Universidade Federal do Ceará
Programa de Pós-Graduação em Engenharia de Teleinformática

Defesa de Tese de Doutorado – Amarilton Lopes Magalhães

Título: Tensor Methods for Channel Estimation in Wireless Communication Systems Aided by Reconfigurable Metasurfaces

Data: 25/06/2026
Horário: 14:00
Local: https://meet.google.com/vxj-tnrg-fuh

Banca Examinadora:
Andre Lima Ferrer de Almeida (Orientador, Universidade Federal do Ceará – UFC)
Fazal-E-Asim (Universidade Federal do Ceará – UFC)
Walter da Cruz Freitas Junior (Universidade Federal do Ceará – UFC)
A. Lee Swindlehurst (University of California, Irvine – UCI)
Bartolomeu Ferreira Uchoa Filho (Instituto Tecnológico de Aeronáutica – ITA)

Resumo:

Recently, reconfigurable intelligent surfaces (RISs) have been envisioned as a key enabling technology for deploying future wireless networks, thanks to their remarkable ability to reshape the wireless propagation environment via software-controlled smart reflections. Additionally, several studies have investigated architectures beyond conventional RISs to improve adaptability, hardware efficiency, and system performance. This work presents contributions spanning different reconfigurable surface architectures, including conventional RIS, hybrid RIS, dynamic metasurface antenna (DMA), and fluid RIS (FRIS). By exploring the multidimensional structure of the signals, we use tensor decompositions to derive semi-blind receivers for joint channel and symbol estimation, given their effectiveness and powerful properties.

Firstly, in the field of conventional RIS, we proposed a channel estimation (CE) framework for a RIS-assisted wireless communication system that reduces the computational complexity of the joint estimation of channels and information symbols compared to the competitor method. Simulation results reveal that the proposed method performs similarly to the competitor while incurring substantially lower computational cost.

Second, for the hybrid RIS (HRIS) approach, a promising concept that blends sensing and reflecting meta-atoms, we formulate novel semi-blind receivers for HRIS-aided wireless communications that enable joint symbol and CE at both HRIS and BS. The proposed receivers provide symbol decoding capabilities for the HRIS and ensure ambiguity-free, separate CE without requiring an a priori training stage. Our findings demonstrate the competitive performance of the proposed solutions at the HRIS and the BS and reveal distinct performance trends across possible HRIS-BS receiver pairs.

Third, we investigate a DMA-based system and propose a CE scheme that extracts separate estimates of the wireless channel and the unknown waveguide propagation vector. Such decoupled estimates can be useful for isolating and compensating for waveguide propagation effects when designing the DMA beamformer. Numerical results corroborate the notable performance of the proposed scheme.

Fourth, we investigate channel estimation for fluid RIS (FRIS), whose dynamically reconfigurable element positions introduce an additional degree of freedom but also new position uncertainties. We propose a two-time-scale CE framework that jointly estimates the individual channels and motion-induced phase coefficients in closed form. To the best of our knowledge, this is the first CE approach for FRIS-assisted systems under position uncertainty, achieving robust performance under unknown position deviations.

The insights obtained from the investigated architectures can provide valuable guidance for receiver design and system optimization in future reconfigurable metasurface systems.

Palavras-chave: Reconfigurable intelligent surface (RIS), hybrid RIS, dynamic metasurface antenna, fluid RIS, tensor decompositions, channel estimation.