Robust TDoA Localization via Quadratic Weighted Least Squares in Degenerate Geometries

Authors

Abstract

This letter presents a zone-optimized Quadratic Weighted Least Squares (QWLS) algorithm for time-difference-of-arrival (TDoA)-based localization using wireless sensors. QWLS extends previously developed Parametric Three-Stage Weighted Least Squares (P3WLS) method by introducing a generalized formulation that remains robust under a broad range of eigenvalue conditions. Unlike P3WLS, which fails when the smallest eigenvector of the system matrix becomes nearly tangent to the TDoA constraint surface, QWLS introduces a second-order correction in the eigenbasis to enforce tangency and maintain physical consistency. We focus on the specific “worst-case” zones where P3WLS becomes unstable due to near-degenerate geometry and demonstrate that QWLS significantly improves localization accuracy in these regions. Simulation results across a mixed indoor-outdoor sensor deployment show that QWLS consistently outperforms both standard Weighted Least Squares and the original P3WLS method in structured failure zones

Additional Files

Published

2026-10-26

Issue

Section

Digital Signal Processing