A Survey of Methods and Architectures for Hybrid Graph Neural Networks for Wearable IoT Monitoring Systems with Adaptive Algorithms and Energy-Efficient WSN Integration

Abstract

Wearable Internet of Things monitoring systems have significantly improved real-time data collection in healthcare, environmental monitoring, and smart living applications. However, the growing complexity of interconnected sensor networks creates major challenges related to scalability, adaptive decision-making, latency, and energy efficiency. Graph Neural Networks have emerged as effective solutions for modelling complex relationships in IoT environments because they efficiently process graph-structured data and capture dependencies among interconnected sensor nodes. Hybrid GNN architectures integrating graph convolutional networks, graph attention networks, and temporal learning models have demonstrated improved performance in analysing dynamic and heterogeneous IoT data. These models enhance sensor relationship representation, monitoring accuracy, and overall system robustness. Adaptive algorithms such as reinforcement learning and optimization-based routing strategies further improve wearable IoT systems by dynamically adjusting network parameters according to changing environmental and communication conditions. These techniques significantly reduce latency and improve system responsiveness. Additionally, energy-efficient Wireless Sensor Network integration has become an important research focus, where AI-driven routing and communication optimization methods help extend network lifetime and minimize power consumption. Despite these advancements, challenges including computational complexity, scalability, and real-time deployment remain significant concerns. Overall, integrating hybrid Graph Neural Networks with adaptive and energy-aware frameworks provides a promising direction for developing intelligent, scalable, and efficient next-generation wearable IoT monitoring systems.