Recovering QPSK Modulated Visible Light Communication Signals Using LSTM-VAE

Visible Light Communication (VLC) systems have gained significant attention for their potential in high-speed wireless communication. However, the performance of VLC signals is often degraded by noise and non-linear distortions, especially at increased distances. In this study, we propose an innovative approach using a Long Short-Term Memory Variational Autoencoder (LSTM-VAE) to generate clean Quadrature Phase Shift Keying (QPSK) modulated VLC signals from noisy and distorted measurements taken at multiple distances. Our model leverages the sequential learning capabilities of LSTM networks to effectively capture temporal dependencies in the signal data, while the VAE framework ensures robust encoding and decoding processes. We demonstrate the effectiveness of our LSTM-VAE model through experiments, showing that it significantly reduces non-linearities, noise and improves signal quality. The proposed method is validated against benchmark dataset, and results indicate an enhancement in signal reconstruction, making it a promising solution for real-world VLC applications.