标题： 基于变分自编码器的自动化仓库中人工智能驱动的无线电传播预测 显示英文标题

标题： AI-Driven Radio Propagation Prediction in Automated Warehouses using Variational Autoencoders

摘要： 接下来的十年将带来无线通信的深刻变革，这由对数据密集型应用日益增长的需求以及新兴技术的快速采用所推动。 为了充分发挥5G及更先进技术的潜力，需要在信号处理技术、创新网络架构和高效频谱利用策略方面取得重大进展。 这些进展促进了新兴技术的无缝集成，推动了工业数字化转型和连接性。 本文介绍了一种基于变分自编码器（VAE）的框架，名为WISVA（使用VAE的智能仓库无线基础设施），旨在准确地对自动化工业4.0环境中的室内无线电传播进行建模，例如在5G无线频段内运行的仓库和工厂地板。 本研究深入探讨了训练数据张量的精心创建，捕捉受各种障碍物影响的复杂电磁（EM）波行为，并概述了所提出的VAE模型的架构和训练方法。 通过其在不同场景下预测信噪干扰比（SINR）热图的能力，包括去噪任务、验证数据集、对未见过的配置进行外推以及之前未遇到的仓库布局，展示了该模型的鲁棒性和适应性。 本文展示了引人注目的重建误差热图，突显了WISVA相比传统自编码器模型的优越准确性。 本文还分析了该模型在处理复杂智能仓库环境中的性能，证明了其作为优化工业4.0无线基础设施的关键推动者的潜力。 显示英文摘要

摘要： The next decade will usher in a profound transformation of wireless communication, driven by the ever-increasing demand for data-intensive applications and the rapid adoption of emerging technologies. To fully unlock the potential of 5G and beyond, substantial advancements are required in signal processing techniques, innovative network architectures, and efficient spectrum utilization strategies. These advancements facilitate seamless integration of emerging technologies, driving industrial digital transformation and connectivity. This paper introduces a novel Variational Autoencoder (VAE)-based framework, Wireless Infrastructure for Smart Warehouses using VAE (WISVA), designed for accurate indoor radio propagation modeling in automated Industry 4.0 environments such as warehouses and factory floors operating within 5G wireless bands. The research delves into the meticulous creation of training data tensors, capturing complex electromagnetic (EM) wave behaviors influenced by diverse obstacles, and outlines the architecture and training methodology of the proposed VAE model. The model's robustness and adaptability are showcased through its ability to predict signal-to-interference-plus-noise ratio (SINR) heatmaps across various scenarios, including denoising tasks, validation datasets, extrapolation to unseen configurations, and previously unencountered warehouse layouts. Compelling reconstruction error heatmaps are presented, highlighting the superior accuracy of WISVA compared to traditional autoencoder models. The paper also analyzes the model's performance in handling complex smart warehouse environments, demonstrating its potential as a key enabler for optimizing wireless infrastructure in Industry 4.0.