A Q-learning-based mobility-aware tree routing protocol in wireless body area networks
TL;DR
This study proposes a Q-learning-based mobility-aware tree routing protocol (QMTR) for WBANs integrated with vehicular ad hoc networks (VANETs) to support reliable transmission of passengers’ health data to remote medical centers.
Abstract
Wireless body area networks (WBANs) are widely used in both medical and non-medical applications. In these networks, routing protocols face specific challenges due to energy constraints, bandwidth limitations, heterogeneity, node mobility, and dynamic topology. Although reinforcement learning (RL), particularly Q-learning, can improve routing performance in WBANs, existing Q-learning-based approaches often suffer from large state spaces, slow convergence, fixed learning parameters, inefficient topology management, and limited capability to recover routing failures. This study proposes a Q-learning-based mobility-aware tree routing protocol (QMTR) for WBANs integrated with vehicular ad hoc networks (VANETs) to support reliable transmission of passengers’ health data to remote medical centers. QMTR employs a two-stage screening mechanism to reduce the learning space. The first screening stage retains only neighboring nodes that provide geographical progress toward the destination, whereas the second stage eliminates routing candidates that may violate the tree structure. Furthermore, the learning rate is dynamically adjusted according to a comprehensive link-quality index consisting of link lifetime, link delay, and link stability. The protocol further incorporates an adaptive hello-message interval to balance topology accuracy and routing overhead, and a recovery mechanism that temporarily relaxes routing constraints to reconnect isolated nodes without restarting the learning process. Extensive simulations are conducted using Network Simulator 3 (NS3) to evaluate QMTR against existing routing schemes, including ARMR, VehiHealth, and GPSR, under varying vehicle speeds and packet-sending rates. In the first simulation scenario, the evaluation outcomes show that QMTR improves packet delivery rate (PDR) and network lifespan by 3.58% and 8.38%, respectively, and reduces end-to-end delay (EED) by 28.15%. Although QMTR introduces approximately 9.57% higher routing overhead than ARMR in this scenario, it achieves better reliability and energy efficiency. In the second simulation scenario, QMTR increases PDR and network lifespan by 3.73% and 6.48%, respectively, and decreases EED and RO by 21.92% and 13.70%, respectively.
