Machine Learning-Assisted Reconfigurable Antenna Architecture for Adaptive Frequency Tuning in Dense Wireless Body Area Networks for Personalized Health Monitoring

Abstract

This article presents a machine-learning-enhanced reconfigurable textile antenna architecture that is specifically developed on the dense Wireless Body Area Networks (WBANs) to support personalized health monitoring. The system combines tunable radiating components, state switching circuit board and multimodal sensing to dynamically adjust themselves to varying on-body propagation conditions. It introduces a frequency-reconfigurable antenna based on textile-based PIN-diode switching, allowing three operation bands to be provided, including medical telemetry, ISM communication, and high-data-rate body-to-edge connection. In crowded WBAN settings, a machine learning model is used to predict optimal antenna state to maximize reliability and minimise communication outage through using real-time channel state indicators and packet-level metrics to train the model. Some of the experimental validation encompasses electromagnetic models simulations, prototype fabrication of textile, SAR compliance and indoor WBAN communication trials. It has been demonstrated that ML-aided tuning can help control the success rate of packets in dense networks by 18 percent and decrease the probability of link outage by 31 percent versus the baseline of using the same antenna type. The combination of the behaviour of tunable antennas with activity-sensitive health sensing (personalized) reflects the viability of the next generation intelligent wearable systems. The suggested architecture preconditions adaptive, user-conscious and context-responsive textile communication platforms.