A Hybrid Quantum–Siamese Framework with Congestion-Aware Contrastive Learning for Explainable Routing in MANETs and WSNs

The implementation of congestion-aware routing in MANETs and WSNs is challenging due to factors such as dynamic topology, unpredictable traffic, unreliable connections, mobile nodes, and limited energy resources. Equipped with relational Siamese representation learning and variational quantum feature transformation, this research introduces an Explainable Hybrid Quantum–Siamese Learning Framework (EHQ-SLF) for adaptive next-hop selection. The system uses a CACRL with margins updated dynamically based on congestion, connection dependability, and mobility to represent current and future nodes, along with joint node-state and link-state properties. In addition, domain-conditioned routing adjusts the decision-making procedure for MANET and WSN settings, and a variational quantum circuit modifies the learned relational representation. To identify trustworthy next-hop candidates across a variety of network scenarios, the method measures forwarding quality and forecasts congestion. To improve decision transparency, Quantum–Siamese Routing Explanation (QSRE) employs counterfactual reasoning, feature attribution, and quantum sensitivity analysis to identify routing decision factors. We evaluate the proposed architecture on several metrics, including packet delivery ratio, latency, energy efficiency, congestion prediction, mobility robustness, and explanation fidelity.