PROJECT TITLE :

CNN-LSTM: Hybrid Deep Neural Network for Network Intrusion Detection System

ABSTRACT:

The importance of network security to our day-to-day interactions and networks cannot be overstated. The importance of having an efficient intrusion detection system has become paramount due to the fact that attackers are continually developing new methods of attack and the size of networks is continually expanding. Developing an efficient IDS required the implementation of machine learning algorithms in a number of studies; however, with the advent of deep learning algorithms and artificial neural networks that can generate features automatically without the need for human intervention, researchers began to rely on deep learning. During the course of our study, we created a hybrid model of an intrusion detection system by combining the capabilities of two different types of neural networks: the Convolutional Neural Network and the Long Short-Term Memory Network. Both of these networks are able to pull out spatial and temporal features, respectively. In order to improve the performance of the model, we included both a batch normalization layer and a dropout layer. The model was trained with the CIC-IDS 2017 dataset, the UNSW-NB15 dataset, and the WSN-DS dataset. The training was based on the binary and multiclass classifications. The efficiency of the system is determined by the confusion matrix, which takes into account a variety of criteria for assessment, including accuracy, precision, detection rate, F1-score, and false alarm rate (FAR). Experiment results that showed a high detection rate, high accuracy, and a relatively low FAR demonstrated the usefulness of the model that was proposed.