Two-Stage Fault Management Framework for IoT-based Wireless Sensor Network through Resilient and Energy-Efficient Approach

The Internet of Things (IoT) is built on the foundation of Wireless Sensor Networks (WSNs), which is the technology that provides a smooth connection and real-time data transmission in smart cities and industrial settings. Nevertheless, IoT-based WSNs are prone to several challenges due to hardware, software, and time-related faults, which affect stability and reliability. In the research, a two-step fault management system is proposed to improve fault detection, classification, and resilience in IoT-based WSNs. At the initial step, the identification of faulty nodes in heterogeneous environments is done with a Siamese-Residual Attention Convolutional Neural Network - Bidirectional Long Short-Term Memory (SRA-CNN-BiLSTM) model. Next, the second step used a Self-Learning Residual Shrinkage Network (SLRSN) to categorize the types of faults effectively. The optimization of the model parameters is performed using a new Clan Cooperative Crested Porcupine-based Optimizer (3CPO) that guarantees high accuracy and convergence rate. Python simulations indicate a 0.98% increase in the accuracy of fault detection and significant energy savings. The framework is therefore a reliable and scalable system of fault management of IoT-enabled WSNs.