Intrusion detection system based on hybridizing a modified binary grey wolf optimization and particle swarm optimization

Nowadays, the world is increasingly becoming more connected and dependent on the Internet and Internet-based services. One of the main challenges of interconnectedness is the security of applications and networks from malicious actors. The security challenge is further compounded by the exponential growth of threats and the increase in attack vectors through interfaces of many newly introduced network services. To deal with the security threats, many solutions have been proposed; yet the existing solutions overwhelmingly fail to detect security threats efficiently with high performance. Accordingly, a hybridization of modified binary Grey Wolf Optimization and Particle Swarm Optimization is proposed in this article. The proposed solution uses two benchmarking datasets, NSL KDD'99 and UNSW-NB15, and the results reveal that the proposed solution outperforms the existing solutions, as the proposed approach improves the detection accuracy by approximately 0.3% to 12%, and the detection rate by 2% to 12%. In addition, it reduces false alarm rates by 4% to 43%, and reduces the number of features by approximately 31% to 75%. Last, the proposed approach reduces processing time by approximately 14% to 22% compared to state-of-that-art approaches.

Automated summary

The proposed hybrid optimization approach outperforms existing solutions in terms of detection accuracy, detection rate, false alarm rate reduction, feature reduction, and processing time.