Intrusion Detection System at Node Level using Swarm Optimization Algorithm

The initial implementation of an anomaly detection model was for one-dimensional data collecting, and it was based on statistical models. When the data’s probability distribution is utilized to build a statistical normal model, and that model is then used to evaluate the data, the models meet the fundamental criteria of representing the reference model. As a subset of intrusion detection, anomaly detection involves keeping tabs on a system’s activity and then labelling it as either “normal” or “anomalous,” depending on whether or not it conforms to predetermined norms. Classifiers that employ thresholds to determine what constitutes abnormal behavior have proven effective. Threshold-based classification is a pruning strategy for reducing the size of a decision tree (DT) to minimize the classifier complexity and boost its predictive performance. In this study, Particle Swarm Optimization (PSO) is applied t o address the issues of network intrusion detection in an effort to prune a Threshold-based categorization. The suggested method is a hybrid strategy that employs PSO for nodes pruning and the Threshold method for classifying network intrusions. In the perspective of network intrusion detection, this dataset has been frequently utilised as a reference standard. In comparison to other existing classifiers, our proposed method achieve well with intrusion detection rate, cache hit rate, cache miss rate and Data Delivery Rate.