A Hybrid Model for Botnet Detection using Machine Learning

Botnet attacks are becoming a growing threat to the security of computer networks, and there is a need for effective and efficient methods for detecting these attacks. In this study, we proposed a hybrid machine-learning model for detecting botnet attacks in network traffic. The approach combines three machine learning algorithms: k-means, rule- based system, and decision tree. The experiment was conducted using the CTU-13 dataset and features extracted from the Barnacles Mating Optimizer for network traffic flow. The experiment results showed that the proposed hybrid machine learning approach can achieve high accuracy levels in detecting botnet attacks, with a computed accuracy of 99%. The combination of different machine learning algorithms can improve the accuracy of the system, making it more robust to different types of botnet attacks. The use of real-world data, such as the CTU-13 dataset, and incorporating features from the Barnacles Mating Optimizer, make the results of this experiment more representative of real-world botnet attack detection computed accuracy 99.32 for kmean, 99.11% for decision tree and 97.14% for rule-based system. Precision for kmean 98.93%, decision tree 98.37% and rule-based 95.93% computed.