Generative Adversarial Networks Based Industrial Protocol Construction in the Fog Computing

Industrial protocols are critical communication elements for networking components in industrial applications, and plays core rule for security audit and anomalous detection. The emerging of fog computing paradigm makes that intelligence moves down to the industrial devices for enhanced local computational capability. In this work, deep convolutional generative adversarial networks (GAN) is applied for designing industrial protocol construction scheme, in which a discriminator and a generator are respectively established to achieve collaborative training and optimization. Siemens S7 protocol payloads are transformed into gray scale images for texture feature extraction, and a data set of industrial protocols are adopted for implementation. The adversarial training model could be uploaded in the industrial cloudlets, and the proposed protocol construction scheme will launch a perspective for establishing honeypots or honeynets in the fog computing.