Anomaly Detection in Wireless Sensor Networks Using Hybrid Machine Learning Models

Abstract

Nowadays, WSNs, or Wireless sensor networks, have become one of the most widely used wireless technologies for the purpose of sensor communication. WSNs are often designed for specialized applications, involving tracking or monitoring, in indoor or outdoor settings where battery capacity is a major concern. Over the past few years, numerous routing schemes have been developed to resolve these challenges. Currently, researchers use a variety of machine learning (ML) approaches to identify anomalies in WSN. The study presents an Anomaly Detection in Wireless Sensor Networks Using Hybrid Machine Learning Model (ADWSN-HMLM). The ADWSN-HMLM approach undergoes data preprocessing, feature extraction, detection, and classification. The benchmark IDS dataset is used to test the experimental results of the ADWSN-HMLM approach. The simulation outcomes showed that the ADWSN-HMLM approach outperformed other approaches. The infrastructure of WSN and the security issues they encounter are conveniently referenced in this study. Along with discussing the difficulties and suggested solutions for enhancing sensors' capacity to recognize threats, attacks, risks, and malicious nodes through their capacity to learn and self-develop using ML techniques, this study also explores the potential benefits of ML approaches in lowering the security costs of WSN across a number of domains. In addition to 98% for regular traffic, the detection accuracy for scheduling, grayhole, flooding, and blackhole assaults is 97.59%, 96.95%, 96.03%, and 97.05%, respectively. These findings demonstrate that the ADWSN-HMLM methodology can offer the WSN effective anomaly recognition.