Visible Light Communication (VLC) Overview

• VLC is an optical wireless communication paradigm that uses LED-based intensity modulation with direct detection for both illumination and data transmission.
• Research highlights detailed channel modeling, modulation techniques like OOK and OFDM adaptations, and standards compliance with IEEE 802.15.7 for robust performance.
• Studies emphasize VLC's complementarity with RF systems, addressing challenges in indoor multipath, vehicular mobility, and energy-efficient network design.

Searching arXiv for recent and foundational VLC papers to support the article. Using the arXiv search tool to retrieve relevant VLC literature. Visible light communication (VLC) is an optical wireless communication paradigm in which visible light emitted by light-emitting diode-based devices is used for simultaneous illumination and data transmission. In the cited literature, VLC operates in the visible band of $400\text{–}800\,\mathrm{THz}$, corresponding to wavelengths of approximately $380\text{–}780\,\mathrm{nm}$, and is ordinarily implemented through intensity modulation with direct detection (IM/DD), with photodetectors, and in some cases CMOS/CCD image sensors, acting as receivers (Baranda et al., 2020). Across indoor networking, intelligent transportation systems, and hybrid heterogeneous networks, VLC is consistently treated as a complement to radio-frequency (RF) systems rather than a wholesale replacement: it offers unlicensed spectrum, immunity to electromagnetic interference, and strong spatial confinement, but it also inherits hard constraints from line-of-sight (LoS), field-of-view (FoV), ambient light, and coverage geometry (Ndjiongue et al., 2019).

1. Physical basis and channel characteristics

In VLC, digital information is conveyed through rapid variations in optical intensity that are undetectable to the human eye. The canonical LoS IM/DD model is expressed as

$y(t) = h \cdot x(t) + n(t),$

where $x(t)$ is the transmitted optical signal, $h$ is the channel gain, and $n(t)$ is additive noise (Baranda et al., 2020). This formulation captures a central asymmetry of VLC relative to RF: optical signaling must remain real and positive, which constrains admissible modulation formats and motivates optical adaptations of conventional baseband schemes (Ndjiongue et al., 2019).

For LoS propagation, the literature repeatedly uses Lambertian models. A representative form is

$$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$

for $\theta < \phi_{\text{max}}$, together with the simplified path-loss model $H_v = K d^{-\varsigma}$ for $d>1$ (Abuella et al., 2020). These expressions formalize the dependence of received power on distance, irradiance angle, incidence angle, optical filtering, and concentrator gain.

The same propagation physics underlies VLC’s principal strengths and limitations. Because light does not penetrate opaque objects or walls, VLC offers high spatial reuse, localized communication, and enhanced privacy and security. The same confinement also imposes short-range coverage, sensitivity to blockage, and strong dependence on LoS or high-quality reflected paths (Baranda et al., 2020). Survey literature further identifies ambient light sensitivity, uplink hardware limitations, and inter-cell interference in dense “attocell” deployments as recurring system-level constraints (Abuella et al., 2020).

2. Modulation, coding, and transceiver realizations

The modulation space of VLC includes On-Off Keying (OOK), Variable Pulse Position Modulation (VPPM), Pulse Position/Amplitude Modulation, Color Shift Keying (CSK), and optical OFDM variants adapted to IM/DD constraints (Abuella et al., 2020). In IEEE 802.15.7-compliant outdoor links, OOK with Manchester line coding is used to maintain constant average optical output and preserve regulatory lighting behavior; one experimentally validated infrastructure-to-vehicle-to-vehicle relaying system reached up to $380\text{–}780\,\mathrm{nm}$0 baud rate with 4-byte packets and no FEC under IEEE 802.15.7 PHY 1 specifications (Nawaz et al., 2019). For higher spectral efficiency in indoor settings, DCO-OFDM is used because of its robustness against multipath propagation and suitability for high data rates; in an automated valet parking demonstrator, 4-QAM DCO-OFDM with $380\text{–}780\,\mathrm{nm}$1, $380\text{–}780\,\mathrm{nm}$2, an IQ rate of $380\text{–}780\,\mathrm{nm}$3, and a data rate of $380\text{–}780\,\mathrm{nm}$4 supported bidirectional vehicular VLC links (Turan et al., 2021).

Open experimental platforms have made these PHY and MAC choices inspectable. "An Open-Source Research Platform for Embedded Visible Light Networking" introduced OpenVLC as a software-defined platform built around the BeagleBone Black, a simple opto-electronic front-end, an LED used for both emitting and sensing light, and a Linux device driver implementing PHY and MAC primitives together with TCP/IP interoperability (Wang et al., 2014). OpenVLC uses OOK with Manchester coding, basic and fast carrier sensing, and a CSMA/CD MAC; the reported MAC-layer throughput is up to $380\text{–}780\,\mathrm{nm}$5, median UDP throughput is $380\text{–}780\,\mathrm{nm}$6, median TCP throughput is around $380\text{–}780\,\mathrm{nm}$7, and 90% of ping packets have RTT below $380\text{–}780\,\mathrm{nm}$8 (Wang et al., 2014).

Receiver design is strongly application-dependent. In wearable computing, no single receiver architecture was found to operate efficiently across the full range of mobility-induced light conditions. Three receiver designs were therefore proposed: an ultra-low-power receiver built around a solar cell and thresholding circuit, and two high-speed photodiode/TIA-based receivers with high-gain and low-gain configurations. The ultra-low-power design operates at approximately $380\text{–}780\,\mathrm{nm}$9, supports up to $y(t) = h \cdot x(t) + n(t),$0 in darkness and about $y(t) = h \cdot x(t) + n(t),$1 under normal indoor illumination, while the low-gain high-speed receiver reaches up to $y(t) = h \cdot x(t) + n(t),$2 and the integrated switching architecture attains up to $y(t) = h \cdot x(t) + n(t),$3 with BER approximately zero (Varshney et al., 2018). This suggests that VLC transceiver design is inseparable from the intended energy, mobility, and illumination regime.

3. Network architectures, standards, and system organization

VLC networking literature distinguishes point-to-point and point-to-multipoint links, relayed and non-relayed systems, simplex and duplex operation, and homogeneous versus heterogeneous deployments (Ndjiongue et al., 2019). A recurring practical pattern is the aggregate architecture in which the downlink is optical and the uplink is RF, since user devices seldom provide convenient optical uplink hardware. One survey formalizes this through a Wide Optical Wireless Personal Area Network (W-OWPAN) architecture in which indoor LED downlinks are combined with RF uplinks, while inter-access-point links may use laser diodes for longer-range point-to-point optical transport (Ndjiongue et al., 2019).

IEEE 802.15.7 remains the principal standards reference in the cited corpus. The standard covers PHY and MAC layers, supports OOK, VPPM, CSK and related line coding and FEC choices, and spans data rates from bits per second to tens of megabits per second; PHY I is described as supporting $y(t) = h \cdot x(t) + n(t),$4, while PHY II and III reach up to $y(t) = h \cdot x(t) + n(t),$5 (Ndjiongue et al., 2019). The literature also notes related ecosystem activity by the Li‑Fi Consortium and VLCC, and identifies the CTTC “Demonstrador SILENCE” as a practical IEEE 802.15.7 implementation using SDR components, OOK in PHY I mode, and physical rates above $y(t) = h \cdot x(t) + n(t),$6 for unidirectional transmission (Baranda et al., 2020).

Architectural work has also emphasized spatial reuse. A software-defined multi-element VLC architecture based on a hemispherical bulb with many directional LEDs supports simultaneous multiple data streams, LoS discovery via VLC “SEARCH” frames and RF “ACK” frames, dynamic association, and hybrid VLC downlink/RF uplink operation (Mushfique et al., 2017). Related multi-element network studies formalize LED assignment, power control, and receiver combining under throughput, proportional fairness, and QoS objectives, with grouping-based optimum combining improving overall SINR by $y(t) = h \cdot x(t) + n(t),$7 to $y(t) = h \cdot x(t) + n(t),$8 (Eroglu et al., 2017).

4. Indoor performance engineering: multipath, rate adaptation, and QoS

Indoor VLC performance is often limited by multipath-induced intersymbol interference (ISI), especially in diffuse environments. One approach uses computer generated holograms (CGHs) to direct 30% of the best source’s output toward a $y(t) = h \cdot x(t) + n(t),$9 zone on the communication floor, thereby increasing the LoS contribution and reducing diffuse reflections (Younus et al., 2018). In an idealistic room, this reduced delay spread from $x(t)$0 to $x(t)$1, raised minimum 3 dB bandwidth from $x(t)$2 to $x(t)$3, and increased the OOK-supported data rate from $x(t)$4 to $x(t)$5 (Younus et al., 2018).

A complementary infrastructure-level strategy uses mirrors. "MirrorVLC: Optimal Mirror Placement for Multi-Element VLC Networks" formulates a two-stage optimization problem in which mirror placement and LED power are first chosen to maximize illumination uniformity, and LED-user association is then optimized to maximize the minimum user SINR (Mushfique et al., 2020). The reported gains are about threefold increase in average illumination and fourfold increase in average throughput relative to the no-mirror baseline (Mushfique et al., 2020). These results indicate that non-LoS energy, if geometrically controlled rather than treated as an impairment, can be converted into a joint illumination-and-throughput resource.

Rate adaptation is another core indoor problem because room-scale SNR can vary sharply with position, FoV, and shadowing. A Raptor-code-based receiver-side adaptation scheme eliminates the need for channel-state information at the transmitter and uses DCO-OFDM to combat multipath. In the modeled room, SNR varied from $x(t)$6 to $x(t)$7, throughput adapted smoothly between $x(t)$8 and $x(t)$9, and the Raptor-based method achieved about 10.6% better average throughput than adaptive modulation and coding (AMC) (Albayrak et al., 2019). By contrast, fixed-rate analysis under statistical queueing constraints studies the case in which the access point is unaware of channel conditions and must satisfy exponential bounds on buffer overflow and delay violation probabilities through cross-layer design (Hammouda et al., 2018).

5. Mobility, handover, and RF/VLC coexistence

Mobility is structurally difficult in VLC because optical cells are small, sharply bounded, and sensitive to blockage. Indoor handover work therefore replaces conventional RSS-threshold triggering with more context-aware mechanisms. One pre-scanning scheme stores a database mapping positions to optimal access points in a static $h$0 room with nine LEDs, estimates the user position from RSS, predicts the next position by

$h$1

and initiates disconnection and future-link association in advance, reducing handover time by eliminating scan and decision phases (Mishra et al., 2018). Another approach uses 2-bit cell IDs and the kurtosis of received cell-ID waveforms to trigger handover at overlap regions; this method is described as adaptive to ambient lighting and robust to co-channel interference (Perera et al., 2023).

Hybridization with RF is therefore not merely a convenience but a recurrent design principle. Survey and optimization papers distinguish “hybrid” RF/VLC systems, where each user connects to either RF or VLC, from “aggregated” systems, where a user may be served simultaneously by both (Aboagye et al., 2022). In aggregated systems, the total user rate is written as

$h$2

and joint access-point assignment, subchannel allocation, and power allocation are optimized while accounting for VLC line-of-sight blockages and inter-cell interference (Aboagye et al., 2022). The reported conclusion is that aggregated RF/VLC systems provide considerable improvement over hybrid designs in energy efficiency, sum-rate, and outage performance (Aboagye et al., 2022).

Energy-aware coexistence further sharpens the complementarity claim. A hybrid RF-VLC indoor access system that jointly optimizes user service and illumination is formulated as an NP-complete problem and addressed with an online algorithm having competitive ratio $h$3 with success probability $h$4 (Khreishah et al., 2018). Simulation results are reported to show that the hybrid system can reduce power consumption by more than 75% compared to using only WiFi or only VLC, particularly because communication over VLC can require little extra power when illumination is already needed (Khreishah et al., 2018). A common misconception is therefore that VLC is evaluated only as an RF substitute; the surveyed literature instead treats it as a heterogeneous component whose value depends on illumination demand, blockage statistics, and uplink architecture (Abuella et al., 2020).

6. Vehicular VLC and transportation systems

Vehicular VLC extends the same optical principles to headlights, taillights, traffic signals, and infrastructure luminaires. A safety-critical ITS demonstrator implements an infrastructure-to-vehicle-to-vehicle relaying chain using a regular LED traffic light, a photodiode receiver, and a digital Active Decode-and-Relay stage on Arduino DUE hardware (Nawaz et al., 2019). In corridor experiments from $h$5 to $h$6, the system achieved PER below $h$7 up to $h$8 at $h$9, minimum observed total latency of $n(t)$0, latency below $n(t)$1 for PER below $n(t)$2 at 99.9% confidence, and latency below $n(t)$3 even at $n(t)$4 (Nawaz et al., 2019).

In indoor automated valet parking, VLC is used as a bidirectional vehicle-to-infrastructure redundancy channel. The proposed system uses a VLC on-board unit to drive vehicle LEDs and receive from infrastructure LEDs, and a VLC roadside unit to send camera video, LiDAR point clouds, and aggregated online maps while receiving CAN Bus and mobility data (Turan et al., 2021). The reported physical-layer result is error-free LoS transmission up to $n(t)$5 with minimum SNR of $n(t)$6; at $n(t)$7, BER was $n(t)$8, and in directed-LoS with reflections BER was $n(t)$9 (Turan et al., 2021). The authors describe the link as jam-free, interference-resistant, location-based, and security-enhancing because communication remains spatially confined (Turan et al., 2021).

Recent work extends vehicular VLC to collective perception. A feasibility study using a custom LED bar and a high-speed digital camera evaluated application-level delay, outdoor range, and motion robustness for $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$0-byte collective perception messages (Nakano et al., 22 May 2026). At $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$1, about $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$2 was required for receipt and processing; error-free communication was observed at $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$3 and $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$4, BER at $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$5 was $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$6, BER remained under $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$7 at $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$8, and BER stabilized at $$H_v = \frac{(m+1)A_R}{2\pi d^2}~ T_s(\theta)~g(\theta)\cos^m(\phi)\cos(\theta),$$9 for vehicle speeds up to $\theta < \phi_{\text{max}}$0 (Nakano et al., 22 May 2026). Channel modeling for street-corner vehicular VLC further shows that non-stationarity from changing vehicle speeds, directions, and mobile scatterers materially alters channel gain and RMS delay spread; in one MIMO model, mean channel gain was $\theta < \phi_{\text{max}}$1 and the RMS delay spread was about 30 times higher than in the comparison model because of long single-bounce paths from mobile scatterers (Chen et al., 2020).

Taken together, these studies position VLC as a mature research area spanning optical PHY design, cross-layer resource allocation, spatially structured networking, and safety-critical vehicular systems. The literature does not describe a single canonical deployment model. Instead, it presents VLC as a family of optical wireless techniques whose practical value depends on how illumination, geometry, mobility, and RF coexistence are jointly engineered.