A Deep Learning-Based End-to-End Algorithm for 5G Positioning

Fifth generation (5G) new radio (NR) signals present unique opportunities for users and devices localization due to their higher bandwidth and more antenna elements. We propose an efficient deep learning-based end-to-end algorithm for 5G or beyond 5G (B5G) positioning. Both the channel geometric parameters and complex impulse response are utilized to exploit the connection between propagation channel and environment. A new multifeature fusion network architecture is proposed. Simulation results and complexity analysis are provided to demonstrate the efficacy of the proposed algorithm.