Performance of distance based and path loss based weighted centroid localization algorithms for video capsule endoscope

Abstract

In this paper, we propose distance and path loss based weighted centroid localization (WCL) algorithms for video capsule endoscope (VCE) using the received signal strength indicator (RSSI). We evaluate the performance of both algorithms considering real channel characteristics of human body. One of the major challenge in RSSI based VCE localization is the shadow fading and multi-path propagation effects of non-homogeneous medium of human body for which the measured RSSI is highly random resulting in high localization error. Again, due to the complex environment of experiment, accurate estimation of the channel parameters is quite difficult. We evaluate the performance of both algorithms in presence of randomness in path loss and estimation errors in channel parameters. To address the randomness issue, we estimate the smoothed path loss using moving averaging filter. Then, we introduce 10–50% errors in channel parameters to analyze the performance of both algorithms. We develop a simulation tool using MATLAB to visualize the results and to compare the performance. We observe significant improvement in performance by applying moving averaging method of smoothed path loss estimation using both algorithms. We also observe that the accuracy of distance based WCL decreases significantly in presence of errors in channel parameters. Whereas path loss based WCL is robust to the errors in channel parameters as it estimates the positions by using the estimated path loss directly without prior precise knowledge of channel parameters.