Local polynomial regression based path loss estimation for weighted centroid localization of endoscopic capsule

Abstract

Wireless video capsule endoscopes (VCEs) are used to diagnose intestinal abnormalities. By accurate localization of the VCE, proper diagnosis of the lesions are possible. In this letter, we propose path loss degree based weighted centroid localization (WCL) for VCE. The main challenge in path loss based VCE localization is the random deviation in the measured path loss caused by shadowing and multi-path propagation due to non-homogeneous medium. To address the randomness issue, we model the expected path loss using linear least square (LLS) method in the experimental phase. Then, we apply local polynomial least square method to estimate the path loss in the test phase. Finally, we propose path loss degree based WCL to estimate the coordinate position of the capsule. We simulate the proposed algorithm and visualize the results using a 3D simulation platform developed in MATLAB. We verify the accuracy of our proposed algorithm using different performance metrics and observe significant improvement over the literature without any prior precise knowledge of channel parameters or distance.