Machine Learning based Blockchain Networks for Combating Security Threats in IoTs

Abstract

The rapid increase in the usage of Internet of Things (IoT) technology and their deep involvement in every aspect of routine life made them a potential target for attackers. Attackers not only target the IoT networks but also go deep into the systems by using these attacks as a catalyst to sabotage the whole network and to disrupt the availability of the services. The main reason behind these sophisticated and modern attacks is that the IoT devices have less computational power and security, lying themselves open towards these attacks. Because of their less security features it is necessary to develop tools and techniques to detect the intrusions within the IoT network. In this thesis, we propose an intrusion detection system (IDS) to combat threats in the IoT network by integrating blockchain network with machine learning algorithms. For this purpose, the machine learning algorithm is trained on an actual data-set for intrusion detection within the system. The blockchain network is used to share the attackers information across various Autonomous Systems (AS). In addition, the spectral partitioning is used to divide the network into different autonomous systems and for identifying border nodes in each autonomous system. An IDS is deployed on these border nodes for traffic monitoring. The results have shown that this technique successfully identifies the threats and shares the attacker’s information through the blockchain network with precision and accuracy. We are hoping to replace the traditional intrusion detection techniques with our approach as it provides better integrity and an optimal way of diving the network to identify the nodes to place the IDS.