Improving performance of intrusion detection system using OpenCL based general-purpose computing on Graphic Processing Unit (GPGPU)

With the increased connectivity of computer devices including mobile devices to the Internet, the volume as well as the type of information that moved through organization network perimeter are also growing significantly. Consequently, this has heightened information security threats to organizations that rely on the Internet to do their businesses. On the other hand, Intrusion Detection System (IDS) being used today to detect these threats is limited by a single threaded application, limiting its capability to cope with the growing network traffic. This research introduced a parallel intrusion detection system based on GPGPU and Open Computing Language (OpenCL), that processes incoming packets concurrently, thus increasing system capacity in a high throughput environment. The experiment result shows that performance of GPU-based design is improved on the average of 58.73x against a single-thread central processing unit (CPU) implementation, and on the average of 13.53x against a multi-thread implementation. At the same time, GPGPU-based system has a performance gain of 29.92 compared with CPU-based system implementation.