Efficient Parallel Muti-pattern Matching Using GPGPU Acceleration for Packet Filtering

Abstract

In the past decades, the Internet usage has increased dramatically. For the network security, the network packet filtering is an important strategy to identify malicious network packets. However, malicious attacks spread much faster than network administrators can respond. The software-only implementations of filter are unlikely to meet the performance goals. Therefore, we develop a novel GPGPU-based parallel packet classification approach by adopting bloom filter to inspect the packet payload by leveraging the computation power of GPGPU. The experiment results present that the proposed algorithm can be significantly enhanced the performance of filtering packets. According to the experimental results, the proposed method can achieve over 5.4 times speed up over the sequential bloom filter on single CPU.