IoT Intrusion Detection System Based on LSTM Model

Abstract

Aiming at the problems of time-consuming feature extraction and general efficiency in the detection of en-crypted traffic by traditional machine learning algorithms, an intrusion detection model based on deep learning long short-term memory network (LSTM) was proposed. First, the malicious encrypted traffic in the CTU-13 data set and the normal traffic in the CICIDS-2017 data set are extracted to form a data set; then the binary classification data set processing is completed based on the secure transport layer protocol; finally, the LSTM and one-dimensional convolutional neural networks are trained. Network, two-dimensional convolutional neural network and convolutional neural network-long short-term memory network four deep learning models. The experimental results show that LSTM has significant advantages over the other three models in five evaluation parameters, the accuracy of key parameters is as high as 99.84%, and it performs well in terms of CPU and memory usage, which meets the security requirements of the Internet of Things.