Error Analysis for Fingerprint-Based Localization

In this paper, the authors derive the theoretical error Probability Density Function (PDF) and Region of Confidence (RoC) conditioned on the on-line signal parameter vector, for a generalized fingerprint-based localization system. As the computations of these terms require the exact expression of the joint PDF for both the device location and the on-line signal parameter vector, which is often not available practically, they propose to approximate this joint PDF by Non-parametric Kernel Density Estimation techniques using the training fingerprints.

Provided by: National University of Singapore Topic: Security Date Added: Mar 2010 Format: PDF

Download Now

Find By Topic