Design of backpropagated neurocomputing paradigm for Stuxnet virus dynamics in control infrastructure

In the present study, a novel application of backpropagated neurocomputing heuristics (BNCH) is presented for epidemic virus model that portrays the Stuxnet virus propagation in regimes of supervisory control and data acquisition (SCADA) networks using multi-layer structure of neural networks (NNs) optimized with competency of efficient backpropagation with Levenberg-Marquardt (LM) method. Stuxnet virus spread through removable storage media that used to transfer of data and virus to device connected to SCADA networks with ability to exploit the whole system. The reference dataset of mathematical model of Stuxnet virus dynamics is generated by the competency of Adams method and used arbitrary for training, testing and validation of BNCH through NNs learning with LM scheme. Comparative study of BNCH with reference results shows the matching of 4-7 decimal places of accuracy and the further validated through mean squared error-based figure of merit, histograms, and regression measures.

Automated summary

In this study, a novel application of backpropagated neurocomputing heuristics (BNCH) is proposed to study the propagation of the Stuxnet virus in SCADA networks. The mathematical model is built using a multi-layer structure of neural networks (NNs) optimized with the Levenberg-Marquardt method, and the reference dataset is generated using the Adams method. Comparative study shows that BNCH achieves good accuracy and is further validated using metrics such as mean squared error.