Research And Implementation Of Network Intrusion Detection Acceleration In Heterogeneous Environment

With the rapid development of science and computing, big data analytics, computer vision, and other areas. Computer performance requirements also rapid grow. And in recent years, the growth of computer CPU performance slowed down, enhance semiconductor technology bottlenecks encountered, power consumption is also growing. In order to meet the growing demand for these areas, the research of using heterogeneous processors to accelerate computing has become increasingly active. GPU manufacturers are constantly exploring and optimizing GPU architecture so that GPU can be more effective for general purpose computing. Related software development framework, tools are constantly improving.In this thesis, the research and implementation of network intrusion detection acceleration in heterogeneous environment, focusing on research of memory model, task scheduling model and multi-pattern string matching algorithm speed optimization methods in heterogeneous environment. Experimental platform for network intrusion detection system was designed and implemented, introduced and implemented two multi-pattern string matching algorithm speed optimization methods in heterogeneous environments.The main content is divided into four parts:1) To study the technical details Open CL framework, research the forefront of heterogeneous processor architecture, study the current memory model and task scheduling model in heterogeneous environment, compared the difference to traditional homogeneous environment. Find and outlook the direction of heterogeneous programming development. Choose three different heterogeneous processors, including AMD's A10 5800 K APU, Freescale IMX6 embedded chip and Intel I7-3770 k. Test compare the performance of the three different processer. 2) Specific the network intrusion detection system as an application to the research. Explore the details of the open source network intrusion detection system, Sonrt. 3) Investigate and survey the research status of optimization methods for the multi pattern string matching algorithm. Design and implementation three newalgorithms. 4) Design and implement an experimental platform for network intrusion detection system, implement and test the optimized multi-pattern string matching algorithm that I introduced.Experiments show that, in heterogeneous environment, the matching speed of the multi-pattern string matching algorithm which I introduced has been improved significantly, can effectively reduce the current hardware cost and power consumption of network intrusion detection system.