Generalized hybrid LiFi-WiFi UniPHY learning framework towards intelligent UAV-based indoor networks

Generalized hybrid LiFi-WiFi UniPHY learning framework towards intelligent UAV-based indoor networks

Advancements in unmanned aerial vehicle (UAV) technology, along with indoor hybrid LiFi-WiFi networks (HLWN), promise the development of cost-effective, energy-efficient, adaptable, reliable, rapid, and on-demand HLWN-capable indoor flying networks (IFNs). To achieve this, a unified physical layer (...

Full description

Saved in:
Bibliographic Details
Main Authors: Rizwana Ahmad, Dil Nashin Anwar, Haythem Bany Salameh, Hany Elgala, Moussa Ayyash, Sufyan Almajali, Reyad El-Khazali
Format: Article
Language:English
Published: KeAi Communications Co., Ltd. 2024-01-01
Series:International Journal of Intelligent Networks
Subjects:
Unmanned aerial vehicles
Visible light communication
Auto-encoders
Deep learning
Neural networks
Online Access:http://www.sciencedirect.com/science/article/pii/S2666603024000277
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Advancements in unmanned aerial vehicle (UAV) technology, along with indoor hybrid LiFi-WiFi networks (HLWN), promise the development of cost-effective, energy-efficient, adaptable, reliable, rapid, and on-demand HLWN-capable indoor flying networks (IFNs). To achieve this, a unified physical layer (UniPHY) capable of simultaneous control communication, data transfer, and sensing is crucial. However, traditional block-based decoders, designed independently for LiFi and WiFi, perform poorly in complex and hybrid LiFi-WiFi-enabled UniPHY systems. In this study, we propose an end-to-end learning framework based on convolutional neural networks (CNNs) for UniPHY, which can be trained to serve hybrid LiFi-WiFi transmissions to improve error performance and simplify UAV hardware. In this work, the performance of the proposed framework is assessed and compared with that of the conventional independent block-based communication system. Furthermore, a comprehensive summary of optimal hyper-parameters for efficient training of our learning framework has been provided. It is shown that, with optimal hyper-parameters, the proposed CNN-based framework outperforms the conventional block-based approach by providing a signal-to-noise ratio gain of approximately 7 dB for the LiFi channel and 23 dB for the WiFi channel. In addition, we evaluate the complexity and training convergence for loss and accuracy.
ISSN:2666-6030

Similar Items