A Novel Approach for Data Collection and Network Attack Warning

Abstract

Network security in general, research on detecting and finding attacks in computer networks in particular, has become a very hot topic. There are a variety of studies on machine learning models to attempt to detect network attacks, but these studies only focused on the models for prediction while the details of collecting data and the steps of processing and extracting information from network packets are not presented. In this research, we have employed and installed an active framework for collecting data using Honeynet and leveraging artificial intelligence algorithms, such as machine learning and deep learning, to detect_attacks in computer networks. We have proposed to use only header information of the network packets for network traffic classification. Our results from the experiments prove that the framework of collecting network packets and detecting attacks in computer networks can be implemented and employed efficiently in practical cases. In addition, DARPA29F extracted from the proposed method with 29 features is a promising dataset to validate the learning algorithms.