A neutrosophic WPM-based machine learning model for device trust in industrial internet of things

The manufacturing industry is one of the most suitable sectors that can significantly benefit from implementing the Industrial Internet of Things (IIoT) concepts and technologies. The extensive automation in the manufacturing industries leads to many heterogeneous applications based on IIoT, requiring effective integration of many heterogeneous systems and seamless operations across the machines. The problem of integration and seamless operation introduces IIoT as a separate field of study in smart manufacturing that brings many challenges, including security, traceability, trust, and reliability. In IIoT, many devices will be connected using wireless and internet infrastructure to communicate with each other. In such a scenario, the IIoT devices' trustworthiness becomes an essential factor in avoiding injection by malicious machines. Therefore, an intelligent computational model is needed to cluster and classify the IIoT devices' trustworthiness accurately. In this paper, we present a trust model based on the neutrosophic weighted product method (WPM) used by IIoT applications to assess IIoT devices' trust score. The developed model assesses devices' trustworthiness based on the spatial knowledge, temporal experience, and behavioural pattern retrieved from the IIoT devices. Furthermore, the model proposes neutrosophic K-NN clustering and neutrosophic support vector machines (SVM) for classifying the extracted characteristics. The proposed neutrosophic SVM method is able to correctly identify the trust boundaries and produces the final trust score. The simulation results show that the proposed trust model detects the misbehaviour of IIoT devices more accurately than other aggregation methods.