Deep Learning Model for Security of IoT Network

Abstract

The Internet of Things (IoT) is the emerging and rapidly rising network of phys-ical objects that are provided IP addresses for network connectivity and having the ability of transferring data between objects and other Internet-based devices and systems. There are billions of IoT devices connected and there is a high cyber security and data privacy risk. Com-puters and mobile devices have many software and security solutions to secure and defend from attacks, but a similar type of security solution is missing to secure IoT networks. In this paper, the One-Dimensional Convolution Neural Network (1DCNN) is proposed to measure efficiency using UNSW-NB15 dataset which is the latest and covered modern attacks data in comparison to NSL KDD and KDDCUP99. For comparison study of performance, we have compared attic Machine learning models with KNN and Naïve Bayes. In each experiment, the model ran up to 200 epochs and with 0.001 learning rate. Deep learning models have outper-formed in comparison to the attic machine learning model.