GPU-based Multi-stream Analyzer on Application Layer for Service-oriented Router
Nishi Laboratory, Graduate School of Science and Technology, Keio University, Japan
The 2013 International Conference on Parallel and Distributed, Processing Techniques and Applications (PDPTA’13), 2013
@inproceedings{ikeuchi2013gpu,
title={GPU-based Multi-stream Analyzer on Application Layer for Service-Oriented Router},
author={Ikeuchi, Kazumasa and Wijekoon, Janaka and Ishida, Shinichi and Nishi, Hiroaki},
booktitle={Embedded Multicore Socs (MCSoC), 2013 IEEE 7th International Symposium on},
pages={171–176},
year={2013},
organization={IEEE}
}
Service-oriented router (SoR) is a new router architecture for providing rich services to Internet users by utilizing useful information extracted from network traffic. In SoR, stream reconstruction and selection is a fundamental process for providing the services in the application layer. After real-time reconstruction of stream data, SoR used a software character string analyzer to extract important required information. One of the promised services is a router-level network intrusion detection system. Because a network consists of hundreds of thousands of data streams, achieving an intended throughput while analyzing these stream data is a critical problem. We propose an acceleration method of string matching based on a heterogeneous system consisting of a CPU and a graphics processing unit. In addition, we designed and implemented a task controller that improves the distribution of POSIX-thread-based processes so that string matching can be performed concurrently depending on the status of the string matching system.
December 4, 2013 by hgpu