Fast Parallel Network Packet Filter System based on CUDA

Abstract

In recent years, with the rapid development of the network hardware and software, the network speed is enhanced to multi-gigabit. Network packet filtering is an important strategy of network security to avoid malicious attacks, and it is a computation-consuming application. Therefore, we develop two efficient GPGPU-based parallel packet classification approaches to filter packets by leveraging thousands of threads. The experiment results demonstrate that the computational efficiency of filtering packet can be significantly enhanced by using GPGPU.