Dynamic Network Traffic Analysis for Cyberattack Detection Using Multiple Deep Learning Techniques

Network security is critical in the development of applications related to the rapid growth of computer networks. With the ongoing advancement and deployment of various cutting-edge technologies, there is a rising security vulnerability in networks, significantly harming the application’s services. Identifying vulnerabilities in real-time to secure the network and data is critical since the attacks could quickly take advantage of them and sever connections from other networks. To offer proper security, networks require an effective Intrusion Detection System (IDS) designed with Artificial Intelligence algorithms. It continuously monitors and alerts network users when security violations are detected dynamically. Most IDS uses advanced AI models to predict malicious activities in cyber networks. These IDS uses several administrators like Security Information and Event Management (SIEM), Hybrid Intrusion Detection System (HIDS), and Network Intrusion Detection System (NIDS), which collect the data from different sources and use them to filter the typical connections from the abnormal ones. The paper’s main objective is to develop an efficient IDS using a deep learning model for analyzing and predicting the abnormalities in network traffic data. The proposed Dynamic Network Traffic Analyzer (DNTA) integrates three different deep learning models: a Deep Convex Neural Network (DCNN), an Artificial Neural Network (ANN) algorithm, and an XGBoost regressor. This IDS analyzes dynamic traffic data to predict malicious connections or traffic to the network. The proposed DNTA model is evaluated with the NSL-KDD dataset, and the results are discussed in detail and compared with similar ones.