MultiPath Component (MPC) tracking aims to trace the path parameter series in time-varying channel, which is vital in wireless mobile network. Conventional MPC tracking solution relies on the heuristic Multipath Component Distance (MCD), which cannot directly characterize the user mobility process. In this paper, Virtual User Position (VUP) is proposed to represent the geometric information of MPCs, which can reflect the user mobility in time-varying channel. Owing to the explicit location information in VUP, a deep learning based MPC Tracking neural Network (MPCTNet) is proposed inspired by similarities between MPC tracking and Multiple Object Tracking (MOT) in computer vision. To train MPCTNet more efficiently, a two-stage hierarchical training scheme is presented as well. Simulation results reveal the superiority of proposed MPCTNet over other conventional MCD based solutions.
A Novel Deep Learning Based Time-Varying Multipath Component Tracking Algorithm
2024-06-24
1275021 byte
Conference paper
Electronic Resource
English
Performance of Turbo Decoding for Time-Varying Multipath Channels
British Library Conference Proceedings | 2003
|The issue of spreading in multipath time-varying channels
Automotive engineering | 1995
|UAV Dynamic Tracking Algorithm Based on Deep learning
IEEE | 2021
|