Edge Computing Enabled Anomaly Detection in IoT Environments Using Federated Learning

This research explores the integration of edge computing and unified learning procedures for peculiarity locations in Internet of Things (IoT) situations. Four inconsistency discovery calculations - Isolation Forest, Local Outlier Factor, One-Class SVM, and Recurrent Neural Organize - are assessed over assorted IoT datasets to survey their adequacy. Results show that the Repetitive Neural Network outperforms other calculations with exactness, review, F1-score, and AUC-ROC values of 0.90, 0.92, 0.91, and 0.95 individually. The study illustrates the possibility of leveraging edge computing and unified learning for productive and privacy-preserving inconsistency locations at the organised edge. By passing on computational assignments to edge contraptions and collaboratively planning models utilizing combined learning, our approach finishes flexible and adaptable irregularity revelation in IoT systems. The investigate contributes encounters into tending to challenges such as inertness, security, and flexibility characteristics in centralized idiosyncrasy disclosure approaches, clearing the way for more capable and privacy-preserving courses of action in IoT applications.