Detection of IoT Botnet Based on Deep Learning
Junyi Liu, Shiyue Liu, Sihua Zhang · 2019
In this paper, we propose a deep learning based approach for IoT botnet detection. We use the damped incremental statistics to extract basic traffic features of IoT devices and apply the Z-Score method to normalize the features. After that, the mangle area maps (TAM) based multivariate correlation analysis (MCA) algorithm is employed to generate dataset. Then we design a convolutional neural network (CNN) to learn the dataset and utilize the trained CNN to detect the traffic. The final experiments show that our approach can distinguish benign traffic and different kinds of attack traffic effectively and reaches the accuracy of 99.57%.