Intelligent Framework for the Future-Proof Heterogeneous Ultra-Dense Networks: a Foresight for the Emergence of 6G

我们正在见证物联网(IoT)数量的惊人增长和革命性技术的出现，这些技术是我们日常生活中必不可少的部分，特别是那些由人工智能(AI)支持的技术。为了满足未来对高达1TB/S的高数据速率的渴求，研究人员和业界已经开始构思6G网络。我们最初的FRQNT团队项目提出了一个新颖的智能框架，考虑到海量的异质连接、吞吐量、高度多样化的需求和服务质量，为6G奠定了基础。提出的解决方案利用了图表示学习(GRL)、网络科学、通信理论和机器学习技术的最新发展，用于在异质超密集网络(H-UDN)中进行集群、资源管理(RM)和随机访问技术，这是基本的6G范式。 该方案是对FRQNT-项目的补充，并通过利用智能反射表面(IRS)的概念来提供超可靠和低延迟的通信，从而补充了其目标。IRS是6G移动通信的关键技术之一，它通过使用低成本的被动反射元件重新配置传播环境，显著提高了无线通信的性能。IRS系统将是H-UDN资源的重要组成部分，并将由原FRQNT项目中提出的技术进行管理。在本补充资料中，我们将扩展我们的H-UDN网络设计和分析，包括IRS辅助的数据传输，旨在提供高度的灵活性和卓越的兼容性。新的信道估计技术将被用于所提出的体系结构。我们将研究H-UDN中IRS参数的不同针对性优化，以改善网络性能指标，如总和速率、能量效率或中断概率。最后，对整个系统进行了性能分析。该项目提供工业4.0中需求最多的领域的前沿培训，并有助于增进人工智能和H-UDN方面的知识。

We are witnessing a formidable growth in the number of Internet of things (IoT) and the emergence of revolutionary technologies that are essential parts of our daily lives, especially those supported by artificial intelligence (AI). To satisfy the future thirst for high-data-rates, as high as 1TB/s, researchers and industry have started conceptualizing 6G networks. Our original FRQNT-Team project proposes a novel intelligent framework that lays down the foundations for 6G, considering the massive heterogeneous-connectivity, throughput, highly-diverse requirements, and quality-of-service. The proposed solutions leverage the recent developments in graph-representation-learning (GRL), network science, communication theory, and machine-learning techniques for clustering, resource management (RM), and random access techniques in the heterogeneous ultra-dense network (H-UDN), which is the essential 6G paradigm. This proposal is a supplement to the FRQNT-project and complements its objectives by leveraging the concept of intelligent reflecting surfaces (IRSs) to provide ultra-reliable and low latency communication. IRS is one of the key technologies for the 6G mobile communications that significantly improves the performance of wireless communication by reconfiguring the propagation environment using low-cost passive reflecting elements. The IRS system will be an essential part of the H-UDN's resources and will be managed by the techniques proposed in the original FRQNT project. In this supplement, we will expand our H-UDN network design and analysis by including the IRS-aided data transmission aiming to provide high flexibility and superior compatibility. Novel channel estimation techniques will be driven for the proposed architecture. We will investigate different pertinent optimization of the IRS's parameters in H-UDN to improve the network performance metrics such as sum-rate, energy efficiency, or outage probability. Finally, a performance analysis will be conducted for the whole system. This project provides leading-edge training in the most demanded fields of industry 4.0, and contribute to advance knowledge in AI and H-UDN.