Non-parametric Method of Path Loss Estimation for Endoscopic Capsule Localization

Abstract

Localization of the wireless video capsule endoscope (VCE) is necessary for proper diagnosis and treatment of the lesions in the gastrointestinal tract. In this paper, we propose path loss based VCE localization algorithm using weighted average of the sensors position. The main challenge in path loss based localization in human body is the scattered random deviation of path loss caused by the shadowing and multipath propagation of non-homogeneous medium. To address the randomness issue of the scattered path loss, we propose three non-parametric methods of path loss estimation using moving averaging, local weighted regression and local Gaussian weighted average. Then we use the degree based estimated path loss to calculate the weight of the sensors position. We propose a heuristic method of degree estimation for the estimated path loss. We develop a simulation platform using MATLAB to evaluate the performance of our proposed methods. The results show significant improvement in accuracy without any prior knowledge of distance related channel parameters.