Dynamic Energy Management in UASN Co-Driven by Layered Cheetah Optimization and Power Prediction

To address the critical energy constraints and complex communication challenges in underwater acoustic sensor networks (UASNs), this letter proposes a joint optimization framework called LCOA-ASPPM. By integrating Layered Cheetah Optimization Algorithm (LCOA) with Attention Optimized Support Vector Regression Power Prediction Model (ASPPM), the framework improves energy efficiency and communication reliability in UASNs. LCOA employs a bio-inspired multi-layer election strategy to optimize cluster head configurations, considering node energy levels, load balancing, spatial distribution, and energy efficiency. ASPPM power prediction system dynamically adjusts transmission power during network operation. It achieves this by applying incremental updates to a pre-trained SVR model, thereby adapting to varying network conditions. Through joint optimization, cluster head elections integrate power prediction results to assess node communication capabilities, while power allocation strategies adapt to cluster structure, thereby creating a mutually reinforcing optimization cycle.