Relay Node Selection Based on Adaptive Neuro- Fuzzy Inference System in Delay Tolerant Networks

With the rapid growth of mobile networks and infrastructure-free environments, delay-tolerant networks (DTNs) have become an effective solution for communication in unstable conditions. However, selecting relay nodes intelligently remains a fundamental challenge for improving efficiency and reducing resource consumption. This study presents an adaptive neuro-fuzzy inference system (ANFIS)-based intelligent relay node selection method for DTNs, referred to as SR-ANFIS. The proposed method considers four key node characteristics—closeness centrality, reachability centrality, selfishness index, and visitation frequency—to calculate the reception probability (CR) and intelligently select the optimal relay node. Simulation results using The ONE simulator demonstrate that SR-ANFIS outperforms existing techniques, achieving approximately 25% higher message delivery, 20% lower delivery delay, and over 25% reduction in message drop and network overhead. These results confirm the effectiveness of the proposed method in enhancing routing performance and resource utilization in dynamic and uncertain environments.