A Novel Adaptive AI-Based Framework for Node Scheduling Algorithm Selection in Safety-Critical Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are vital to a wide range of applications, spanning from environmental monitoring to safety-critical systems. Ensuring dependable operation in these networks critically depends on selecting an optimal node scheduling algorithm; however, this remains a major challenge since no single approach performs best under all conditions. To address this issue, this paper proposes an AI-driven framework that evaluates scenario-specific functional requirements—such as coverage, connectivity, and network lifetime—to identify the optimal node scheduling algorithm from a pool that includes Hidden Markov Models (HMMs), BAT, Bird Flocking, Self-Organizing Maps (SOFMs), and Long Short-Term Memory (LSTM) networks. The framework was evaluated using a neural network trained on simulated data and tested across five real-world scenarios: healthcare monitoring, military operations, industrial IoT, forest fire detection, and disaster recovery. The results clearly demonstrate the effectiveness of the proposed framework in identifying the most suitable algorithm for each scenario. Notably, the LSTM algorithm frequently achieved near-optimal performance, excelling in critical objectives such as network lifetime, connectivity, and coverage. The framework also revealed the complementary strengths of other algorithms—HMM proved superior for maintaining connectivity, while Bird Flocking excelled in extending network lifetime. Consequently, this work validates that a scenario-aware selection strategy is essential for maximizing WSN dependability, as it leverages the unique advantages of diverse algorithms.