Firewall engine based on Graphics Processing Unit

Abstract

Firewalls are very important elements in network security. Working of firewall rules for enterprise network has become complex, error-prone and time-consuming. Firewall filtering rules have to be carefully written and organized in order to correctly implement the security policy. The main issue is the slow filtering action during heavy load. To reduce the time consumption there is a very urgent need of optimized firewall engine, which runs on GPU's. We mainly focus on the creating parallel algorithms for desktop firewall which reduces the time consumption and at the same time it can allow for strong threat detection, intrusion detection of incoming packets. In our paper we have created parallel optimized algorithms for intrusion detection, threat detection, packet filtering and network address translation which runs on NVIDIA's GPU card and it is based on CUDA programming. For our experimental analysis, we have created test packets and for virus scanning we have used the virus-signatures from Clam-AV.