Skip to main navigation Skip to search Skip to main content

A Robust Machine Learning Based UWB AOA Estimation Method

  • Wenmin Zeng*
  • , Jialin Zhang*
  • , Tingting Zhang*
  • *Corresponding author for this work
  • School of Electronics and Information Engineering, Harbin Institute of Technology
  • Peng Cheng Laboratory

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In current main-stream ultra wideban (UWB) devices, hardware imperfections such as antenna mutual coupling, cross-polarization and signal distorations may prevent from achiving high accuracy angle of arrival (AOA) estimation. Due to the difficulties in hardware imperfection modeling, we present a non-parametric support vector regression (SVR) based robust AOA solution in this paper. A set of high relevant features extracted from received signals are adopted as the input for the rat swarm optimizer (RSO) - SVR method. Comprehensive experimental measurements are carried out, which show its performance advantages, particular in cases where traditional phase difference of arrival (PDOA) does not work well.

Original languageEnglish
Title of host publication2024 IEEE 99th Vehicular Technology Conference, VTC2024-Spring 2024 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798350387414
DOIs
StatePublished - 2024
Externally publishedYes
Event99th IEEE Vehicular Technology Conference, VTC2024-Spring 2024 - Singapore, Singapore
Duration: 24 Jun 202427 Jun 2024

Publication series

NameIEEE Vehicular Technology Conference
ISSN (Print)1550-2252

Conference

Conference99th IEEE Vehicular Technology Conference, VTC2024-Spring 2024
Country/TerritorySingapore
CitySingapore
Period24/06/2427/06/24

Keywords

  • Angle of arrival (AOA)
  • Support Vector Regression (SVR)
  • hardware imperfections
  • ultra-wideband (UWB)

Fingerprint

Dive into the research topics of 'A Robust Machine Learning Based UWB AOA Estimation Method'. Together they form a unique fingerprint.

Cite this