A carregar...
Projeto de investigação
Sem título
Financiador
Autores
Publicações
Joint Channel and Nonlinearity Estimation for Memoryless Nonlinear Systems
Publication . Mokhtari, Zahra; Dinis, Rui; Hu, Sha; Kapetanovic, Dzevdan; DEE - Departamento de Engenharia Electrotécnica e de Computadores; Institute of Electrical and Electronics Engineers (IEEE)
System nonlinearity due to hardware impairments has always been a challenging issue. Distortion cancellation and iterative detection based receivers such as the Bussgang Noise Cancelling (BNC) receiver are used to detect the original data in the presence of strong nonlinear (NL) effects. However, these receivers require knowledge of the system nonlinearity which is usually unknown in practical systems. Bussgang decomposition and its general form denoted Generalized Bussgang decomposition (GBD), have been commonly used to model system nonlinearity. In GBD the nonlinearity output is decomposed as the sum of uncorrelated terms of increased orders and provides spectral characteristics of the useful and distortion terms. In this paper we consider nonlinearity at the transmitter side and model it with GBD. We aim to estimate the scalar weights in the GBD to later use them at the BNC receiver. However, knowledge of the channel is required to make a reliable estimate of the NL parameters. On the other hand the pilots for channel estimation are affected by the system nonlinearity, which can preclude reliable channel estimation. Therefore, in this paper we propose a joint channel and NL parameter estimation technique by designing appropriate training signals for each estimation phase (i.e. channel estimation and NL parameter estimation). We also derive a closed form expression for the average power of residual distortion in GBD with estimated parameters to see how well this model can characterize the nonlinearity. The results show that the proposed estimation technique has good accuracy and enables quasi-ideal performance for a BNC receiver.
On the Optimum Detection of MIMO-SVD Signals With Strong Nonlinear Distortion Effects at the Transmitter
Publication . Gonçalves, João; Nogueira, M. Teresa; Dinis, Daniel; Guerreiro, João; Dinis, Rui; Faculdade de Ciências e Tecnologia (FCT); Institute of Electrical and Electronics Engineers (IEEE)
Multiple-Input Multiple-Output (MIMO) architectures are now widely adopted in wireless systems, providing substantial capacity benefits by harnessing spatial diversity and spatial multiplexing. Nonetheless, the large Peak-to-Average Power Ratio (PAPR) associated with common pre-processing techniques, like Singular Value Decomposition (SVD), increases the system's susceptibility to nonlinear distortion. Conventional receiver designs that mitigate this distortion often neglect the fact that it has useful information on the transmitted data. Maximum Likelihood (ML) detection offers the capability to take advantage of the nonlinear distortion, but its inherent complexity is prohibitively high. This paper introduces a new MIMO receiver design aimed at exploiting the diversity introduced by the transmitter nonlinearities. It also provides an approximate bound on the achievable ML Bit Error Rate (BER) performance. Our results indicate that the proposed receiver can have a performance close to the ML receiver with just a few iterations.
A Low Complexity Linear Precoding Method for Extremely Large-Scale MIMO Systems
Publication . Berra, Salah; Benchabane, Abderrazak; Chakraborty, Sourav; Maruta, Kazuki; Dinis, Rui; Beko, Marko; Faculdade de Ciências e Tecnologia (FCT); Institute of Electrical and Electronics Engineers (IEEE)
Massive multiple-input multiple-output (MIMO) systems are critical technologies for the next generation of networks. In this field of research, new forms of deployment are emerging, such as extremely large-scale MIMO (XL-MIMO), in which the antenna array at the base station (BS) is of extreme dimensions. As a result, spatial non-stationary features emerge as users view just a section of the antenna array, known as the visibility regions (VRs). The XL-MIMO systems can achieve higher spectral efficiency, improve cell coverage, and provide significantly higher data rates than standard MIMO systems. It is a promising technology for future sixth-generation (6G) networks. However, due to the large number of antennas, linear precoding algorithms such as Zero-Forcing (ZF) and regularized Zero-Forcing (RZF) methods suffer from unacceptable computational complexity, primarily due to the required matrix inversion. This work aims to develop low-complexity precoding techniques for the downlink XL-MIMO system. These low-complexity linear precoding methods are based on Gauss-Seidel (GS) and Successive Over-Relaxation (SOR) techniques, which avoid calculating the complex matrix inversion and lead to stable linear precoding performance. To further enhance linear precoding performance, we incorporate the Chebyshev acceleration method with the SOR and GS methods, referred to as the Cheby-SOR and Cheby-GS methods. As these proposed methods require optimizing parameters, we create a deep unfolded network (DUN) to optimize the algorithm parameters. Our performance results demonstrate that the proposed method significantly reduces computational complexity from to O K2, where K represents the number of users. Moreover, our approach outperforms the original algorithms, requiring only a few iterations to achieve the RZF bit error rate (BER) performance.
Throughput analysis of movable antenna systems with mobility exploitation models
Publication . Candeias, Pedro; Oliveira, Rodolfo; DEE - Departamento de Engenharia Electrotécnica e de Computadores; Elsevier Science B.V., Amsterdam.
Movable antenna systems offer a promising approach to enhancing wireless communications by dynamically adjusting antenna positions to optimize signal reception. This paper explores the performance of a movable antenna at the receiver side considering a multi-tap propagation scenario in both Single-Input Single-Output (SISO) and Multiple-Input Multiple-Output (MIMO) systems. We introduce and evaluate four distinct mobility patterns that dictate antenna movement. The mobility patterns are part of the exploitation phase, aimed at probing the achievable capacity at different antenna locations. Additionally, we account for the time required to physically move the antenna in the system throughput, considering the worst-case scenario where the transmission is temporarily paused during the antenna motion. Our analysis assesses the performance of mobility pattern heuristics by examining their ability to balance the tradeoff between capacity gains from exploring new antenna positions and the downtime due to antenna movement. Simulation results show that specific antenna mobility patterns can achieve up to 70% of the SISO's optimal throughput or 77% of the MIMO's optimal throughput. The results reported in this paper show that simple mobility patterns more than double or triple the MIMO's or SISO's throughput compared to a scenario where the antenna remains fixed in a random position, respectively, underscoring the significant potential of the antenna mobility patterns in enhancing the MA system performance.
Human Posture Classification through Active Radars
Publication . Rodrigues, Miriam Salomé Ribeiro; Oliveira, Rodolfo
Over the years, the subject of human monitoring has attracted increasing interest in various
contexts. The continuous surveillance of less independent individuals such as the elderly,
children, or people with some kind of disability or medical condition is the motivation
for this dissertation. However, in these cases, privacy is an important factor, which is
why Radio Frequency (RF) Sensing has emerged as an alternative technology to other
types of sensors. This work tackles human posture recognition through the use of an
Frequency Modulated Continuous Wave (FMCW) radar. Two different postures are
defined to recognize an individual who is seated or standing. An extra scenario is also
categorized, in order to recognize an empty room, with no individual present.
In the first stage of this work, a visualization tool is presented to understand the data
behavior in addition to the classification of human postures. The visualization tool uses
techniques like the Simple Moving Average (SMA) and helps evaluate its effectiveness in
filtering the sampled signal. Additionally, the tool supports the visualization of how radar
signals alter depending on the subject’s proximity to the radar or the posture chosen.
In the second stage, the dissertation focuses on the classification of postures in real-
time. Like in most recent sensors and technologies, the FMCW radar also generates data in
a short period of time. In this particular case, the developed system receives a new sample
every 250 ms, which is seen as an upper bound of the algorithms’ computation time. In
order to cope with these requirements, this work compares Deep Learning techniques such
as Artificial Neural Networks (ANNs) with other mathematical solutions, such as Cross-
correlation, to handle the classification decision in real-time. The proposed techniques
are evaluated in terms of classification accuracy and computation time, showing that
deep learning techniques can achieve slightly higher classification accuracy but at the
price of longer computation times. An important aspect also studied in this work is
the importance of training datasets. The approach of using training samples obtained
at various distances to the radar shows higher reliability of classification when deep
learning techniques are adopted and can be practically experienced through the prototype
developed to demonstrate the outcomes of this work.
Unidades organizacionais
Descrição
Palavras-chave
Contribuidores
Financiadores
Entidade financiadora
Fundação para a Ciência e a Tecnologia
Programa de financiamento
Concurso de Projetos de I&D em Todos os Domínios Científicos - 2022
Número da atribuição
2022.08786.PTDC