Intelligent Scheduling Mechanism for UWSNS using Mayfly-Lion Optimization and DSCN

Underwater Wireless Sensor Networks (UWSNs) support long-term aquatic observation but are afflicted with extreme resource scarcity, adverse channel conditions, and extremely unpredictable traffic priorities. To overcome these issues, we propose a hybrid intelligent scheduler—Mayfly Lion Optimization-Based Shepherd Convolution Neural Network (MLO-SCNN)—which combines swarm-intelligence search and deep, context-aware learning. The envisioned framework initially embeds node-level and queue-level features (delay tolerance, residual energy, hop distance, packet criticality) and supplies them to a Shepherd Convolutional Neural Network whose attention-facilitated "shepherd" module assigns priority to packets based on urgency. A Mayfly– Lion cooperative meta-heuristic at the same time adapts SCNN hyper-parameters and adjusts transmission queues for optimal performance while minimizing a multi-objective cost involving energy, latency, and packet delivery ratio (PDR), balancing exploration (Mayfly) and exploitation (Lion). Large-scale NS-3 acoustic-channel simulations involving 200 mobile sensor nodes reveal that MLO-SCNN increases average PDR by 28.4 %, reduces end-to-end latency by 17.6 %, and reduces per-node energy expenditure by 22.1 % over state-of-the-art PSO-CNN and EDF schedulers, while optimizing 34 % more quickly than single-algorithm optimizers. Such improvements extend network lifetime by 19 % and ensure QoS even in the presence of bursty, priority-mixed traffic. The findings prove that integrating deep learning with a two-phase evolutionary search presents an adaptive, priority-conscious scheduling solution appropriate to the extreme and dynamic underwater environment.