2^{*\left(0\right)}$. 2: repeat 3:    $t\leftarrow t+1$, $k\leftarrow 1$; 4:    Solve subproblem (P2-a) using the Proximal Gradient Descent (PGD) algorithm and find the power allocation matrix $P^{\left(t\right)}$; 5:    repeat 6:        Update $x_1^{*\left(t\right)}\leftarrow w{\gamma }_k\left(P^{\left(t\right)}\right)$; 7:        Calculate $\eta \left(x_1^{*\left(t\right)}\right)$ based on (38); 8:        Calculate $\xi \left(x_1^{*\left(t\right)}\right)$ based on (39); 9:        Update $x_2^{*\left(t\right)}\leftarrow {\sigma }_k^2+{\rho }_k^2{\sum }_{i=1,i\ne k}^K{\left(h_k^{\left(1\right)T}{P}^{\left(t\right)}{f}_i\right)}^2$; 10:       $k\leftarrow k+1$; 11:    until $k>K$ 12: until  $\parallel P^{(t+1)}-P^{\left(t\right)}{\parallel }_F<{ϵ}_2$ 13: Set $P^{*}\leftarrow P^{\left(t\right)}$.

Thanks to the MM algorithm, the SE has been significantly improved, achieving a nearly linear SE gain with respect to the number of IRS units and reflection factors. Additionally, the impact of geometric factors on IRS performance is highlighted, providing a detailed analysis of IRS locations and room sizes [41].

Previous papers have primarily focused on Single-Input–Single-Output (SISO) OWC systems. SISO systems often struggle with limited robustness, particularly against channel blocking, which can significantly impact their performance in high-capacity applications. To address these limitations, Ref. [51] investigates an IRS-assisted Multiple-Input–Single-Output (MISO) OWC system with a focus on channel gain improvements. The authors model the mirror array-based IRS-aided MISO-OWC channel shown in Figure 2 and formulate an optimization problem to reconfigure the direction of each IRS element. The IRS setup consists of a 50 × 30 array of mirrors, each measuring 0.1 m × 0.1 m, uniformly distributed within the simulated indoor environment. The goal is to maximize the asymptotic capacity at high SNR for the system employing Intensity Modulation/Direct Detection (IM/DD) with peak-power constraints. The non-convex optimization problem is converted into a Quadratic Programming (QP) problem with hemispherical constraints, which can be solved by calculating the maximum eigenvalue of an equivalent matrix. Simulation results show that the asymptotic capacity of the MISO-OWC channel improves with IRSs, with the NLOS channel gain increasing significantly as the number of mirrors increases. Specifically, when the number of mirrors increases from 15 to 1500, the NLOS channel gain increases from 0.09 to 0.22. The study also examines the impact of the distance between the receiver and IRS, showing that closer proximity results in greater performance improvement, achieving up to a $35\%$ capacity gain compared to systems without IRSs.

While prior research has predominantly centered on using IRSs in indoor OWC systems, Ref. [52] extends the application of IRS-based OWC to Vehicle-to-Vehicle (V2V) communication. It studies the effect of the number of IRS elements on the Energy Efficiency (EE) in an OWC system for parallel vehicles. The authors design a system with the transmitter on the right headlamp of one vehicle, the receiver between the headlamps of another, and the IRS on a streetlight pole shown in Figure 3. The achievable rate and power consumption are analyzed under different numbers of IRS mirror elements. An optimization problem for maximizing the EE is formulated, and the Binary Search–Conditional Iteration (BSCI) algorithm shown in Algorithm 2, which iteratively calculates a range and uses binary search to find the optimal solution, is proposed to efficiently find the optimal number of elements. The IRS system consists of a 60 × 60 array of mirrors, each measuring 0.01 m × 0.01 m, totaling 3600 mirror elements. Simulation results demonstrate that the proposed method significantly improves EE, using only a small fraction of the total mirrors available and requiring minimal computational effort compared to traditional methods.

Algorithm 2 BSCI algorithm [52]

Require:  Parameter values for the system components (e.g., LED, PD, IRS), initial iterative range $N_{min}$ and $N_{max}$, formulas for calculating the range and objective function. Ensure:  Optimal value $N_{\mathrm{opt}}$ and maximum objective $E{E}_{max}$. 1: Calculate the iterative range N based on the provided parameters. 2: Calculate $R\left(N\right)$ using the iterative range N. 3: for $N=N_{min}$ to $N_{max}$ do 4:    Calculate $EE\left(N\right)$ based on the objective function. 5:    if $EE\left(N\right)<EE(N-1)$ then 6:       $N_{\mathrm{opt}}$ does not exist; 7:       break; 8:    else if $EE\left(N\right)\ge EE(N-1)$ then 9:       $N_{\mathrm{opt}}=N$; 10:     $E{E}_{max}=EE\left(N\right)$; 11:   end if 12: end for 13: Use the binary search method to refine $N_{\mathrm{opt}}$ and $E{E}_{max}$. 14: Output $N_{\mathrm{opt}}$ and $E{E}_{max}$.

Unlike previous studies that have focused on time-domain characteristics, Ref. [54] investigates the Frequency-Domain (FD) channel characteristics of IRS-assisted OWC systems. A comprehensive tapped-delay line channel model for IRS-assisted OWC systems is proposed, and the frequency-selective channel characteristics due to the IRS-induced time delay are validated through experiments. Key findings indicate that the IRS array enhances OWC system performance when it operates as a narrowband system or when the strength of the reflected channel via the IRS array is significantly larger than the LOS channel. In wideband OWC systems, however, similar or worse performance compared to the LOS-only case is observed. The study also presents an achievable rate analysis using Direct Current-biased Optical Orthogonal Frequency-Division Multiplexing (DCO-OFDM), deriving a closed-form expression for the achievable rate. The results show that with a path loss ratio of zero, the achievable rate is 2.86 Gbps. Increasing the path loss ratio to two raises the achievable rate to over 4 Gbps. In scenarios where the IRS’s reflected signal is stronger than the LOS path, the achievable rate consistently improves. The channel gain measurements in various configurations exhibit low-pass characteristics due to the limited bandwidths of optical front-ends, with systems incorporating both LoS and IRS paths showing greater frequency fluctuations. These findings imply that IRS activation should be channel-dependent or require a gain control mechanism.

In addition to performance improvements in OWC systems, an IRS also offers valuable benefits for optical wireless-based positioning systems, such as Visible-Light Positioning (VLP) systems. In [40], the authors investigate a VLP system where an OWC receiver with a single photo-detector estimates its position using Received Signal Strength (RSS) measurements from multiple LED transmitters, including signals from both LOS and reflected paths via IRSs. The study’s main contributions include developing an achievable Cramer–Rao Lower Bound (ACRLB) expression for the 3D positioning accuracy, proposing a Maximum Likelihood (ML) estimator for position estimation, and optimizing the IRS orientation when LOS paths are blocked. The IRS setup consists of 21 × 21 arrays of units, each measuring 4 cm × 2 cm, placed on the four walls of the room, totaling 1764 IRS units. Additionally, an N-step localization algorithm, which iteratively refines the estimated position through multiple steps leveraging different IRS configurations, is introduced to improve positioning accuracy. The results show that the Root-Mean-Square Error (RMSE) of the ML estimator can be reduced to 4.2 cm with optimized IRS configurations, compared to 27.5 cm with the lowest noise variance and standard configurations. Extensive simulations demonstrate the impact of IRS parameters and LOS path blockages on positioning accuracy, showing that while IRSs offer limited improvements in the presence of sufficient LOS signals, they are crucial for accurate positioning when LOS paths are obstructed.

2.2.2. Link Reliability Optimization

Having explored the role of IRSs in optimizing the channel gain and positioning accuracy in OWC systems, IRS-based OWC systems have also been widely studied with a focus on link reliability optimization. Refs. [47,48] presents a comprehensive analysis of FSO systems assisted by IRSs, focusing on improving link reliability by reducing the outage probability. Unlike previous studies, this work incorporates pointing errors, IRS plane jitter, and obstruction of obstacles in the performance evaluation. Key contributions include deriving closed-form expressions for SISO and SIMO IRS-assisted systems, proposing a multi-branch IRS-assisted system adaptable to environments with obstacles, and providing asymptotic Probability Density Function (PDF) and Cumulative Distribution Function (CDF) expressions for IRS channel coefficients. The study also develops an optimal power allocation scheme and demonstrates the system performance gains with increasing intelligent channels. Numerical results validate the accuracy of the derived expressions and highlight the impact of system parameters on BER and outage probability. Specifically, it is shown that with a transmission power of 30 dBm, the outage probability is reduced from ${10}^{-2}$ to ${10}^{-5}$ with the assistance of IRSs. Similarly, Refs. [18,42] investigate the use of IRSs to optimize link reliability in FSO systems. The authors first specify the phase-shift distribution across the IRS for a desired beam reflection, demonstrating an equivalence to a mirror-assisted FSO system. They derive the Geometric and Misalignment Loss (GML) as a function of IRS characteristics and develop statistical channel models accounting for the impact of building sway on the transmitter, receiver, and IRS. The models are applicable to both 2D and 3D systems and include outage probability analysis considering atmospheric turbulence. Simulations validate the models and provide key insights for IRS-assisted FSO systems design. For instance, with a 10 cm IRS for a 1 km link, beam truncation is negligible. Additionally, the analysis of outage probability shows that identical position variances of the transmitter, IRS, and receiver affect the channel differently, depending on their relative positions.

While the aforementioned studies focus on enhancing link reliability in FSO systems, IRS technology can also be effectively applied to Underwater Optical Wireless Communication (UOWC) systems to achieve similar improvements. In [50], an IRS-assisted UOWC system was proposed to overcome signal losses in underwater environments caused by obstructions like aquatic plants, underwater vehicles, seamounts, and environmental conditions such as turbidity and salinity. These factors can cause significant burst errors and random errors in UOWC systems. The study characterized the IRS-assisted UOWC channel by combining underwater turbulence models, beam attenuation, occlusion due to obstacles, and pointing errors. It derived the PDF and CDF expressions for the SNR in a system with N IRS elements. Analytical expressions for outage probability were also derived and validated through Monte Carlo simulations. The results demonstrated that IRS could significantly reduce the outage probability; for example, at an average SNR of 40 dB, a system with 16 IRS elements could reduce the outage probability from $0.9\times {10}^{-2}$ to $0.8\times {10}^{-3}$. Moreover, in [53], the authors investigate the outage probability performance of IRS-assisted UOWC links, considering the effects of attenuation, pointing errors, and turbulence. Unlike [50], which focuses on the combined impact of occlusion and environmental factors, that study specifically models the underwater turbulent medium using the Oceanic Turbulence Optical Power Spectrum (OTOPS) model, which incorporates practical values for average temperature and salinity concentration in earth basins. The study finds that integrating IRSs into the underwater medium significantly enhances link reliability under various adverse conditions. Notably, the results show that without IRSs, the average BER is approximately $7\times {10}^{-3}$. However, with the addition of IRS elements, the outage probability decreases dramatically, to $1.5\times {10}^{-4}$ with 20 IRS elements, $1.9\times {10}^{-5}$ with 100 IRS elements, and further to $2.4\times {10}^{-6}$ and $9.9\times {10}^{-7}$ with 500 and 1000 IRS elements, respectively. This substantial improvement shows the potential of IRS technology in mitigating signal loss and enhancing communication reliability in challenging underwater environments.

2.2.3. Data Rate Optimization

In addition to optimizing the channel gain and link reliability in OWC systems, IRSs have also been studied for enhancing data rates in these systems. In [22], an IRS-aided indoor OWC system is proposed to address the LOS blockage problem while accounting for random receiver orientations. That study is the first to quantitatively explore the use of IRSs to enhance data rates and improve system reliability in the absence of an LOS path. The IRS setup consists of 10 × 30 mirrors, each measuring 0.1 m × 0.1 m, strategically placed to maximize system performance. The authors develop an optimization problem to configure the orientation of IRS elements to maximize the achievable rate. Due to the non-convex nature of the problem, a sine–cosine-based optimization algorithm is introduced, which is designed to solve complex optimization problems by using cooperative and competitive search agents and finding the global optimal solution. Simulation results demonstrate significant improvements in data rate and outage performance, with the proposed design achieving up to $397\%$ improvement in data rate and a $50\%$ reduction in outage compared to wall-only reflections. The study also highlights the impact of the number of blockers on system performance and confirms that the IRS-aided system outperforms no-IRS systems by up to $28.84\%$ in data rate across various scenarios.

In addition, in [49], the authors consider an IRS-aided OWC system that includes both NLOS paths specularly reflected by the IRS and LOS paths. Unlike previous methods that focused on optimizing mirror orientation, this study leverages the high spatial resolution of specular reflection paths and introduces a discrete matrix to represent the association between LEDs and IRS units. The IRS allocation is formulated as a binary programming problem. By pre-mapping IRS unit coefficients to emergence angles using a look-up table, the coefficients can be efficiently found using a reverse look-up table. A relaxing greedy algorithm, which iteratively improves the solution by selecting the best option at each step and relaxing constraints when necessary, is proposed to solve this binary programming problem, with the achievable sum rate as the evaluation metric. Complexity analysis and numerical results show that the proposed algorithm significantly reduces computational complexity compared to brute-force methods and performs well even with many users. Results indicate that using 256 IRS units can improve the achievable sum rate by nearly 12 megabits per second (Mbps) in high-SNR regimes, outperforming random selection and distance-based methods, with an additional gain of about 2 Mbps over the distance greedy method. The base data rate without IRS units and under shadowing conditions is approximately 5 Mbps.

Moreover, handover management in indoor VLC systems using IRS is investigated in [57], specifically addressing scenarios where the signal is blocked. The study proposes an active handover mechanism where, as users move indoors, IRS can reflect and redirect signals when the LOS path is obstructed by obstacles, ensuring that users continue to receive signals. The paper employs an Artificial Neural Network (ANN) to optimize the angle allocation of the IRS to enhance data rates and reduce handover latency during the transition. Experimental results demonstrate that using IRSs can significantly increase data rates (up to 60%) and reduce the number of handovers (up to 12%) compared to systems without IRSs. Additionally, in scenarios with high mobility, soft handover outperforms hard handover in maintaining connectivity and performance.

2.2.4. Security Optimization

In addition to enhancing the channel gain, link reliability, and data rate, IRS technology also holds significant promise for optimizing security in OWC systems. In [21], the authors propose an IRS-aided OWC system to enhance the secrecy performance. They model the reflected channel gain in an OWC system using an intelligent controllable mirror array as the IRS and derive a lower bound on the achievable secrecy rate for the IRS-aided peak power-constrained OWC system employing IM/DD. The study focuses on optimizing mirror orientations to maximize the difference between the channel gains of the legitimate user and the eavesdropper. The orientation optimization problem is transformed into a reflected spot position-finding problem, significantly reducing complexity. Simulation results show that without the IRS, the LOS channel gains of both legitimate users and eavesdroppers vary similarly. Specifically, when eavesdroppers are close to legitimate users, the channel gain difference decreases from 0.05 to 0 and remains close to 0 over a large range of approximately 1.6 m, resulting in limited secrecy rates and large insecure areas. With the IRS, the sum channel gain difference is significantly enlarged, leading to improved secrecy rates. Specifically, when using the IRS, the sum channel gain difference decreases from approximately 0.19 to 0, which only occurs when the eavesdropper is very close to the legitimate user. This critical distance is only 0.2 m, compared to 1.6 m without the IRS, showing a significant reduction in insecure areas. This results in a larger difference in the sum channel gain between the legitimate user and the eavesdropper, thereby enhancing the achievable secrecy rate.

Similarly, in [56], the authors study a mirror array IRS-aided secure OWC system with multiple transmitters and several independent two-dimensional rotatable mirror elements located at different positions. The goal is to create an optimal reflecting environment through mirror array IRS rotation control, enhancing legitimate user transmission while restricting eavesdropper reception. The study provides a closed-form expression of the achievable secrecy rate as a function of mirror orientations and AP-mirror element association. The optimization problem to maximize the secrecy rate is handled by a Reflected Spot Search (RSS) transformation method, which simplifies and splits the problem into subproblems. These subproblems are solved using an improved Particle Swarm Optimization (PSO) algorithm and a traverselike method. Additionally, the dual task of communication and illumination is considered, with an analysis of illumination uniformity. Simulation results show that the proposed RSS method achieves better secrecy performance than other IRS configuration methods with minimal impact on illumination uniformity. For instance, the results indicate that the legitimate user’s IRS gain is positive and close to their LOS gain, while the eavesdropper’s IRS gain remains zero unless their positions overlap. The simulation results are very similar to those of their previous study [21]. When eavesdroppers are close to legitimate users, the channel gain difference decreases from 0.05 to 0 and remains close to 0 over a large range of approximately 1.6 m without IRS. However, with the IRS, the sum channel gain difference decreases from approximately 0.6 to 0, which only occurs when the eavesdropper is very close to the legitimate user. This critical distance is only 0.2 m, compared to 1.6 m without the IRS, showing a significant reduction in insecure areas. Notably, the main difference between the two studies is that the sum channel gain difference is significantly enhanced when using IRSs.

2.2.5. Energy Optimization

Furthermore, IRS technology can also be employed to optimize energy in OWC systems. An indoor VLC system that utilizes a mirror-array IRS to enable Simultaneous Lightwave Information and Power Transfer (SLIPT) is studied in [55]. The main objective is to minimize energy consumption during user data transmission by optimizing the transmit power. To achieve this, the paper proposes a solution that leverages an ANN to predict user location and receiver orientation, thereby optimizing the beamforming matrix and transmitting power accordingly. The experimental results demonstrate that the SLIVER system with IRSs can reduce the transmit power by up to $60\%$ compared to conventional methods when dealing with user mobility, receiver orientation changes, and signal blockage.

3. Metasurface-Based IRSs in OWC Systems

In addition to mirror array-based IRSs, metasurface-based IRSs also offer significant potential for enhancing OWC systems. The operation of a metasurface-based IRS is fundamentally determined by the interaction between light waves and the engineered materials constituting the metasurface. Key factors influencing this interaction include dielectric properties, permeability, permittivity, and refractive index. High-permittivity materials like titanium dioxide (TiO₂) are used to control the phase shift and amplitude of reflected waves, enabling precise beam steering. While permeability variations are minimal in optical frequencies, they still contribute to the overall electromagnetic response. Permittivity, which dictates a material’s response to electric fields, is critical for phase modulation, especially in tunable designs like liquid crystal (LC)-based IRSs. Refractive index variations allow for control over the reflection angle and efficiency, essential for precise beam steering in OWC. These structures can modify various properties of the incident light, such as phase, amplitude, and polarization [58]. The effectiveness of these modifications is intricately linked to the careful selection of materials with specific dielectric, permeability, permittivity, and refractive index properties, leading to improved signal propagation, interference mitigation, and overall system performance.

In OWC systems, a mirror array-based IRS primarily relies on the reflection of light, and a metasurface-based IRS not only reflects signal light but can also be applied to refract signal light. Both metasurface and mirror-array IRSs can be positioned between the transmitter and receiver to utilize the reflective properties of light for signal redirection. A metasurface IRS further extends its functionality by employing refractive properties, allowing for advanced signal beam shaping [59,60] and control of the receiver’s FOV [28,61]. This versatility enables a metasurface IRS to be placed not only between the transmitter and receiver but also directly at the transmitter for beam shaping and at the receiver for signal concentration. The dual capability of controlling light through both reflection and refraction, combined with the careful selection of materials with specific dielectric, permeability, and permittivity properties, provides metasurface IRSs with a distinct advantage in optimizing OWC system performance across various deployment scenarios.

Table 4. Research progress in OWC systems based on metasurface-based IRSs.

Year	Application Type	Operation Wavelength	Dimensions	Ref.
2019	Reconfigurable beam system for NLOS OWC	1520–1620 nm	N/A	[62]
2020	Electro-optically tunable multifunctional metasurfaces	1522 nm	400 nm periodicity, various layer thicknesses	[63]
2021	Liquid-crystal IRS-based receivers	600 nm	0.75 mm thicknesses	[61]
2021	Electrically Tunable Polarization-Dependent Metasurfaces for dynamic optical applications	450 nm, 532 nm, 635 nm	400 nm period, 500 nm height TiO2 nanofins	[64]
2021	Perovskite metasurface for multiscale-optimized VLC	535–580 nm	135 nm depth, 295–460 nm period	[65]
2022	Beam-steering metasurfaces for full-duplex optical communication	1550 nm	2 mm × 2 mm	[66]
2022	Liquid-crystal IRS-based receivers	510–670 nm	0.75 mm thicknesses	[28]
2023	Metasurface-enhanced coherent multichannel OWC	1550–1555.8 nm	801.6 $\mathsf{\mu }$m × 801.6 $\mathsf{\mu }$m	[59]
2024	UOWC with metasurfaces auto-focusing Airy beam transmitter	440–640 nm	300 nm period, 800 nm height TiO2 nanopillars	[67]
2024	Tunable metasurfaces with fluid responsiveness	1550 nm	500 $\mathsf{\mu }$m × 500 $\mathsf{\mu }$m	[60]

summarizes the recent studies and key findings based on metasurface-based IRSs.

3.1. Signal Reflection with Metasurface-Based IRSs

Recent research has begun to explore the potential of metasurface-based IRSs for enhancing OWC systems through signal reflection. While the number of studies investigating this approach is relatively limited, here, we review some of the latest papers focusing on the implementation and benefits. The study [63] demonstrates a multifunctional electro-optically tunable metasurface designed for operation in the near-infrared wavelength regime. The metasurface is constructed with a layered structure comprising a gold (Au) back-reflector, an aluminum oxide (Al₂O₃) dielectric spacer, and an indium tin oxide (ITO) active layer, topped with gold fishbone nanoantennas. The ITO layer plays a crucial role, as its permittivity can be electrically tuned to achieve significant phase modulation, particularly near the epsilon-near-zero regime. This design allows for dynamic wavefront control at a subwavelength scale, enabling the metasurface to be reprogrammed for various optical functions without the need for mechanical movement. The authors demonstrate that by adjusting the voltage applied to individual metasurface elements, the device can achieve a wide range of phase shifts, leading to precise beam steering and variable focal lengths, as shown in Figure 4. The spatial distribution of the phase shift was meticulously designed to achieve precise control of the reflected beam, showcasing the potential for enhancing optical wireless communication systems through intelligent signal manipulation.

Moreover, paper [66] presents an innovative approach to enhancing high-speed full-duplex OWC systems using mass-manufactured beam-steering metasurfaces as shown in Figure 5. The metasurfaces are based on Silicon-On-Insulator (SOI) technology, featuring silicon nanobricks optimized for efficient phase modulation through Pancharatnam–Berry phase control. The SOI structure includes a 2 $\mathsf{\mu }$m thick silicon dioxide layer as an insulator, supporting 340 nm thick crystalline silicon nanobricks. These materials were selected not only for their structural stability but also for their superior optical properties, particularly in P-B phase modulation. The metasurfaces enable flexible beamforming by manipulating the polarization state of the incident light, supporting both point-to-point and point-to-multipoint communications through Wavelength-Division Multiplexing (WDM). Electromagnetic field modulation is achieved by optimizing the geometric parameters of the nanostructures, resulting in precise control over the beam direction. A key characteristic of this design is its compactness and scalability, which are made possible by standard Complementary Metal–Oxide–Semiconductor (CMOS) fabrication processes and allow for large-scale production on 8-inch SOI wafers. The metasurfaces, each measuring only 2 mm × 2 mm, achieve dynamic beam steering by changing the polarization of incident light, achieving a downstream capacity of up to 100 Gbps with a large beam-steering angle of $\pm {40}^{\circ }$. Up to 14 user channels can be supported and 10 Gbps upstream is achieved per user channel. Experimental results show that the power loss fluctuation for ten wavelengths at three different polarization states is less than 1.6 dB, demonstrating the system’s flexibility and stability, and highlighting its potential for large-scale commercial applications.

In addition, Ref. [24] systematically compares the performance and applications of mirror-array IRSs and metasurface IRSs in OWC systems. The authors propose an analytical framework to study the capabilities of adaptive metasurfaces and mirror array-based reflectors in directing radiated power toward a specific detector. They derive the phase gradients for metasurface arrays and the orientation of mirror array elements needed to focus the incident power onto the detector center. Expressions for irradiance on the detector plane for both reflector types are provided, along with a new metric to evaluate reflectors’ performance based on the received power. The study quantifies the received power gain compared to LOS scenarios and examines the impact of the number of reflecting elements and detector location through simulations. The findings reveal that using different types of IRSs (mirror array and metasurface) can significantly increase the received optical power. Specifically, the reflectors’ gain performance compared to the LOS link can achieve a factor of 30 in most areas under different positions of the two types of IRS. The mirror array maintains received power gains better at smaller distances, while the metasurface’s performance degrades as the detector gets closer. However, the mirror array’s superiority decreases with increasing distance between the reflector and the receiver.

3.2. Signal Refraction with Metasurface-Based IRSs

In addition to the light reflection capability, metasurface-based IRSs can also manipulate light through refraction, providing a powerful tool for enhancing OWC systems. Metasurfaces can achieve precise control over the spatial phase and polarization of incident light, allowing for advanced signal beam shaping and redirection. The refraction capability enables metasurface-based IRSs to be placed at various points in the communication link, including at the transmitter for beam shaping and at the receiver for signal concentration, thereby optimizing the FOV and improving overall system performance. This section summarizes recent advancements in using refraction-based metasurface IRSs in OWC systems.

3.2.1. Beam Shaping with Metasurface IRSs

Placing a metasurface-based IRS at the transmitter end allows for beam shaping [64,65], enabling the transmitted signal to be directed in a highly controlled manner. This approach can enhance the directivity of the signal, reduce interference [23], and ensure that the optical power is concentrated towards the intended receiver [62], thereby improving the efficiency and reliability of the communication link. The study [64] presents an Electrically Tunable Multifunctional Polarization-Dependent Metasurface (ETPM) designed for the visible range, integrating birefringent nematic LC with TiO₂ nanofin-based geometric phase metasurfaces. The ETPM features a vertically stacked structure where TiO₂ nanofins, encapsulated in poly(methyl methacrylate), enhance the refractive index contrast and improve device efficiency. The LC layer is electrically controlled to dynamically adjust phase retardation, enabling precise polarization manipulation and beam steering. The integration of these materials and structures allows for efficient electromagnetic field manipulation, critical for dynamic optical applications. This work has significant implications for OWC systems, where the tunable metasurface can be applied to IRS for beam shaping to enhance signal directionality and communication reliability in dynamic environments.

In addition, the authors propose an adaptive UOWC link using multi-wavelength lasers and an ultra-broadband metasurface for converting Gaussian beams to Circular Auto-Focusing Airy Beams (CAFAB) as shown in Figure 6 in [67]. The metasurface demonstrates uniform conversion efficiency across the visible light spectrum, enhancing beam propagation and mitigating occlusions. The study evaluates the performance of Red, Green, and Blue (RGB) CAFAB lasers in UOWC scenarios, highlighting their ability to maintain reliable, high-speed, and long-range communication even in the presence of underwater obstacles and air bubbles. Experimental results show that the proposed RGB Airy beam based UOWC system outperforms traditional Gaussian beam systems. Specifically, when the obstacle diameter is 0.43 ${\omega }_0$ (where ${\omega }_0$ is normalized to a maximum transmitter beam aperture of approximately 14 mm), the RGB Airy beam system achieves a data rate that is 7.3 Gbps higher than the RGB Gaussian beam system and 17.8 Gbps higher than the single blue laser Gaussian beam-based link. The Received Optical Power (ROP) for RGB channels with the metasurface is significantly higher compared to channels without the metasurface. Specifically, the red channel shows a 3.4 dB ROP improvement, while the green and blue channels demonstrate enhancements of 9.66 dB and 4.74 dB, respectively. Additionally, the data rate of RGB Airy beams remains superior even as the obstacle size increases, demonstrating the potential of IRSs for robust UOWC transmission in challenging environments.

In addition, a study in [60] introduces a fluid-responsive tunable metasurface designed to significantly enhance the performance of OWC systems. The metasurface is constructed from CMOS-compatible silicon nanobars on a fused quartz substrate, achieving a high cross-polarization efficiency of 98% at a 1550 nm wavelength. The design utilizes fluid infiltration to adjust the refractive index, allowing for real-time tunability and varifocal control, which is critical for dynamic beam steering and improving signal directivity. The fluid-responsive nature of the metasurface enables precise control over the focal length and beam direction by selecting different isotropic fluids, demonstrating a novel approach to electromagnetic wave manipulation. The study’s experimental validation shows a strong agreement between numerical simulations and measured results, confirming the metasurface’s ability to enhance the OWC network’s reliability and adaptability. The integration of spin-decoupled metasurfaces with fluidic environments provides a versatile platform for on-demand beam steering, making it suitable for advanced optical applications, including high-fidelity communication and information encryption.

Furthermore, in [59], a polarization-insensitive beam-steering metasurface was employed in a bidirectional coherent OWC system. The metasurface was composed of CMOS-compatible silicon nanobars on a fused quartz substrate, optimized for high cross-polarization efficiency at a working wavelength of 1550 nm. By employing fluid infiltration to adjust the refractive index, the metasurface achieved real-time tunability, enabling dynamic beam steering and improved signal directivity. The fluid-responsive nature allowed for precise control over the beam focus and direction by using different isotropic fluids, demonstrating the significant potential for light wave manipulation. Experimentally, the metasurface achieved a measured transmission efficiency of $17\%$, and the overall system showed enhanced OWC network reliability and adaptability. Moreover, the metasurface had a Polarization-Dependent Loss (PDL) of less than 1.2 dB, and all users maintained BERs below the soft decision Forward Error Correction (FEC) limit of $2.4\times {10}^{-2}$. This integration of optical coherent communication with metasurface technology reduced system complexity and cost while ensuring high data rates, rendering it a promising solution for future high-performance wireless communications. The ability to steer the beam accurately ensured that the signal was directed where it was most needed, enhancing the overall system efficiency and user experience.

3.2.2. FOV Enhancement with Metasurface IRSs

Positioning a metasurface-based IRS at the receiver end increases the receiver’s sensitivity to incoming signals, mitigates alignment issues, and enhances the overall robustness of the OWC system by concentrating the received optical power. In [28], a novel LC IRS-based receiver was designed to enhance the signal strength and data rate without using power amplifiers. The LC IRS could steer the incident light within the PD’s FOV by adjusting its refractive index. Simulation results showed that the LC IRS-based receiver could achieve up to $731\%$, $688\%$, and $591\%$ improvement in data rates for light wavelengths of 510 nm, 550 nm, and 670 nm, respectively, compared to ordinary receivers. The proposed optimization algorithm outperformed benchmark schemes and achieved optimal performance in fewer iterations. The study also highlighted the importance of maintaining the incident angle below ${60}^{\circ }$ to ensure higher transition coefficients and better overall channel gain. In addition, the paper [27] studied the LC-based IRS receiver in VLC systems to enhance signal detection and extend transmission range. The study focused on how adjusting the optical refraction properties of LC can amplify received light signals, thereby improving signal detection capabilities. The paper discussed the physicochemical characteristics of LC mixtures and proposed an LC-based IRS structure that could modulate its optical properties through an external electric field, enabling the amplification of low-intensity signals. The findings indicated that using an LC-based IRS receiver could extend the transmission range of VLC systems by up to 6.56 m.

Furthermore, in [61], the authors explore the use of IRSs in VLC systems to enhance the FOV of receivers, focusing on replacing traditional lens systems with dynamically tunable meta-elements. The paper introduces two key types of IRSs: metalens with electrically stretchable artificial muscles and an LC-based IRS structure. These elements enable the dynamic control of light beams by adjusting their physical dimensions and refractive indices, offering significant improvements in beam steering and signal reception. The IRS structure leverages advanced materials, such as titanium dioxide (TiO₂) embedded in LC cells, to achieve precise control over the light’s refraction and diffraction, thereby enhancing the electromagnetic field manipulation within the VLC system. This design allows for a flexible and efficient adjustment of the beam direction, improving the receiver’s ability to detect signals across a wide range of incidence angles. Around a wavelength of 600 nm, an IRS with a 0.2 mm physical depth can completely absorb incoming light, whereas increasing the depth to 0.8 mm reduces this absorption by about $90\%$, delivering nearly all the light intensity to the PD. These findings underscore the potential of metasurface IRSs in significantly improving the FOV and enhancing VLC system performance.

Overall, the application of IRS technology in OWC systems has evolved, starting from 2020, where researchers primarily used mirror array-based IRSs to optimize channel gain and outage probability, particularly in free space and indoor environments. This period focused on enhancing communication performance, with significant improvements in channel gain and data rate. By 2021, the application of IRS technology expanded, not only continuing to optimize channel models and security but also introducing metasurface-based IRSs for the first time. Early research on metasurface IRSs focused on dynamic optical applications and liquid crystal-based IRS receivers, gaining wider application in 2022, including full-duplex optical communication and multiscale-optimized VLC systems. These studies demonstrated the potential for more efficient transmission through material science advancements and signal manipulation. By 2023 and 2024, the application of IRS technology had expanded from traditional FSO to UOWC and full-duplex multichannel optical communication systems. The use of metasurface IRSs for adaptive beam focusing and dynamic adjustment highlighted the significant potential of IRS technology in addressing complex dynamic environments. Overall, the application of IRS technology in OWC systems has expanded with advancements in material science and control algorithms, showcasing substantial performance improvements across various environments.

4. Conclusions and Future Scope

This survey discussed two types of IRSs in OWC systems, each with its own advantages and limitations. Mirror array-based IRSs have shown effectiveness in reflecting incident light to enhance communication performance, addressing challenges such as signal blockage and interference mitigation. Studies have demonstrated significant improvements in channel gain and data rate, particularly in indoor OWC systems and V2V communication scenarios. Additionally, mirror array-based IRSs can enhance link reliability by reducing outage probability, even in complex environments with significant physical obstructions. Since mirror array-based IRSs rely primarily on signal reflection, they have a simple structure and easy deployment. However, their limitation lies in their inability to manipulate the amplitude of light, being restricted to altering only the spatial phase or direction of the light path. Additionally, mirror array-based IRSs can only be placed within the communication channel.

Metasurface-based IRSs offer further advantages by not only reflecting light but also manipulating it through refraction, due to the metasurface-based IRS being rooted in new meta-materials. This capability can manipulate the phase, amplitude, and direction of the incident light, enabling advanced signal beam shaping and focusing, enhancing the FOV, and improving overall system performance. The ability to place metasurface IRSs at various points in the communication link, including the transmitter and receiver, provides flexibility in system design and optimization. However, its complexity and dependence on material advancements pose major challenges.

Recent studies have leveraged various algorithms to control the IRS for optimal system performance, often based on static system models. However, real-world systems are dynamic, influenced by factors such as human movement in indoor OWC systems, weather changes, and turbulence in FSO systems. Thus, the ability to dynamically adjust and control IRSs in response to environmental changes is crucial.

Conventional feedback methods typically rely on predefined models and periodic measurements, which can be inadequate in rapidly changing environments. Machine learning-based methods offer a more dynamic alternative, capable of predicting and adapting to environmental variations using real-time data. However, machine learning approaches require significant data and computational resources and can introduce processing latency. Effective feedback control requires timely and accurate data acquisition about the channel conditions. This can be achieved through integrated sensors or external devices, ensuring low-latency communication with the IRS. Machine learning methods, while powerful, necessitate substantial computational resources and efficient algorithms to manage data processing in real time.

Moreover, dynamic environments present significant challenges for channel modeling with IRSs in OWC systems. Moving objects, varying weather conditions, and atmospheric turbulence can alter channel characteristics. Developing adaptive channel models that can predict and respond to these changes is essential. Hence, advancing feedback mechanisms, exploring machine learning-based approaches, optimizing data acquisition, and addressing channel modeling challenges are crucial for the future development of IRS-OWC systems.

In conclusion, the integration of IRSs in OWC systems holds significant promise for enhancing communication performance across various metrics. Both mirror array-based and metasurface-based IRSs have demonstrated their potential in optimizing channel gain, link reliability, data rate, and security. Ongoing research and development, particularly in dynamic control and feedback mechanisms, are crucial for realizing the full potential of IRS technology in future OWC systems.