Pervasive Wireless Intelligence Beyond the Generations (PerCom)

从历史上看，新一代移动通信系统大约每十年构思一次。随着全球5G部署的推进和大流行带来的限制对经济的抑制，全球经济复苏需要强大的下一代解决方案来支持全球各地的智能制造、采矿、运输和农业，即使在偏远或恶劣的地理位置也是如此。子孙后代的成功将取决于三个主要方面：强劲的市场需求、坚实的商业模式和新技术。因此，这项前沿研究设想了激进的天空地一体化网络(SAGIN)解决方案，原因如下。至于市场需求，只有大约一半的世界人口能够接入互联网。作为关键基础设施的一部分，无线通信支持宽带市场的时机已经到来。其次，萨金建筑将卫星、飞机和地面车辆结合在一起。该模式在将无线宽带红利作为关键基础设施的一部分方面具有很强的潜力。第三，为了克服Sagin的各种关键部署问题，本项目旨在以下三个方面提供整体解决方案：(1)在Sagin下首次提出用于联合雷达和通信(RADCOM)的新波形，以便它们都从集成中受益，重复使用相同的硬件和相同的有限频谱，以支持引人注目的新应用。(2)开发可重构智能表面(RIS)形式的新型超材料，以提高Sagin覆盖率。显然，RISS由大量低成本的无源元件组成，这些无源元件不使用有源RF链来反射信号，从而消除了信号处理延迟和功耗。我们通过优化RIS模式来创新RIS配置。这种彻底的解决方案消除了对每个RIS元素的高复杂性信道估计的需要，这促进了在构思RIS辅助的SAGIN的相干/非相干自适应方面的重大突破。(3)针对Sagin的多目标优化问题，提出了一种新的深度学习工具，在任何给定的Sagin场景下，都可以通过信道编码获得接近容量的性能。显然，与将高复杂度的信道编码设计分解为两个级联卷积码的Turbo编码的成功类似，该项目提出了两个DNN的串/并级联，其中一个对另一个进行校正。这有助于在任何信道编码方案的上下文中指导其过程的高级DL设计，从而使接近容量的无线智能成为现实。该项目将是英国有史以来第一个超越显著权衡的项目，它通过找到最优解决方案的整个帕累托前沿来实现，最优解决方案的定义是所有最优解决方案的集合，其中任何优化的性能指标都不能在不影响另一个的情况下得到进一步改进。例如，这比纯粹的功率最小化或吞吐量最大化具有更大的挑战性。因此，该项目将使学术界和工业界受益并受到鼓舞。

Historically speaking, a new generation of mobile communication system has been conceived roughly every decade. As the global roll-out of 5G is progressing and the economy is stifled by the restrictions imposed by the pandemic, the global economic recovery requires powerful next-generation solutions right across the planet in support of smart manufacturing, mining, transport and agriculture even in remote or hostile geographic locations.The success of future generations will depend on three major aspects: strong market demand, solid business model and new technologies. Hence this frontier research conceives radical space-air-ground integrated networking (SAGIN) solutions for the following reasons. As for the market demand, only about half of the world population has Internet access. The time has come for wireless communication to support the broadband market as part of the critical infrastructure. Secondly, the SAGIN architecture incorporates satellites, planes and vehicles on the ground. This model has strong potential in extending the wireless broadband dividend as part of the critical infrastructure. Thirdly, to overcome a variety of critical deployment issues of SAGIN, this project aims to provide holistic solutions in the following three aspects: (1) New waveforms are proposed for joint radar and communication (RadCom) for the first time under SAGIN, so that they both benefit from the integration in terms reusing the same hardware and the same limited spectrum in support of compelling new applications. (2) Exploiting a new breed of meta-materials in form of reconfigurable intelligent surfaces (RIS) for improving the SAGIN coverage. Explicitly, RISs consist of a large number of low-cost passive elements, which reflect signals without using active RF chains, hence eliminating the signal processing delay and power-consumption. We innovate in RIS configuration by optimizing the RIS patterns. This radical solution eliminates the need for high-complexity channel estimation for each RIS element, which facilitates a major breakthrough in conceiving coherent/non-coherent adaptivity for RIS-assisted SAGIN. (3) New deep learning (DL) tools are conceived for the multi-objective optimization problems of SAGIN, where near-capacity performance can be achieved with the aid of channel coding in any given SAGIN scenario. Explicitly, in analogy to the success of turbo coding that breaks up a high-complexity channel coding design into two concatenated convolutional codes, this project proposes the serial/parallel concatenation of two DNNs, where one makes corrections for the other. This facilitates an advanced DL design that guides its process in the context of any channel coding scheme, which makes near-capacity wireless intelligence a reality.This project will be the first ever in the UK that goes way beyond striking trade-offs by finding the entire Pareto-front of optimal solutions, which is defined as the set of all optimal solutions, where none of the optimized performance metrics can be further improved without compromising another. This is vastly more challenging than pure power-minimization or throughput maximijzation, for example. Hence the project will benefit and inspire both the academic and industrial community.