Predicting Device Anomalous Condition in a Collaborated Industrial Environment

Abstract

The industrial environment augments resource-constrained devices to bring services closer to autonomous devices. However, over time, these devices get overburdened due to computational workload, which results in degraded network performance. Therefore, the devices are programmed to share resources with nearby devices. However, due to real-time collaboration, there is the possibility that the device moves to an undefined state and starts behaving maliciously. This can impact the entire collaborative environment laid to meet the industrial products deadline. In this work, we proposed an industrial simulation framework that enables the resource-sharing environment and identifies the undefined device behavior. Further, our detection scheme is based on an intelligent model trained on device behavior through the machine-in-a-loop mechanism and deployed at network intersections, i.e., edge nodes. The proposed technique improved the efficiency of the collaborative network by 30%.