This paper, published in May 2020, presents a novel way to handle the "edge user" problem in dense Visible Light Communication (VLC) networks by combining the spectral efficiency of NOMA with the flexibility of hybrid RF relaying.
TL;DR
The authors propose a cooperative NOMA (Co-NOMA) scheme for multicell VLC networks where "strong" users act as relays, forwarding data to "weak" edge users via radio-frequency (RF) links. By iteratively optimizing power allocation and link selection, the system significantly mitigates inter-cell interference, boosting both total network throughput and user fairness compared to standard NOMA.
What Problem Does This Paper Address?
Visible Light Communication (VLC) uses LED ceiling lamps to transmit data, offering high speeds and unregulated spectrum. However, in a multicell environment (like an office with many lamps), users at the "cell edge" suffer from severe inter-cell interference (ICI) because they receive signals from multiple overlapping lamps.
While Non-Orthogonal Multiple Access (NOMA) allows multiple users to share the same frequency/time resources, edge users in a VLC-only NOMA system still struggle with low signal-to-interference-plus-noise ratios (SINR). Most existing research focuses on single-cell scenarios, ignoring the complex interference patterns of real-world multicell deployments. This paper tackles the multicell challenge head-on by introducing a hybrid VLC/RF cooperative relaying mechanism.
Methods
The authors model a system with $N$ access points (APs), each serving one "strong" user (near the center) and one "weak" user (near the edge). The core innovation is the Cooperative NOMA (Co-NOMA) link: the strong user decodes the weak user's message and forwards it using an RF link, powered by energy harvested from the VLC signal's DC bias.
To maximize the total network sum-rate, the authors developed a two-tier optimization strategy to solve a non-convex, mixed-integer problem:
- Inner Loop (User Level): For a fixed AP transmit power, the authors derived closed-form solutions to determine the optimal power split between the strong and weak users. Crucially, they also solve the link selection problem—deciding whether the weak user should receive data directly via VLC or through the hybrid VLC/RF relay.
- Outer Loop (Network Level): The authors use the Golden Section method to iteratively adjust the transmit power of each AP. This manages the interference "budget" across the entire room, ensuring that one AP's high power doesn't drown out its neighbors.
Datasets
The paper relies on Monte-Carlo simulations rather than physical datasets.
- Setup: 16 APs arranged in a grid on a ceiling.
- Evaluation: Results are averaged over 200 different random user distributions.
- Parameters: Standard VLC channel models (Lambertian radiation) and RF parameters (AWGN noise, specific bandwidths) are used, following established literature in the field.
Key Insights and Ideas
- The "Relay" as an Interference Buffer: By switching a weak user from a direct VLC link to an RF relay link, the system effectively "bypasses" the optical interference zone at the cell edge.
- Energy Harvesting Synergy: The DC bias required for VLC illumination isn't just wasted; here, it's used for energy harvesting to power the RF relaying, making the cooperation "self-sustaining" from a power perspective.
- Fairness vs. Throughput: Standard NOMA often prioritizes total throughput, which can leave edge users with nearly zero data. This Co-NOMA approach uses a Quality of Service (QoS) constraint ($R_{th}$) to ensure every user meets a minimum speed requirement without tanking the network's total performance.
- Interference Resilience: The study found that as the average interference level ($\alpha$) increases, standard NOMA performance collapses, while the proposed Co-NOMA scheme remains remarkably stable because it can dynamically shift users to the RF relay.
Key Results
- Sum-Rate Gains: The proposed Co-NOMA scheme consistently outperformed standard NOMA. In high-interference scenarios, the sum-rate for Co-NOMA remained high (around 650-700 Mb/s) while standard NOMA dropped significantly.
- Fairness: Using Jain’s fairness index, the authors showed that Co-NOMA maintains a fairness score near 0.9 (where 1.0 is perfect equality), whereas standard NOMA fairness drops toward 0.4 as interference increases.
- Convergence: Algorithm 2 (the outer loop) is computationally efficient, typically converging to a local optimum within 25 iterations for a 16-AP system.
Limitations and Future Work
- User Cooperation Assumptions: The model assumes strong users are always willing to act as relays. In the real world, battery drain (even with energy harvesting) and privacy concerns might make users reluctant to forward others' data.
- Fixed User Pairing: The paper assumes each AP serves exactly two users (one strong, one weak). Future work could explore dynamic user grouping where APs serve varying numbers of users.
- Hardware Complexity: Implementing a simultaneous VLC-receiver, energy-harvester, and RF-transmitter in a mobile device adds significant hardware complexity and cost not fully explored here.
- Mobility: The simulation assumes static users; the impact of user movement on the stability of the Golden Section optimization remains an open question.
Summary updated