Improving the accuracy of detection botnet attacks in Internet of Things network by using MLP neural network
Subject Areas : ICT
1 -
2 -
Keywords:
Abstract :
Due to the increasing use of the Internet of Things around the world and the exponential increase in the number of devices connected to the network and the communication between them, the potential for security problems is increasing. Considering that many personal and public devices are connected to this network, any security problem can have unpredictable and significant consequences. Internet of Things applications include smart cities, smart transportation, responsive environments, and some other specific things that are directly controlled by users or digital devices, cyber-attacks through the Internet of Things and smart digital devices is the most important threat for these networks. So far, numerous researches have been conducted to detect Internet of Things attacks, in particular botnet attacks, as one of the most important attacks in this field. But the lack of a method that uses machine learning methods with high accuracy and low error to detect these attacks is strongly felt. In this research, by using the N-BaIoT dataset and Python simulator for modeling and also using deep learning methods and MLP neural network to evaluate and train the data (using the objective function and training), the neural system was used for detecting botnet attacks. This method obtained accuracy 90.35, precision 85.99, recall 90.53 and f1-score 87.50. Compared to other machine learning methods including random forest algorithm (RF), support vector machine algorithm (svm), K nearest neighbor algorithm (knn), XGBOOST algorithm, AdaBoost algorithm, the best result was obtained in all 4 evaluation parameters.