Abstract

In the context of the Internet of Things (IoT) and edge computing, system robustness is paramount. This study introduces a refined Convolutional Neural Network (CNN) model enhanced with a black window optimization (BWO) algorithm for feature selection and an Enhanced Gazelle Optimization (EGO) algorithm for hyperparameter tuning. These techniques were specifically selected to address the need for improved fault tolerance and redundancy in complex network environments. Our methodology encompasses an extensive preprocessing stage to clean and normalize a diverse dataset that is representative of typical IoT/edge computing scenarios. The BWO algorithm was pivotal in pinpointing critical features, and the EGO algorithm was adept at fine-tuning the settings of the model to ensure optimized performance. The CNN architecture was carefully structured to process high-dimensional data, which is essential for identifying patterns that signal faults or necessary redundancies in network operations. The results of this study were significant. Post-optimization, the CNN model demonstrated a remarkable improvement in all performance metrics, with the accuracy increasing to an impressive 99.98%. The macro-and weighted average precision, recall, and F1-scores similarly reflected high efficiency, with nearly perfect scores. In the simulated fault scenarios, the model exhibited robustness with a rapid Mean Time to Detect (MTTD) and Mean Time to Recover (MTTR), highlighting its capability to swiftly identify and react to network anomalies. The findings of this study underscore the practical utility of bio-inspired algorithms for enhancing deep-learning models for real-world applications. The optimized CNN model shows not only a theoretical advancement in fault detection and system redundancy but also proves to be a pragmatic solution for real-time IoT/edge computing systems. This research contributes to the body of knowledge by demonstrating the effectiveness of combining CNN architectures with advanced optimization techniques, paving the way for more resilient and autonomous computing infrastructure in the IoT ecosystem.