TrustSys: Trusted Decision Making Scheme for Collaborative Artificial Intelligence of Things
Geetanjali Rathee,Sahil Garg,Georges Kaddoum,Bong Jun Choi,Mohammad Mehedi Hassan,Salman A. AlQahtani
DOI: https://doi.org/10.1109/tii.2022.3173006
IF: 12.3
2022-11-13
IEEE Transactions on Industrial Informatics
Abstract:Many IoT-based applications have inherited the artificial intelligence of things (AIoT) techniques to explore new services and benefits of smart recording and monitoring generated information. However, hundreds of hacking incidents caused by highly sophisticated attackers have generated serious risks, where they compromised various IoT sensors for their benefits, impeding the growth of AIoT. Various security schemes have been proposed in the literature; however, it is critical to determine the legitimacy of AIoT devices in real-time scenarios during the initial deployment of the network. Therefore, this article aims to provide a secure, reliable, and trusted decision-making scheme using multiattribute methods in collaborative AIoT. The proposed system uses backpropagation and Bayesian's rule to ensure a fast and accurate decision. In addition, agent-based modeling and population-based modeling trust schemes are used to compute the legitimacy of the communicating model. Further, the proposed system is validated over various security measures against the various decision-based conventional methods such as Fuzzy c-means, REPTree, and random tree in terms of time, accuracy, replay attack, data falsification attack, recall, region of convergence, and F-Measure. The proposed mechanism achieves 93% improvement over accuracy and attack identification against existing mechanisms.
automation & control systems,computer science, interdisciplinary applications,engineering, industrial