Detecting Grid-Abuse Attacks by Source-based Monitoring

While it provides the unprecedented processing power to solve many large scale computational problems, GRID, if abused, has the potential to easily be used to launch (for instance, denial-of-service) attacks to other computer systems. This paper introduces a methodology to prevent GRID infrastructure from succumbing to such activities. Our system is designed to detect, classify, and control GRID-abuse attacks. The current implementation of our system uses system calls as raw features, while the chi-square statistical method is used for attack detection, and fisher linear discriminant analysis is used for attack classification. The performance evaluation shows our system can achieve a high detection rate (over 92%) with very low false alarm rate (virtually zero), and small detection delay (about 10 seconds). For the detected attacks, our system classifies them with average accuracy over 82%.

• Type: Conference
• Year: 2004
• Where: ISCAPDCS
• Authors: Jianjia Wu, Dan Cheng, Wei Zhao
• Keywords: Control Grid-abuse Attacks, ISCAPDCS 2004, ISCAPDCS 2007, Scale Computational Problems, Unprecedented Processing Power