LSTM: A Deep Learning Based Approach for the Classification of Intrusions in IoT Based Networks

  • P.S.Nandhini, S.Malliga, N.Monisha Sri, K.Jeevitha, C.Sakthivel

Abstract

Internet of Things (IoT) is a collection of heterogeneous objects or devices. The devices in IoT are energy constrained and are connected through the Internet for collecting and sharing the data. This leads to severe processing in the devices or the nodes. Due to limited processing resources, the standard security mechanisms could not be implemented in the devices. This leads to various kinds of attacks such as Denial of Service, User to root, Remote to Local and probe attacks. In this paper, LSTM a deep learning approach is used for the classification of different kinds of attacks that occur in the IoT networks.  LSTM is a type of recurrent neural network in which the information flows through a mechanism called as cell states. It can forget or remember the things selectively. The parameters such as accuracy, recall, precision and F1 Measure are calculated for the proposed algorithm. The above parameters are calculated for Artificial Neural Networks and are compared.

Published
2020-05-01
How to Cite
P.S.Nandhini, S.Malliga, N.Monisha Sri, K.Jeevitha, C.Sakthivel. (2020). LSTM: A Deep Learning Based Approach for the Classification of Intrusions in IoT Based Networks. International Journal of Advanced Science and Technology, 29(9s), 26 - 33. Retrieved from https://sersc.org/journals/index.php/IJAST/article/view/12998