Enabling security for the Industrial Internet of Things using deep learning, blockchain, and coalitions

AbstractAbstractIn a wireless Industrial Internet of Things (IIoT) network, enforcing security is a challenge due to the large number of devices forming the network and their limited computation capabilities. Furthermore, different security attacks require specifically tailored security protocols to prevent their occurrence. As an alternative to these conventional centralized security protocols, the application of Blockchain (BC) and Deep learning (DL) for securing IIoT networks hold great potential. BC facilitates security by being an immutable record of the changes happening in a network. Coalition Formation theory aids decentralization and promotes energy efficiency. And to enforce a state‐of‐the‐art attack detection technique, Deep learning provides an adaptive and reliable platform. Thus, in this paper, a security framework that facilitates generalized security for the IIoT network using BC and Coalition Formation theory is proposed. Additionally, we promote a sophisticated deep learning‐based classification algorithm to efficiently classify malicious and benign devices in IIoT scenarios. In the proposed model, connection links can only be established if the details of the connection are mined on the BC by the “sender” device. Therefore, we propose a Proof of Reliance algorithm that dynamically increases the computational difficulty to prevent malicious devices from attacking the network. Through simulations, it is experimentally proven that malicious devices can never attack the network when the proposed framework is employed for IIoT security.In this paper, a framework that enables generalized security for the IIoT network using interest‐based and physical‐aware coalitions, and Blockchain has been proposed. A deep learning‐based technique for malicious IIoT device identification as a sophisticated attack detection technique has also been proposed. The ideas involving Proof of Reliance, Checking Algorithm and Deep Neural Networks have shown that the proposed framework is able to curtail the malicious devices from attacking the network. View Figure