CNN-LSTM-AM approach for outdoor wireless optical communication systems

This paper introduces the enhancement of Visible Light Communications (VLC) for V2V using artificial intelligence models. Different V2V scenarios are simulated. The first scenario considers a specific longitudinal separation and a variable lateral shift between vehicles. The second scenario assumes...

Full description

Saved in:
Bibliographic Details
Published in:Scientific reports Vol. 15; no. 1; pp. 32178 - 26
Main Authors: Abdelsattar, Montaser, Amer, Eman S., Ziedan, Hamdy A., Salama, Wessam M.
Format: Journal Article
Language:English
Published: London Nature Publishing Group UK 01.09.2025
Nature Publishing Group
Nature Portfolio
Subjects:
639/166
639/166/987
Artificial intelligence
Bit error rate (BER)
Communication channels
Communications systems
Convolutional neural network (CNNs)
Datasets
Deep denoising autoencoder (DDAE)
Deep learning
Gated recurrent unit (GRU)
Generative adversarial network (GAN)
Humanities and Social Sciences
Intelligent systems
multidisciplinary
Neural networks
Power consumption
Roads & highways
Science
Science (multidisciplinary)
Transmitters
Vehicles
Deep learning
Convolutional neural network (CNNs)
Deep denoising autoencoder (DDAE)
Generative adversarial network (GAN)
Gated recurrent unit (GRU)
Bit error rate (BER)
ISSN:2045-2322, 2045-2322
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper introduces the enhancement of Visible Light Communications (VLC) for V2V using artificial intelligence models. Different V2V scenarios are simulated. The first scenario considers a specific longitudinal separation and a variable lateral shift between vehicles. The second scenario assumes random longitudinal separation and a specific lateral shift between vehicles. Significant obstacles that impair performance and dependability in V2V communication systems include bit errors, high power consumption, and interference. By combining Convolutional Neural Networks (CNNs), Generative Adversarial Network (GAN), Gated Recurrent Unit (GRU), and Deep Denoising Autoencoder (DDAE), this paper suggests a deep learning-based system to address these issues. The framework comprises four modules, a power reduction module that uses a GAN to generate low-power signals while maintaining signal quality; a performance enhancement module that uses GRU, a Bit Error Rate (BER) reduction module that uses a DDAE to denoise the received signal and minimize errors; and an interference cancellation module that uses a CNN-based U-Net to separate the desired signal from interference. It is shown that the suggested model significantly improves throughput, power efficiency, BER reduction, and interference cancellation. In dynamic and noisy contexts, our study offers a reliable and scalable way to improve the performance and dependability of V2V communication systems. The CNN-U-Net-GAN-GRU-DDAE model outperforms other models, including CNN-U-Net, CNN-U-Net-GAN, and CNN-U-Net-GAN-GRU, achieving the best results by an average percentage 13.6%, 14.4% and 4.2% respectively. By comparing this work with previous works, we deduce that the improving average percentage for our work by 31.7%.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:2045-2322
2045-2322
DOI:10.1038/s41598-025-16828-2