Deep Packet Inspection With CUDA Technology

With the development of network technology, people’s lives more and more can not be separated from the network, when the network technology to bring us convenience at the same time, it contains the security risks can not be ignored. Deep packet inspection as a traffic management, security and network analysis and other aspects in the more mature technology is widely used in the big network operators, Internet service providers and the efficiency of the core algorithm has become a research focus. And now GPU high performance parallel computing ability is more and more popular, the use of CUDA technology to achieve parallel improvement algorithm has become the preferred program to improve the speed of implementation.In this paper, we will study and implement an efficient DPI algorithm from two aspects. One was design and improve the pattern matching algorithm based on SDCH dictionary compression, this is the basis for the realization of SDCH deep packet inspection, using CUDA technology to improve the parallel of the algorithm can greatly improve the efficiency of the algorithm, so that it can be better to face the actual problem; Two was design the scheme of Glushkov automata bit parallel method which is a regular expression matching algorithm for highly parallel improvement, to improve the efficiency of regular expression matching algorithm, so as to speed up the speed of DPI algorithm based on regular expression matching.Finally verify the correctness of the two algorithms, and compared the improved algorithm on the GPU and the original algorithm on the CPU in execution efficiency differences, which further proves the improved results by using CUDA, and DPI algorithm in the face of the actual complexity of the work to provide a strong guarantee of efficiency.