Enhance The Network Intrusion Detection System Classification Performance Using Three-Dimensional Virtualization

Abstract

Security has become an important part of an organization's information system. Network Intrusion Detection Systems (NIDS) are critical detection systems used as a countermeasure to protect data integrity and system availability from attacks, intruders. Detecting computer network intrusion attacks has become a difficult problem in solving network security. Traditional method error classification is a common problem with machine learning intrusion detection. Improvements in machine learning models are hampered by a lack of insight into the reasons behind this misclassification. This proposed three-dimensional virtualization framework using a Deep Q-Learning Neural Network (DQLNN) to analyze the malicious node, attacker path and traffic conjunction. This method categorizes the traffic based on network parameters such as traffic shaping, bandwidth, total rate, and rate bound. In this process, during the data transmission to verify the network path and request the user to identify the attackers. To introduce a Monte Carlo method to provide random samples to reduce the prediction time delay. The Markov chain model to identify traffic and probability transition states on the network. In this chain, the model constructs the traffic state tree to predict the higher traffic state and attacker states. This model proposed Xgboost classification to predict the traffic path and classify the attackers. After the attacker classified to construct the 3D visual representation to reflect the various attack and traffic. To use a benchmark intrusion detection dataset, which is KDDCup'99 and the accuracy of the classifiers was estimated using the k-fold cross-validation method. In this overall system performance to provide more security and server network detection compared to the existing method.