Electromagnetic signal processing for 6G smart radio environments

The evolution of wireless communications is increasingly driven by the dual imperatives of technological advancement and sustainable development. Meeting the stringent performance requirements envisioned for future sixth-generation (6G) networks, including ultra-high data rates, extreme reliability, ultra-low latency, and massive connectivity, necessitates a fundamental paradigm shift beyond conventional digital-centric architectures, which face a "digital bottleneck" due to the rapidly increasing energy consumption of massive antenna arrays and baseband processing units. This dissertation investigates Electromagnetic Signal Processing (ESP) and Smart Radio Environments as transformative approaches to overcome these limitations. A comprehensive physics-informed analytical framework is developed, integrating electromagnetic (EM) theory with signal processing and information-theoretic models. This framework enables rigorous characterization of the degrees of freedom (DOF) of the wireless channel in the near-field regime and provides a structured methodology to guide the design of scalable, energy-efficient, and high-performance wireless systems. The main contributions include: (i) task-specific exploitation of the DOF offered by the near-field channel, demonstrating significant benefits across multiple wireless applications, including orbital angular momentum (OAM) transmission via large intelligent surfaces for holographic communications, radio environment imaging for high-resolution sensing using extremely large multiple-input multiple-output arrays and reconfigurable intelligent surfaces, and grant-free coded spatial random access schemes for massive connectivity in industrial Internet of Things networks; (ii) design and optimization of metasurface-based EM processing architectures, including active transmitting reconfigurable intelligent surfaces and stacked intelligent metasurfaces, supporting near-orthogonal OAM mode generation, dynamic waveform control, and rapid reconfiguration, thereby enabling scalable, low-complexity EM-domain processing. These findings provide a first step toward harnessing programmable EM environments to realize sustainable, intelligent, and task-oriented 6G networks. The methodologies developed establish a foundation for further exploration, offering actionable insights into open challenges identified throughout this thesis. This work opens a pathway toward transformative 6G systems where extreme performance, sustainability, and intelligence are seamlessly integrated through ESP.