Location of mobile terminals based on LS-SVM with survey points in NLOS environment

The performance of mobile location systems is often limited by large errors primarily caused by non-line-of-sight (NLOS) propagation conditions. Nonparametric kernel-based estimators for location of mobile terminals using measurements of propagation delays have been proposed to mitigate the NLOS error. This paper proposes a novel estimation algorithm based on non-parametric density approximation of the measurements using survey data taken from the propagation environment. This method estimates the joint density function based on the Least Square Support Vector Machine (LS-SVM) solving linear operator equations. Simulation results indicate that the proposed algorithm provides better location accuracy even in severe NLOS conditions.