Abstract
WSNs have increasingly become part and parcel of the current and emerging communication systems, hence the need for solid measures to protect this critical infrastructure from evolving cyber threats. This research investigates the development of a Supervised Machine Learning-based IDS using MLP, SVM, and LR to increase the detection of network attacks for prevention purposes. A new method is proposed that entails the integration of an inter-dataset evaluation strategy that harmonizes two heterogeneous data structures and simultaneously aims to protect models and sensitive data. The conceived framework assesses the accuracy of a variety of machine learning metrics, such as accuracy, precision, recall, and the F1 score. As we can see from the results, detection efficiency was significantly improved with the incorporation of inter-dataset routing. MLP achieved a maximum accuracy of 99.1% with a recall of 100% in cross-dataset testing, showcasing its robustness in identifying all positive instances. SVM demonstrated a precision of 99.4% in certain scenarios, effectively minimizing false positives and enhancing classification confidence. Logistic Regression also showed stable precision values, contributing to consistent detection performance. These analyses stress how existing techniques for machine learning algorithms used in IDS designs cover a wide variety of purposive uses, with one being the prevention of distributed denial-of-service attacks. The practical implications of this research include the recommendation of the usage of proper algorithms justified for the desired security level in the networks to the administrators of the networks and security entailment specialists. The work done consolidating the result and outcome of this study points a way forward for subsequent studies on WSN security, with the view of providing a springboard for the development of intrusion detection systems in view of the dynamic nature of cybersecurity.
Keywords
Cybersecurity Threats
Intrusion Detection Systems
Network Attack Prevention
Support vector machine
Wireless Sensor Networks