Pergamos - Library and Information Center of National and Kapodistrian University of Athens

Unit:

Κατεύθυνση Smart Telecom and Sensing Networks
(SMARTNET)
Πληροφορική

Deposit date:

2021-09-26

Year:

2021

Author:

Botirov Khojiakbar

Supervisors info:

Dimitris Syvridis, Professor, Department of Informatics and Telecommunications , National and Kapodistrian University of Athens.
Panagiotis Stamapoulos, Professor, Department of Informatics and Telecommunications , National and Kapodistrian University of Athens.
Stylianos Sygletos, Associate Professor, Aston Institute of Photonics Technologies, Aston University.

Original Title:

AI/ML assisted Li-Fi communication systems for the future 6G communication systems

Languages:

English

Translated title:

AI/ML assisted Li-Fi communication systems for the future 6G communication systems

Summary:

Information and communication technologies are developing rapidly, and tremendous growth along with advancements was observed over the last few decades. Requirements for bandwidth and capacity of current networks are overgrowing due to the increase in the use of high-speed Internet, video conferencing, streaming, Internet of things, etc. An ever-growing demand for increasing data volumes and multimedia services has led to an overload in the traditional radio frequency (RF) spectrum is used, and there is a need for transition from RF carrier to optical media. In this work, a novel Deep Neural Network (DNN) was proposed to mitigate nonlinearities caused by Perovskite material-based components of Li-Fi communication system, and measurement of Perovskite Photodiodes (PePD) the Optical Communications Laboratory in the National and Kapodistrian University of Athens. Due to the analysis of the PePDs bandwidth measurement, the highest cut-off frequency was measured 36,25kHz at 635nm wavelength. The proposed DNN showed promising results in comparison with Support Vector Machines (SVM) model, both models were trained on the dataset generated by OFDM based - Li-Fi systems. This technique successfully mitigates the nonlinearity of the PePD and the interference generated by the multipath channel. The simulation results reveal that the proposed scheme outperforms conventional techniques in terms of BER performance demonstrating the potential and validity of DL in the Li-Fi system.

Main subject category:

Technology - Computer science

Keywords:

Li-Fi, PePD, PeLED, DNN

Index:

Yes

Number of index pages:

Contains images:

Yes

Number of references:

Number of pages:

File: