Improving the Routing Security in Wireless Sensor Networks using Neutrosophic Set and Machine Learning Models

Numerous methods have been put forth to identify and safeguard routing data because Wireless Sensor Networks (WSNs) are susceptible to attacks during data transfer. To create an artificial intelligence-based attack detection system for WSNs, we provide a unique stochastic predictive machine learning technique in this research that is intended to identify unreliable events and untrustworthy routing properties. Our approach makes use of real-time feature analysis of simulated WSN routing data. We create a strong foundation for categorization. Our approach's primary benefit is the development of an effective machine learning (ML) technique that can analyze and filter WSN traffic to stop dangerous and suspicious data, lessen the significant variation in the routing information gathered, and quickly identify assaults before they happen. We use the XGBoost and Random Forest (RF) models with different parameters. Then the bipolar neutrosophic set is used to deal with uncertainty and vague information. The neutrosophic set is used to rank the ML models and select the best one.