Abstract
As a result of the Internet of Things’ (IoT) explosive growth secure routing, energy optimization, and privacy preservation in resource-constrained environments have become major challenges. High overhead, static decision-making and susceptibility to malevolent attacks are common problems with traditional routing protocols. Federated Learning-Assisted Encrypted Routing based on Cost Function (FL-ERCF), an improved routing protocol that combines encrypted transmission with intelligent, privacy-preserving cluster head (CH) selection, is proposed in this paper to address these issues. The proposed protocol consists of three key operations: link quality assessment based on Received Signal Strength Indicator (RSS)I, SNR, and variance measurements trustbased clustering led by federated learning(FL) that has been trained using distributed IoT nodes to dynamically select the most suitable CHs and secure data transmission via a lightweight symmetric encryption algorithm that has been improved with digital certificates (DCs). FL can preserve the privacy of nodes at the local level and provide strong robustness to model flexibility and network dynamics. The proposed FL-ERCF protocol is implemented in the NetSim simulator and compared with stronger protocols such as Energy Efficient Secure Routing (EESR) and Hybrid Secure Routing (HSR). The performance evaluation demonstrates an improved packet delivery ratio (PDR) and reduced routing overhead and throughput even against on adversary attack. In addition, the adaptive and secure nature of FL-ERCF also makes it suitable for other mobile robotic networks like drone swarms and industrial robots, where trust, energy efficiency, and mobility are essential.