Non-Parametric Approach Using ML Estimated Path Loss Bounded WCL for Video Capsule Endoscope Localization

Abstract

Wireless video capsule endoscope (VCE) is used to diagnose intestinal abnormalities. The physicians need to know the exact location of the images sent by the VCE for proper diagnosis and treatment of abnormalities. In this paper, we propose path loss bounded weighted centroid localization (PB-WCL) algorithm to estimate 3-D coordinate position of the VCE. The main challenge in VCE localization is the tissue characteristics of human body and the signal attenuation due to the complex non-homogeneous environment. The path loss suffers from high randomness due to the shadow fading and multi-path propagation effects caused by non-homogeneous tissues. To address the highly random scattered path loss issue, we propose maximum likelihood estimated path loss by using the received RF signal packets transmitted from randomly distributed neighboring positions of the VCE. Then we propose path loss bounded weight of the reference positions of sensor receivers which is not dependent on any channel parameters or unknown prior. Finally, we estimate the position of VCE using PB-WCL and apply two boundary conditions on the estimated positions to improve the accuracy. We develop a simulation platform using MATLAB to present and validate the results. We observe that our proposed position bounded PB-WCL shows significantly improved performance and outperforms the existing literature.