Capacities and performance improvement of localization methods using the co-prime based antenna arrays reinforced by machine learning

In the context of MIMO beamforming for 5G/6G, the estimation of the direction of arrival using antenna arrays will be crucial for retrieving directional information from sources and equipment. However, existing methods will face challenges, especially when dealing with many coherent sources or paths. Recent techniques based on sparse arrays have been proposed. In particular, techniques based on co-prime arrays can increase the capacity to estimate the direction of arrival of a greater number of sources for a given array dimension. However, most of these methods only consider the receiving array of antennas and cannot be directly used to localize coherent sources. Furthermore, these methods assume that the antenna positions are precisely known, which is only an approximation for many practical applications. It has to be said that for the frequency ranges of millimeter wave communication systems, the inaccuracy of antenna arrays will have much more negative influences than for lower frequencies.

In this project, we will focus on three main objectives:

Studying sparse array techniques at both the transmitter and receiver levels to enhance the ability of MIMO systems to localize as many sources as possible, particularly coherent sources.

Investigating the sensitivity of the proposed method to the imprecision of the antenna array positions, especially for millimeter wave systems.

Exploring machine learning-based methods to improve the robustness of the proposed techniques with respect to the imprecise knowledge of the antenna array positions.

Funding category: Financement public/privé
ANR/PEPR
PHD title: Doctorat en traitement du signal et télécommunication
PHD Country: France