Efficient signal detection in massive MIMO systems using low-complexity techniques
A carregar...
Ficheiros
Data
Autores
Título da revista
ISSN da revista
Título do volume
Editora
Resumo
As técnicas massive MIMO (Mulit-Input, Multi-Output) serão fundamentais nas comunicaçõessem fio de próxima geração, como 5G e 6G, pois permitem aumentar a capacidade e as eficiências espectrais e de potência. Contudo os sistemas massive MIMO requerem alta complexidade de implementação, nomeadamente devido aos requisitos de processamento de sinal para a receção. Nesta tese exploramos e propomos algoritmos de deteção iterativa avançados para sistemas massive MIMO, com foco na eficiência de potência e redução da complexidade de implementação. Começamos por introduzir um algoritmo de detecção iterativa eficiente baseado no método de Aceleração por Sobrerrelaxação Modificada (MAOR), sendo obtidos os valores ótimos para os parâmetros de aceleração e relaxamento que controlam a velocidade de convergência. Além disso, aplicamos uma solução inicial baseada em autovalores, em vez da comum aproximação por matriz diagonal, melhorando a taxa de convergência. Ao incorporar a aceleração por polinómios de Chebyshev, conseguimos um desempenho próximo do obtido por recetores MMSE (Minimum Mean-Squared Error) com complexidade muito mais reduzida, superando os recetores massive MIMO existentes. De seguida, estendemos o uso da aceleração de Chebyshev a vários métodos de deteção de baixa complexidade para esquemas massive MIMO. Quando comparadas com os métodos mais avançados atuais, as técnicas propostas apresentam um melhor desempenho em termos de probabilidade de erro, em conjunto com menor complexidade computacional. Além disso, desenvolvemos dois detetores iterativos, AC-AOR (....) e AC-SOR (...) que empregam o método acelerado de Chebyshev, e propomos redes DUN (Deep Unfolding Networks) para otimização dos parâmetros dos algoritmos propostos. As DUNs mostram melhorias significativas de desempenho e os processos de treino rápido e escalável superam os recetores convencionais em múltiplos canais, incluindo em condições não ideais. Adicionalmente, consideramos um sistema massive MIMO com técnicas SC-FDE (Single-Carrier with Frequency Domain Equalizaiotn) empregando recetores IB-DFE (Iterative Block Decision-Feedback Equalization). Introduziu-se uma nova estrutura de recetor IB-DFE compatível com os métodos iterativos existentes para evitar a inversão de matrizes, incorporando um método iterativo de Richardson acelerado por Chebyshev. Estas técnicas têm uma complexidade substancialmente reduzida quando comparadas com técnicas MMSE, apresentando desempenho similar, mesmo sob condições imperfeitas de informação do estado do canal. De forma geral, nossos métodos propostos demonstram avanços significativos na detecção iterativa para sistemas massive MIMO, oferecendo melhor desempenho, menor complexidade e robustez em várias condições de canal.
In next-generation wireless communication systems such as 5G and 6G, massive multiple- input multiple-output (MIMO) is already established as a key technology due to its ability to enhance capacity and efficiency. Despite these advantages, massive MIMO suffers from high complexity on the detector side. In this thesis, we explore and propose advanced iterative detection algorithms for massive MIMO systems, focusing on efficiency, low complexity, and enhanced performance. The research is divided into chapters, each addressing different aspects of iterative massive MIMO detection. Firstly, we introduce an efficient iterative detection algorithm based on the Modified Accelerated Over-Relaxation (MAOR) method. We derive the optimal values for the acceleration and relaxation parameters that control the convergence speed. Additionally, we apply an initial solution based on eigenvalues instead of the common diagonal matrix approximation, improving the convergence rate. By incorporating Chebyshev polynomial acceleration, we achieve near-minimum Mean Squared Error (MMSE) performance with reduced complexity, outperforming existing massive MIMO detectors. Next, we extend the application of Chebyshev acceleration to various low-complexity detection methods for massive MIMO schemes. Our proposed technique demonstrates superior error rate performance and reduced computational complexity compared to state-of-the-art methods. Furthermore, we develop two iterative detectors, AC-AOR and AC-SOR, enhanced through the accelerated Chebyshev method, and propose DeepUnfoldingNetworks (DUNs) for optimizing algorithm parameters via training. The DUNs show significant performance improvements and scalable, fast training processes, outperforming conventional detectors across various channel scenarios, including non-ideal conditions.Additionally, we consider a massive MIMO system with single carrier frequency domain equalization (SC-FDE) modulations employing Iterative Block Decision Feedback Equalization (IB-DFE) receivers. We introduce a novel IB-DFE receiver structure compatible with existing iterative methods to avoid matrix inversion, incorporating a Chebyshev accelerated Richardson iterative method. Our complexity analysis indicates substantial computational savings, with simulation results showing BER performance close to that of exact receivers, even under imperfect channel state information. In general, our proposed methods demonstrate significant advancements in iterative detection for massive MIMO systems, offering improved performance, reduced complexity, and robustness under various channel conditions.
In next-generation wireless communication systems such as 5G and 6G, massive multiple- input multiple-output (MIMO) is already established as a key technology due to its ability to enhance capacity and efficiency. Despite these advantages, massive MIMO suffers from high complexity on the detector side. In this thesis, we explore and propose advanced iterative detection algorithms for massive MIMO systems, focusing on efficiency, low complexity, and enhanced performance. The research is divided into chapters, each addressing different aspects of iterative massive MIMO detection. Firstly, we introduce an efficient iterative detection algorithm based on the Modified Accelerated Over-Relaxation (MAOR) method. We derive the optimal values for the acceleration and relaxation parameters that control the convergence speed. Additionally, we apply an initial solution based on eigenvalues instead of the common diagonal matrix approximation, improving the convergence rate. By incorporating Chebyshev polynomial acceleration, we achieve near-minimum Mean Squared Error (MMSE) performance with reduced complexity, outperforming existing massive MIMO detectors. Next, we extend the application of Chebyshev acceleration to various low-complexity detection methods for massive MIMO schemes. Our proposed technique demonstrates superior error rate performance and reduced computational complexity compared to state-of-the-art methods. Furthermore, we develop two iterative detectors, AC-AOR and AC-SOR, enhanced through the accelerated Chebyshev method, and propose DeepUnfoldingNetworks (DUNs) for optimizing algorithm parameters via training. The DUNs show significant performance improvements and scalable, fast training processes, outperforming conventional detectors across various channel scenarios, including non-ideal conditions.Additionally, we consider a massive MIMO system with single carrier frequency domain equalization (SC-FDE) modulations employing Iterative Block Decision Feedback Equalization (IB-DFE) receivers. We introduce a novel IB-DFE receiver structure compatible with existing iterative methods to avoid matrix inversion, incorporating a Chebyshev accelerated Richardson iterative method. Our complexity analysis indicates substantial computational savings, with simulation results showing BER performance close to that of exact receivers, even under imperfect channel state information. In general, our proposed methods demonstrate significant advancements in iterative detection for massive MIMO systems, offering improved performance, reduced complexity, and robustness under various channel conditions.