Estimativa de direção de chegada de sinais utilizando análise de componentes independentes.

Abstract

This work constitutes with the study of an algorithm to estimate the direction of arrival (DoA) of signals using Independent Component Analysis (ICA). The DoA estimation is applied in several problems with electromagnetic sources, such as sensor networks and processing of smart antennas, being a topic investigated due to applications in Internet of Things (IoT) and in the fifth generation of cellphones (5G). The classical methods used to obtain DoA, as the Multiple Signal Classification (MUSIC) and the Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) have some restrictions. The detection of overlapping sources (which has the same DoA) is one limitation of these methods, since they can not estimate the presence of more than one source if they radiate in the same direction. Another constraint is that these methods require more sensors than electromagnetic sources to safely estimate DoA. In the classical ICA formulation was added the term related to phase delay to the signal matrix. This parameter is subsequently used to determine the direction of arrival. In this work, simulations were performed comparing the ICA method studied with the classical methods MUSIC and ESPRIT. It was possible to validate that the ICA can overcome the restrictions of the classic methods in relation to the estimation of overlapping sources and to work with a smaller number of sensors, that is, the same number of sensors and sources. Measurements were made with antennas to experimentally confirm the effectiveness of the extension of the ICA method to detect electromagnetic sources in these situations. Although the technique presents a greater error compared to algorithms like MUSIC, the algorithm was able to detect the DoA of all the sources in adverse situations, besides being able to relate each DoA to its respective electromagnetic source.