| Title |
A Research on Increasing Transmission Capacity in Li-Fi via Autoencoder-based Compression and Sampling |
| Authors |
김세린(Serin Kim) ; 원용욱(Yong-Yuk Won) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.11.39 |
| Keywords |
Compressed data transmission; Autoencoder; Quantization; Li-Fi; Transmission capacity |
| Abstract |
This paper proposes a software-based compressed transmission architecture to enhance the data transmission capacity and reconstruction reliability of Li-Fi (Light Fidelity) systems based on Visible Light Communication (VLC), a next-generation wireless communication technology envisioned for the 6G era. To address the frequency response limitations and signal distortion inherent in Li-Fi systems, an efficient transmission scheme is designed by integrating an autoencoder-based image compression technique with NRZ-OOK modulation. The latent vector generated by the encoder is converted into a digital signal through quantization and sampling, while the receiver performs hard-decision-based dequantization and decoding to reconstruct an image closely resembling the original. The proposed approach is evaluated through Python-based simulations by varying parameters such as the latent dimension (Z_dim), quantization level, and samples per bit under different received SNR conditions. The results show that when the quantization level is set to 256 and the number of samples per bit is 8, the system achieves an average PSNR above 25 dB and an average SSIM above 0.95, even with a 96% compression ratio, provided that the received SNR exceeds 10 dB. Moreover, reducing the quantization level from 256 to 64 resulted in only a marginal degradation (less than 0.1), while doubling the transmission capacity. Increasing the number of samples per bit also improved robustness against channel noise. Overall, the proposed autoencoder-based compressed transmission scheme demonstrates that, despite a PSNR drop of approximately 9 dB compared to uncompressed transmission, it can achieve over a 50-fold increase in transmission capacity. This confirms its effectiveness in significantly improving throughput under constraints of limited time and bandwidth, making it a promising solution for handling the data explosion anticipated in 6G environments. Even in the presence of indoor reflections and obstacles, stable reconstruction performance was maintained when the received SNR exceeded 10 dB, ensuring reliable communication. These findings underscore the potential of compressed Li-Fi transmission for indoor deployment, particularly in secure environments such as government institutions, hospitals, and military facilities. |