Performance Optimization for Mobile Wireless Sensor Networks Routing Protocol Using Adaptive Boosting With Sensitivity Analysis

Abstract

Mobile Wireless Sensor Networks (MWSNs) are employed in diverse applications, including remote patient monitoring systems (RPMS). In RPMS, biomedical sensors collect physiological data from patients outside clinical settings, and the data is transmitted wirelessly to healthcare providers for informed decisions. However, most routing algorithms focus on optimizing routing in static RPMS, neglecting mobile RPMS. This paper introduces an approach to improving the efficiency of MWSN algorithms, with a focus on the Termite Hill Routing Algorithm (THA) applied in RPMS. The investigation employs methods of sensitivity analysis to reveal how crucial parameters, such as the quantity of nodes, speed of nodes, and distribution of nodes affect the behavior and throughput of the algorithm. The paper introduces a novel methodology, Enhanced Regression-based Gradient Boosting (ERGB), which optimizes the algorithm’s parameters and enhances performance. ERGB is a unique combination of regression-based adaptive gradient boosting with sensitivity analysis and a robust machine-learning algorithm. It identifies and ranks the most critical factors that affect throughput in the constantly changing network environment of mobile RPMS. The study found that the network topology size and the source node speed are the most critical parameters impacting the algorithm, piquing the audience’s interest in this innovative approach. The study compared the optimized THA with default parameters and two other algorithms (AODV and Bee Sensor) used with optimized parameters. The results demonstrate significant improvements in throughput, reaching a maximum of about 2.6 Kb/s compared to 0.3 Kb/s with default parameters.