Deep learning-enhanced NLOS NOMA-VLC system based on chirp spread spectrum modulation

Area of Science:

• Wireless communication technologies
• Optical wireless communication systems
• Smart city infrastructure

Background:

• Visible Light Communication (VLC) is a key technology for smart cities, complementing radio frequency (RF) networks.
• Non-orthogonal Multiple Access (NOMA) enhances spectral efficiency in VLC.
• Non-line-of-sight (NLOS) VLC faces challenges like signal attenuation and multipath effects, hindering reliability.

Purpose of the Study:

• To propose a robust NLOS NOMA-VLC system using Chirp Spread Spectrum (CSS) modulation.
• To develop a novel demodulator, DAG-SICNet, for signal compensation and recovery in mobile VLC environments.
• To address challenges in obtaining accurate channel state information for mobile VLC.

Main Methods:

• Implementation of an NLOS NOMA-VLC system with CSS modulation.
• Development of a denoising-attention-graph neural network combined with successive interference cancellation (DAG-SICNet) based demodulator.
• Joint signal compensation and recovery using the proposed DAG-SICNet architecture.

Main Results:

• The DAG-SICNet demodulator effectively mitigates linear and nonlinear impairments from NLOS propagation.
• Interference between multiple users in the NOMA-VLC system is successfully eliminated.
• Achieved data rates of 40.7 Mbps (User1) and 24.3 Mbps (User2) over 1.5 m, demonstrating high reliability.

Conclusions:

• The proposed NLOS NOMA-VLC system with CSS modulation and DAG-SICNet offers a reliable solution for obstructed wireless links.
• The DAG-SICNet demodulator significantly improves signal quality and user data rates in challenging VLC environments.
• This approach enhances the robustness and performance of VLC for future smart city applications.