نوع مقاله : مقاله پژوهشی
نویسندگان
1 دانشجوی دکتری، گروه مهندسی کامپیوتر، دانشکده کامپیوتر، دانشگاه آزاد اسلامی،ارومیه، ایران
2 استادیار،گروه مهندسی کامپیوتر، دانشکده کامپیوتر، دانشگاه آزاد اسلامی،ارومیه، ایران
3 دانشیار،گروه مهندسی کامپیوتر، دانشکده کامپیوتر، دانشگاه آزاد اسلامی،ارومیه، ایران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
The increase in the number of Internet of Things devices, high speed, and large volume of generated data has led to network security issues and the identification of cyber attacks in these networks becoming one of the key challenges in this area. Intrusion detection systems have been proposed as a solution to tackle this problem. Proper selection of features in creating intrusion detection models can significantly increase detection accuracy. In this article, a binary algorithm and an improved chimpanzee algorithm have been designed for feature selection. The chimpanzee algorithm is designed for solving continuous problems and cannot be efficient in solving binary problems. It also suffers from local optima and slow exploration, exploitation, and convergence in this algorithm. Therefore, changes need to be made in this algorithm to solve binary problems. Hence، in this article an improved version of the chimpanzee algorithm for discrete problems and feature selection has been designed and implemented for intrusion detection and network-based attacks in Internet of Things networks. The proposed method reduces features by an average of 60 percent and successfully detects attacks with ac curacies of 99.3%, 99.6%, and 99.9% in the Ton-IoT، UNSW-NB15، and IoTID20 datasets, significantly reducing the false alarm rate of detected attacks. Statistical analysis using the Kruskal-Wallis test showed that the proposed method converges faster compared to the comparison methods.
کلیدواژه‌ها [English]
)