AI‐Driven Cognitive Routing and Delay‐Aware Packet Scheduling for Next‐Generation Wireless Sensor Networks

ABSTRACT
Delays, energy‐constrained, and dynamic environments continue to present a challenge to wireless sensor networks (WSNs), which are an infrastructure in the next generation of smart infrastructure. The traditional routing and scheduling techniques are inadequate to the management of such issues since they are fixed‐point and lack flexibility. A new framework, named Artificial Intelligence‐Driven Cognitive Delay‐Aware Routing and Scheduling Framework (AICD‐RSF), is presented in this paper; it entails the use of context‐aware routing that depends on graph attention network (GAT), delay prediction relying on spiking neural network (SNN), Belief Rule‐Based Trust Evaluation (BRB‐TL), time‐slot preemption with the use of quantum‐inspired particle swarm optimization (QPSO), and decentralized model optimization with the help of AICD‐RSF is demonstrated to work outstanding against the classical protocols like AODV and DSR in a simulated NS‐3 network scenario. The end‐to‐end delay had gone down by up to 51.94 and the percentage rate of delivering the packets up to 98.3 and the use of energy was reduced by up to 41.1. The mean absolute error of delay prediction also had a possibility of 2.96 ms, and this was able to confirm the efficiency of inference of the SNN model. In addition to that, the use of trust‐based path selection was more secure, and the federated learning model was above 92.7% accurate and with low communication overhead. The model provides a solution to the key failures of the existing WSN models, including offering low‐power, self‐learning, and scalable one, which can be employed in high‐priority real‐time system applications, such as battlefield surveillance, smart healthcare, and critical infrastructure.