Performance Comparison of Various End-to-end Learning Technologies with a Bandwidth-Limited OWC System

Yuan Wei,Chaoxu Chen,Li Yao,Haoyu Zhang,Ziwei Li,Chao Shen, Unwen Hang,Nan Chi,Jianyang Shi

OPTICS EXPRESS（2024）

Cited 0|Views0

Abstract

Recently, end-to-end (E2E) learning methodologies have garnered significant attention as a compelling approach to attain global optimal communication within the domain of 6 G native intelligent systems. Nevertheless, a precise evaluation of the diverse E2E techniques is still lacking, leading to uncertainties regarding their applicable scenarios and effectiveness. In this paper, we present a comprehensive comparison applying three advanced E2E methods including the autoencoder-based geometric shaping (AEGS) model, comprehensive autoencoder (CAE) model, and wave-wise auto-equalization (WWAE) model in a real bandwidth-limited optical wireless communication (OWC) system. A novel attention-based comprehensive noise joint channel estimator (ACNJCE) is proposed to serve as a universal channel model adaptable to the existing E2E methods. Based on traditional carrier-less amplitude and phase modulation (CAP) modulation, AEGS, WWAE, and CAE are compared under the conditions of 2 GBaud and 3 GBaud respectively. The final results demonstrate that the CAE exhibits the capability to autonomously allocate bandwidth and achieves the highest dynamic adjustment range, which is increased by 69% compared with CAP based on neural network (NN) equalization. In contrast, AEGS has obvious advantages in terms of received optical power (ROP) gain. Based on bit-power loading discrete multi-tone modulation (DMT) modulation, WWAE can effectively compensate the signal spectrum after modulation order optimization and finally achieves the highest data rate under the condition that the - 3 dB bandwidth of the channel is only close to 1 GHz. The BER of WWAE with DMT at this rate is 25.2% of that using the NN equalization. Furthermore, experimental results under turbulent conditions reveal that AEGS exhibits superior and more stable performance amidst the perturbations caused by turbulence due to its ability to achieve end-to-end autonomous optimization while integrating traditional modulation and bringing additional shaping gain. According to our knowledge, this marks the first comprehensive evaluation and comparison of existing major E2E algorithms and traditional communication algorithms in a real OWC system. (c) 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

Translated text

求助PDF

上传PDF

Bibtex

AI Read Science

AI Summary

AI Summary is the key point extracted automatically understanding the full text of the paper, including the background, methods, results, conclusions, icons and other key content, so that you can get the outline of the paper at a glance.

Example

Background

Key content

Introduction

Methods

Results

Related work

Fund

Key content

Pretraining has recently greatly promoted the development of natural language processing (NLP)
We show that M6 outperforms the baselines in multimodal downstream tasks, and the large M6 with 10 parameters can reach a better performance
We propose a method called M6 that is able to process information of multiple modalities and perform both single-modal and cross-modal understanding and generation
The model is scaled to large model with 10 billion parameters with sophisticated deployment, and the 10 -parameter M6-large is the largest pretrained model in Chinese
Experimental results show that our proposed M6 outperforms the baseline in a number of downstream tasks concerning both single modality and multiple modalities We will continue the pretraining of extremely large models by increasing data to explore the limit of its performance

Upload PDF to Generate Summary

Must-Reading Tree

Example

Generate MRT to find the research sequence of this paper