Traditional intrusion detection systems are insufficient to identify recent and modern sophisticated attempts with unpredictable patterns. The ability to reliably detect modern cyberattacks is vital. Current machine learning-based intrusion detection methods in the field of information technology cannot keep up with the exponential growth of network data and features. For the optimum and decreasing selection of high-dimensional incursion characteristics, deep convolutional neural networks (DCNN) can be an efficient approach. Traditional convolutional neural networks (CNN) are still limited to several parameters and are susceptible to local optimality. In this paper, we propose a DCNN model to detect attacks and test it on a Software Defined Network (SDN) environment. We use the InSDN dataset, specifically developed for the SDN environment. Additionally, the model has been trained using the CIC-IDS2017 and the CIC-DDoS2019 datasets to demonstrate the applicability of the model. Our model outperforms most of the recent attack detection methods and could achieve a 99.99% accuracy rate with only a 0.0016 loss rate.
International Journal of Information Technology
Springer
BJIT-D-22-03013
19-December-2022
26-February-2023
30-May-2023
10.1007/s41870-023-01332-5
https://link.springer.com/article/10.1007/s41870-023-01332-5
If you find this code and paper useful, kindly consider to cite from your valuable work.
Hnamte, V., Hussain, J. An efficient DDoS attack detection mechanism in SDN environment. Int. j. inf. tecnol. (2023). https://doi.org/10.1007/s41870-023-01332-5