Improving intrusion detection system in the internet of things using a combination of convolutional neural network and cuckoo algorithm

The Internet of Things (IoT) refers to the connection of various devices to each other via the internet. Conceptually, the IoT can be defined as a dynamic, self-configuring network infrastructure based on standards and participatory communication protocols. The main goal of the IoT is to lead towards a better and safer community. However, one of the fundamental challenges in developing the IoT is the issue of security, and intrusion detection systems are one of the main methods to create security in the IoT. On the other hand, Convolutional Neural Network (CNN), with its specific features, is one of the best methods for analyzing network data. This network is a type of deep neural network composed of multiple layers that can ultimately reduce the dimensions of features. Additionally, the cuckoo algorithm has parameters required for configuration in the initial search, which are very few and can naturally and efficiently cope with multi-state problems. In this paper, a new method for intrusion detection in the IoT using CNN and feature selection by the cuckoo algorithm is presented. Simulation results indicate the satisfactory performance of the proposed method.