QUDOS: quorum-based cloud-edge distributed DNNs for security enhanced industry 4.0

Distributed machine learning algorithms that employ Deep Neural Networks (DNNs) are widely used in Industry 4.0 applications,such as smart manufacturing. The layers of a DNN can be mappedonto different nodes located in the cloud, edge and shop floor forpreserving privacy. The quality of the data that is fed into and processed through the DNN is of utmost importance for critical tasks,such as inspection and quality control. Distributed Data ValidationNetworks (DDVNs) are used to validate the quality of the data.However, they are prone to single points of failure when an attackoccurs. This paper proposes QUDOS, an approach that enhancesthe security of a distributed DNN that is supported by DDVNsusing quorums. The proposed approach allows individual nodesthat are corrupted due to an attack to be detected or excluded whenthe DNN produces an output. Metrics such as corruption factor andsuccess probability of an attack are considered for evaluating thesecurity aspects of DNNs. A simulation study demonstrates thatif the number of corrupted nodes is less than a given thresholdfor decision-making in a quorum, the QUDOS approach alwaysprevents attacks. Furthermore, the study shows that increasing thesize of the quorum has a better impact on security than increasingthe number of layers. One merit of QUDOS is that it enhancesthe security of DNNs without requiring any modifications to thealgorithm and can therefore be applied to other classes of problems.