Intelligent Routing for GCN‐GRU and SAC‐Driven Software‐Defined Healthcare IoT

ABSTRACT In a healthcare Internet of Things (IoT) system, generally, data from sensors are to be communicated to cloud/edge servers for analysis. However, increasing reliance on healthcare IoT has increased the demand for efficient and intelligent data routing solutions to address challenges such as energy consumption, latency, and reliability. This work proposes an advanced routing framework that leverages software‐defined networking (SDN) to enhance adaptability and control in healthcare networks. The system integrates a graph convolutional network–gated recurrent unit (GCN‐GRU) model for accurate network state prediction and a soft actor‐critic (SAC) reinforcement learning approach for dynamic routing decisions. By combining traffic prediction with multi‐objective optimization, the proposed method minimizes energy consumption, reduces end‐to‐end latency, and improves the packet delivery ratio. This framework prioritizes critical healthcare data, ensures efficient resource utilization, and accommodates the unique requirements of wearable and bedside devices. Mininet simulations show the GCN‐GRU‐SAC framework achieves a throughput of 57 pkts/s, residual energy of 58 J, an average end‐to‐end delay of 25 ms, and a packet delivery ratio (PDR) of 96.5%, outperforming protocols like AODV (32 ms, 96.0% PDR), DDSRP (35 ms, 93.0% PDR), RLBEEP (30 ms, 96.0% PDR), IARP (20 ms, 93.0% PDR), and others (e.g., DC‐ACOP: 55 pkts/s, 54 J). This framework improves energy efficiency, prioritizes critical data, and ensures reliable transmission, offering a scalable solution for dynamic healthcare IoT environments.