Trabajo fin de estudios
| Título | Conformado de Haz Robusto para Sistemas 6G |
| Tutor | David Morales Jiménez |
| Estado | Ofertado |
| Tipo | TFM_Máster Ing. Telecomunicación |
Descripción:
Adaptive beamforming is a fundamental technology in modern multi-antenna wireless transceivers and has been a key enabler of the high connectivity and transmission speeds that support our daily communication needs. Whilst beamforming methods have been in place and well-established for years, we will need to design new robust methods which can accommodate the increasing complexity and dimensions of future (e.g., 5G and beyond) wireless communications; for instance, to meet the growing demands in terms of connectivity and data rates, emerging 5G systems will employ massively-large antenna arrays and large-dimensional signals which make traditional beamforming methods no longer suitable. Regardless of the criteria (e.g., maximum SNR or minimum MSE), optimal adaptive beamforming requires the inverse covariance matrix of the aggregated multi-user interference and noise. This is unknown in practice and needs to be estimated from training (observed) samples. Such estimation becomes particularly challenging in emerging wireless systems where the number of training samples (limited by latency requirements) is not far greater than the number of antennas (which will be massive). In such scenarios, adaptive beamforming based on the traditional sample covariance matrix (SCM) estimator fail remarkably, and new estimators based on diagonal loading or linear shrinkage are being proposed to deal with finite-sampling issues.
In this project, we will investigate new adaptive beamforming methods based on linear shrinkage covariance estimators which are robust to the inherent finite-sampling issues of emerging wireless communications. An exhaustive analysis of the performance of such methods will be carried out, considering different optimization criteria (maximum SNR and minimum MSE) and different system parameters (numbers of training samples and antennas); the new methods will be compared with traditional beamforming methods to assess the benefits of high-dimensional covariance estimation. The investigation will require to develop a suitable simulation platform in MATLAB and basic knowledge of statistics and probability.
This project is best suited to a student with an interest in the mathematical analysis and design of wireless systems. A high proportion of MATLAB programming is expected thus the student should be very confident in the use of basic MATLAB statistics routines and be capable of building a suitable simulation and testing platform.
Volver
