Using Deep Learning Technology to Optimize VPN Networks Based on Security Performance

Abstract

In recent years, network developments and user growth have increased network security problems and techniques. The trend in network security is towards web-based networks, given Internet users' diverse origins, unpredictable persons are more likely to participate in malevolent activities. Security and privacy safeguards are implemented using many technologies. This paper proposes using a virtual private network (VPN) to secure particular communications across vast networks. VPNs restrict unauthorised connections, benefiting secured hosts. Through a VPN network, connections can be kept hidden and external connections prohibited. The influence of a virtual private network (VPN) on a standard network's performance is studied by producing and assessing CBR, HTTP, and FTP payloads. The evaluation used throughput and time delay as performance measures after analysis of the finding, deep learning (DL) can predict attacks. Because that learns attack patterns during training to effectively forecast attacks. To detect attacks, deep learning-based attack prevention model was created This method uses Nave Bayes and FFNN to enhace network performance. The results show that VPNs affect packet latency and performance differently depending on the data type. The FFNN algorithm detects intrusions with 98% accuracy.