Sum Rate and Fairness Optimization in RIS-Assisted VLC Systems

Salah Abdeljabar, Mahmoud Wafik Eltokhey, Mohamed-Slim AlouiniIEEE Open J. Commun. Soc 2024 · DOI

TL;DR

This paper explores how "smart mirrors"—known as Reconfigurable Intelligent Surfaces (RIS)—can solve the biggest weakness of Visible Light Communication (VLC): its extreme sensitivity to physical blockages. By using Genetic Algorithms and Particle Swarm Optimization, the researchers demonstrate that RIS can simultaneously boost total network speed and ensure fair data access for all users, even when they are moving.

What Problem Does This Paper Address?

Visible Light Communication (VLC) is a promising frontier for 6G, using the vast, unregulated spectrum of light from everyday LED fixtures to transmit data. It’s fast, secure, and immune to radio frequency interference. However, it has a "glass jaw": it relies heavily on a clear Line-of-Sight (LOS). If a person walks between the LED and the receiver, the connection can drop instantly.

To fix this, researchers are turning to Reconfigurable Intelligent Surfaces (RIS). Think of these as arrays of small elements on a wall that can be electronically controlled to reflect light toward specific users. The challenge lies in configuration: How do you tilt or assign these hundreds of elements to maximize the total data rate (sum-rate) without leaving users in the "dark" corners of the room (fairness)? This is a complex, non-convex optimization problem, especially when users are mobile.

Methods

The authors tackle the problem by modeling two distinct types of RIS elements:

  1. Specular Reflecting RIS: These act like tiny mirrors, reflecting light in a narrow beam. This requires a discrete optimization approach to decide which element serves which user.
  2. Diffuse Reflecting RIS: These act like Lambertian surfaces (similar to a matte white wall), scattering light more broadly. This requires continuous optimization of the elements' physical tilt (yaw and roll angles).

To solve these, they employed two meta-heuristic algorithms:

  • Genetic Algorithm (GA): Used for the specular RIS to handle the discrete "association" problem. It mimics natural selection to find the best combination of transmitter-RIS-receiver links.
  • Particle Swarm Optimization (PSO): Used for the diffuse RIS to find the optimal continuous angles for each element, mimicking the social behavior of bird flocks to converge on a solution.

They used a multi-objective utility function that allows network operators to tune the balance between raw speed and user fairness using a weighting parameter ($\eta$).

Datasets

As is standard for theoretical communication papers, the authors used a simulated environment rather than a static dataset. They modeled a 5m × 5m × 3m room with:

  • Two LED transmitters.
  • Four users with random locations and device orientations.
  • A square RIS array with up to 144 elements.
  • User mobility was simulated using the Random Waypoint (RWP) model, a common benchmark for evaluating mobile network performance.

Key Insights and Ideas

  • The Precision vs. Robustness Trade-off: Specular RIS provides much higher signal gains because it concentrates light, but it is "brittle"—it requires perfect alignment. Diffuse RIS provides lower gain but is far more forgiving of movement and orientation errors.
  • Fairness isn't Free: Maximizing the total sum-rate often means giving all the "light" to the user with the best connection. By introducing Jain’s Fairness Index into the optimization, the authors show you can significantly improve the experience for the "worst-off" users with only a minor hit to the total network speed.
  • Mobility Complexity: In mobile scenarios, there is a direct trade-off between how often you run the optimization (computational cost) and the resulting network performance. The paper quantifies the "robustness" of their PSO approach even when the optimization rate is slowed down.

Key Results

  • Performance Boost: Utilizing 144 RIS elements improved the network sum-rate by up to 40% and the Fairness Index by 20% compared to a standard LOS-only system.
  • Outage Reduction: The "link outage ratio" (the probability that a user's speed falls below 10 Mbps) was reduced by 20% to 42% when using optimized RIS.
  • Diffuse Superiority in Mobility: For mobile users, the optimized diffuse RIS offered a 100% sum-rate improvement over random or fixed orientation schemes.
  • Convergence: Both GA and PSO algorithms were shown to be efficient, converging to near-optimal solutions within 30–40 iterations.

Limitations and Future Work

The authors acknowledge a few areas for further exploration:

  • Hybrid Surfaces: The current study treats specular and diffuse surfaces separately. Future work could investigate "hybrid" RIS that combines both properties.
  • Multi-cell Interference: The model assumes a single-room environment; real-world deployments would need to account for interference from neighboring VLC cells.
  • Hardware Constraints: The study assumes ideal control over RIS elements. Practical limitations, such as the power consumption of the RIS controllers and the speed of the rotary motors (for specular mirrors), need to be addressed in physical prototypes.

Summary updated

BibTeX
@article{abdeljabar2024sum,
  author = {Salah Abdeljabar and Mahmoud Wafik Eltokhey and Mohamed-Slim Alouini},
  title = {{Sum Rate and Fairness Optimization in RIS-Assisted VLC Systems}},
  journal = {IEEE Open J. Commun. Soc.},
  year = {2024},
  doi = {10.1109/ojcoms.2024.3389093},
}
Figure