LAUSR

With the advent of Industry 4.0, high-quality consumer electronics are being efficiently produced by simulating a virtual model connected to a physical object using the digital twin (DT) technology. Furthermore, high-performance cloud computing technology is being used to simulate resource-intensive consumer electronics DT. Virtual machine (VM)-based DT simulation can simulate various DTs in parallel, but VM is heavy due to the hypervisor and has a slow startup time. In contrast, container-based DT simulation is lightweight and fast-driving and can elastically utilize computing resources. However, it causes a scaling delay problem and affords a degraded DT simulation performance. Therefore, this study proposes intelligence resource scaling (IReS) for efficient consumer electronics DT simulation in a high-performance cloud computing environment. IReS continuously monitors the container’s workload and predicts computing resource requirements using a DLinear model to respond to future workloads. Through predicted computing resource requirements, IReS calculates the optimal number of replicas that can handle future workloads and then performs horizontal autoscaling. The evaluation of the IReS performance shows that it elastically provided computing resources according to the workload changes regardless of the execution time of the consumer electronics DT simulation and optimized the resource cost for DT simulation.