Название: Deep Reinforcement Learning for Reconfigurable Intelligent Surfaces and UAV Empowered Smart 6G Communications
Автор: Antonino Masaracchia, Khoi Khac Nguyen, Trung Q. Duong, Vishal Sharma
Издательство: The Institution of Engineering and Technology
Год: December 2024
Страниц: 293
Язык: английский
Формат: pdf (true)
Размер: 11.1 MB

Reconfigurable intelligent surface (RIS) has emerged as a cutting-edge technology for beyond 5G and 6G networks due to its low-cost hardware production, nearly passive nature, easy deployment, communication without new waves, and energy-saving benefits. Unmanned aerial vehicle (UAV)-assisted wireless networks significantly enhance network coverage.

Resource allocation and real-time decision-making optimisation play a pivotal role in approaching the optimal performance in UAV- and RIS-aided wireless communications. But the existing contributions typically assume having a static environment and often ignore the stringent flight time constraints in real-life applications. It is crucial to improve the decision-making time for meeting the stringent requirements of UAV-assisted wireless networks. Deep reinforcement learning (DRL), which is a combination of reinforcement learning and neural networks, is used to maximise network performance, reduce power consumption, and improve the processing time for real-time applications. DRL algorithms can help UAVs and RIS work fully autonomously, reduce energy consumption and operate optimally in an unexpected environment.

This co-authored book explores the many challenges arising from real-time and autonomous decision-making for 6G. The goal is to provide readers with comprehensive insights into the models and techniques of deep reinforcement learning and its applications in 6G networks and internet-of-things with the support of UAVs and RIS.

The book chapters are arranged into four parts according to their topics. More specifically, the first two parts provide background and fundamentals on Artificial Intelligence and Deep Reinforcement Learning, while the last two illustrate how these principles can be applied to solve complex optimisation problems in the context of UAV-enabled and RIS-assisted networks. In Part I, there are two chapters that provide an essential introduction to Artificial Intelligence and deep neural networks. Specifically, Chapter 1 illustrates the relationship between Artificial Intelligence, Machine Learning, and Deep Learning. After introducing these concepts, Chapter 2 delves into the principles of deep reinforcement learning.

The basis provided in Part I will be very helpful for readers to understand the contents of Part II, which contains four chapters illustrating the most relevant aspects of deep reinforcement learning. In particular, this part starts with a brief introduction to Markov decision processes in Chapter 3. Subsequently, it delves deeper into the concepts of value function approximation and policy search methods, illustrated in Chapters 4 and 5, respectively. Finally, it concludes with Chapter 6, which provides an introduction to actor-critic learning.

Part III consists of five chapters focused on covering the aspects of UAV-enabled networks. This part starts with Chapter 7, which provides an overview of the potentialities and benefits of implementing UAV-enabled networks and also illustrates the main challenges and urgent research directions in this area. Chapter 8 demonstrates how the use of distributed deep deterministic policy gradient can solve the problem of power control in UAV-to-UAV communication, while Chapter 9 explores the potential of multi-agent deep reinforcement learning in solving the energy efficiency optimisation problem in UAV-to-UAV communications. Subsequently, Chapter 10 illustrates how the problem of real-time energy harvesting and communication scheduling in UAV-enabled networks can be addressed using a deep deterministic policy gradient method. Finally, Chapter 11 shows how the problem of 3D trajectory design and data collection in UAV-assisted networks can be efficiently solved through a deep reinforcement learning approach.

On the other hand, Part IV focuses on illustrating how deep reinforcement learning can be efficiently used to solve optimisation problems in RIS-assisted networks. An overview of RIS-assisted 6G communications is provided in Chapter 12, highlighting the main advantages and related challenges of integrating RIS surfaces into next-generation networks. Following this introductory part, the most relevant optimisation problems in RIS-assisted networks are addressed. Specifically, Chapter 13 illustrates a deep reinforcement learning approach for solving the joint optimisation problem of power allocation and phase-shift matrix for RIS-assisted D2D communications. Chapter 14 deals with joint resource allocation in RIS-assisted UAV communications for IoT with wireless power transfer using deep reinforcement learning. Finally, Chapter 15 explains how multi-agent learning can be beneficial for solving the joint power allocation and phase-shift matrix optimisation in networks supported by RIS and multi-UAVs.

Deep Reinforcement Learning for Reconfigurable Intelligent Surfaces and UAV Empowered Smart 6G Communications is aimed at a wide audience of researchers, practitioners, scientists, professors and advanced students in engineering, computer science, information technology, and communication engineering, and networking and ubiquitous computing professionals.




ОТСУТСТВУЕТ ССЫЛКА/ НЕ РАБОЧАЯ ССЫЛКА ЕСТЬ РЕШЕНИЕ, ПИШИМ СЮДА!