Machine Learning Assisted Intrusion Detection System against Slow Rate Http/2 Dos Attacks

Abstract

The recently defined HTTP/2 protocol has additional benefits over HTTP/1.1 and is intended to make effective use of
TCP's transmission rate. Its threat pathways are still not fully recognized, though. Research on security flaws or
vulnerabilities is extremely limited. It's critical to comprehend potential risks. Slow rate DoS (Denial of Service)
attacks are among the most successful. In this research, we use machine learning to develop an intrusion detection
system against slow rate HTTP/2 DoS assaults. We use one class SVM method for classification of attacks. Proposed
OIDS system has two modules in which first one is feature extraction module and one-class SVM classification
module. We contrasted our suggested model with state-of-the-art models including Random Forest, ANN, KNN,
Decision Tree, and Naïve Bayes. Measures including as accuracy, sensitivity, specificity, precision, TPR, FPR, and
detection rate are used to assess these models. With excellent precision, the suggested model (OIDS) performed
better.