Transformer-Aided Mobile Positioning for 6G Ultra-Dense Networks

Abstract

In this paper, we present a mobile positioning technique within the context of 6G wireless networks, utilizing multiple base stations to overcome the limitations of traditional positioning methods. At the core of our approach is cooperation of multiple base stations to capture the angle-delay channel power matrix (ADCPM) from multiple base stations for a mobile's position, leveraging spatial diversity and environmental interactions to enhance positioning accuracy. The application of the Transformer tailored for matrix-based data to these multi-dimensional ADCPM datasets enables a comprehensive analysis of environmental features and spatial layouts for understanding multiple signal paths. From numerical simulations on a ray tracing simulator modeling a 3GPP standard environment, we demonstrate that the proposed technique achieves an average positioning error within a sub-meter range.