نوع مقاله : کامپیوتر - داده کاوی
نویسنده
استادیار،دانشگاه گلستان، گرگان، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسنده [English]
With the increasing progress of Internet of Things technology and the need to establish secure communication between devices connected to each other, the security of exchanged data has become a critical issue. Attackers and cyber hackers always try to create problems in Internet of Things networks by infiltrating systems and manipulating them. Therefore, a lot of research has been done by many researchers to deal with these threats. The accuracy and execution time of the existing methods are not optimal, and the need for a combined method that, in addition to increasing the accuracy, also reduces the required time seems essential. In this paper, a hybrid method using the dragonfly algorithm and gradient boosting is presented to improve the accuracy of the Internet of Things system. The dragonfly algorithm leads to the elimination of ineffective features by improving new solutions and increasing population diversity, which prepares the data to achieve high accuracy in classification. So that those features that lead to a decrease in classification accuracy are left out. After that, by using gradient boosting, which is a factor with high speed and accuracy for detection (classification), the data classification operation is performed, which is the main phase of attack detection. The simulation results show that the classification accuracy in binary and multi-class mode is equal to 99.994% and 99.992%, respectively, which indicates the superiority of the proposed method over other previous methods.
کلیدواژهها [English]
)