Experiments on Hybrid Multi-Color VLC/WiFi Network Optimization

This project investigates resource allocation in an indoor hybrid wireless network composed of multi-color VLC (Visible Light Communication) and WiFi access points. The goal is to maximize network performance by optimizing user-to-AP assignments and bandwidth allocation using reinforcement learning techniques.

• Room dimensions: 10m × 10m × 3m
• Receiver plane height: 1.2m
• VLC APs: 16  |  WiFi APs: 1
• Default number of UEs: 25
• Default FoV: 45°

• Transmit optical power:
  • Red: 20 dBm × 0.333
  • Green: 20 dBm × 0.38
  • Blue: 20 dBm × 0.287
• Bandwidth: 20 Hz
• Semi-angle at half power: 60°
• Optical filter gain: 1
• PD area: 1 cm²
• O/E conversion:
  • Red: 0.44 A/W Green: 0.23 A/W Blue: 0.15 A/W
• Power range: 50 μW to 10 mW

• Transmit power: 20 dBm
• Bandwidth: 20 MHz
• Noise spectral density: –174 dBm/Hz
• Receiver range: –125 dBm to 50 dBm

Simulation episodes: 700

Repetitions per configuration: 10

Fixed: N_UE = 25

FoV Range: 30° to 90°

Purpose: Evaluate impact of PD field of view on performance

Fixed: FoV = 45°

N_UE Range: 1 to 25

Purpose: Observe how increasing user density affects performance

• STP (System Throughput)

• AUS (Average User Satisfaction)

• SFI (Service Fairness Index)

• USR (User Satisfaction Rate)

Proposed:

Multi-Agent Reinforcement Learning (MARL):

• Each UE is treated as an agent deciding whether to connect to VLC or WiFi. The MARL model is integrated with a heuristic VLC bandwidth allocator to enhance throughput and fairness.

Baseline (Benchmark):

MCRAIC:

• A heuristic multi-cell resource allocation mechanism based on interference control.

The experiments aim to maximize system performance and fairness in indoor heterogeneous multi-color VLC/WiFi networks through intelligent access selection and resource allocation.

Figure 1: FoV vs. System Throughput (STP)

Figure 2: FoV vs. Average User Satisfaction (AUS)

Figure 3: FoV vs. Service Fairness Index (SFI)

Figure 4: FoV vs. User Satisfaction Rate (USR)

Figure 5: Number of UEs vs. System Throughput (STP)

Figure 6: Number of UEs vs. Average User Satisfaction (AUS)

Figure 7: Number of UEs vs. Service Fairness Index (SFI)

Figure 8: Number of UEs vs. User Satisfaction Rate (USR)