CASTELO: Convex Approximation based Solution To Elliptic Localization with Outliers

Wenxin Xiong*, Zhang-Lei Shi, Hing Cheung So, Junli Liang, Zhi Wang

*Corresponding author for this work

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

3 Citations (Scopus)

Abstract

This short communication considers mitigating the negative effects of possibly unreliable path delay measurements acquired in non-line-of-sight (NLOS) environments on the positioning performance, a problem deserving further investigation within the expanding research area of elliptic localization. We present CASTELO, a Convex Approximation based Solution To Elliptic Localization with Outliers, to achieve such a goal. Our proposal corresponds to a mixed semidefinite (SD)/second-order cone (SOC) programming formulation derived from an error-mitigated nonlinear least squares (LS) location estimator, presenting itself as a remedy for the neglect of positivity of NLOS biases suffered by the majority of currently fashionable outlier-handling approaches. In terms of analytical discussions, we provide rationales supporting the incorporation of the SOC constraints, which serve to tighten the problem obtained after SD relaxation, and conduct a complexity analysis for the ultimate mixed SD/SOC programming formulation. Simulations are carried out to confirm the strong ability of CASTELO to attain reliable elliptic localization in the presence of NLOS outliers. © 2024 Elsevier B.V.
Original languageEnglish
Article number109380
JournalSignal Processing
Volume218
Online published5 Jan 2024
DOIs
Publication statusPublished - May 2024

Research Keywords

  • Convex approximation
  • Elliptic localization
  • Non-line-of-sight
  • Outlier

Fingerprint

Dive into the research topics of 'CASTELO: Convex Approximation based Solution To Elliptic Localization with Outliers'. Together they form a unique fingerprint.

Cite this