Tensor methods for array processing and channel estimation in wireless communications systems

• Autor(a) principal: Gomes, Paulo Ricardo Barboza
• Data de Publicação: 2018
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da Universidade Federal do Ceará (UFC)
• Texto Completo: http://www.repositorio.ufc.br/handle/riufc/37929
• topic: Teleinformática; Sistemas de comunicação sem fio; Tensor modeling; Array processing; Wireless communications systems; Channel estimation

In several applications in the field of digital signal processing, for example, wireless communications, sonar and radar, the received signal has a multidimensional nature which can intrinsically include on its structure many dimensions such as space, time, frequency, code, and polarization. In view of this, modern processing techniques which exploit all the signal dimensions can be developed to improve the system performance due to more accurate parameter estimation (for example: direction of departure, direction of arrival, delay, Doppler frequency, channel coefficients, phase noise) with powerful identifiability conditions. In this context, this thesis proposes new tensor modeling approaches for array processing and channel estimation applied to wireless communications systems. In the first part of this thesis, devoted to multidimensional sensor array and radar processing, we propose a new tensor-based preprocessing technique for noise supression which significantly reduces the noise effect in matrix and tensor data leading to more accurate estimates of the desired parameters. Then, new tensor methods capitalizing on the PARAFAC, Tucker and Nested-PARAFAC decompositions are formulated, from which new algorithms for joint direction of departure and direction of arrival estimation are proposed. In the second part of this document, tensor modeling approaches are developed to solve channel estimation problems in MIMO wireless communications systems. Firstly, we propose a new closed-loop and multi-frequency channel training framework that concentrates the processing associated with joint downlink and uplink channel estimation at the base station. We also show that the received closed-loop signal can be modeled as the PARAFAC decomposition of a third-order tensor. Then, the PARAFAC decomposition is also exploited to modeling a more realistic MIMO communication system that considers phase noise perturbations at each transmit and receive antenna. Receiver algorithms for channel and phase noise estimation are formulated. Simulation results are presented to illustrate the performance of the proposed receivers which are compared to state-of-the-art approaches.