Machine Learning for Enhancement of Security in Internet of Things Based Applications

Abstract

Internet of Things (IoT) is a new paradigm in which billions of smart devices can communicate with each other and exchange information with zero to minimal human intervention. From the pace at which technological advancement is happening in the field of IoT, it is estimated that by the end of 2020, IoT will be seen in every major real-time application, including healthcare, Intelligent Transportation Systems (ITS), drones, agriculture, smart homes, etc. However, factors like attempts to minimize the cost of IoT technologies, and also the integration and collaboration of various heterogeneous components from almost all the multidisciplinary fields, has led to the introduction of security challenges, and vulnerabilities which need to be handled and addressed, as they are not, it could lead to serious damage such as information and privacy leakages, traffic congestions, and even accidents claiming the lives of people. There exist various security solutions that have already proven their worth in other networks, like the use of encryption, authentication, cryptography, access controls. However, considering the inherent and peculiar features of IoT, their deployment in IoT are believed to be ineffective. "Machine learning" and "deep learning" are the two terms that have drawn considerable attention from researchers over the last couple of years because of their vast usability in almost every area of research. Machine learning- and deep learning-based solutions can considerably enhance the security of IoT network and its associated devices. So, in this chapter, recent advances in the field of machine learning and deep learning for enhancing the security of the IoT network is thoroughly examined. Various possible security threats that exist in IoT based network along with the corresponding existing security solutions for those threats provided by different researchers are presented. To provide the potential future research directions in the use of the machine and deep learning in security enhancement of IoT based applications, all the possible opportunities, possible modifications, advantages, and shortcoming for every existing solution has been enumerated and are presented.