Machine learning–based detection in wireless sensor networks
Mansour Lmkaiti, Houda Moudni, Hicham Mouncif · 2025
Wireless sensor networks (WSNs) have emerged as vital tools for monitoring environmental and physical conditions across various industries. However, their efficiency is counterbalanced by inherent security challenges, particularly concerning data integrity and transmission. Denial of service (DoS) attacks pose a significant threat to the reliability of WSNs. Intrusion detection systems (IDSs) employing machine learning (ML) techniques offer promising solutions to address these challenges. This chapter conducts a comprehensive review of relevant literature, emphasizing the pressing need for robust intrusion detection systems in resource-constrained WSNs. Leveraging advancements in academic performance prediction models, our study proposes an ML-based approach tailored to detect specific assaults in WSNs, trained on the IDSAI dataset. Our contributions aim to bolster WSNs against evolving security threats, ensuring the integrity and confidentiality of critical data. Furthermore, we explore various ML algorithms, including the random forest classifier, decision tree classifier, extra trees classifier, gradient boosting classifier, and XGB classifier, evaluating their performance to identify the most suitable algorithm for WSN security. Through meticulous analysis and experimentation, we highlight the gradient boosting classifier&s;s superior accuracy despite longer training times, offering insights into optimizing WSN security systems.