Energy-Aware Routing Via Transformer-Based Spectrum Learning For Iot Mesh Networks

Energy efficiency and spectrum utilization are critical challenges in large-scale Internet of Things (IoT) mesh networks due to dense node deployment, dynamic traffic patterns, and limited battery resources. Conventional routing protocols often fail to adapt to rapidly varying spectrum conditions and energy constraints, leading to increased packet loss and reduced network lifetime. This paper proposes an Energy-Aware Routing framework based on Transformer-Driven Spectrum Learning (EAR-TSL) for IoT mesh networks. The proposed approach leverages a transformer architecture to learn long-range temporal and spatial dependencies in spectrum availability, interference patterns, and node energy states. By integrating attention-based spectrum prediction with adaptive routing decisions, the framework dynamically selects energy-optimal paths while minimizing spectrum congestion and transmission overhead. Extensive simulations demonstrate that the proposed EAR-TSL scheme significantly improves network lifetime, packet delivery ratio, and throughput while reducing end-to-end delay and energy consumption when compared with traditional energy-aware and machine learning-based routing protocols. The results confirm that transformer-based spectrum intelligence provides a scalable and robust solution for next-generation IoT mesh networking environments.