Quality Checks After Production: TSN-based Industrial Network Performance Evaluation

The automation of quality checks of a product in a smart manufacturing environment poses several challenges for the underlying industrial network. Such a network involves a variety of connected devices, e.g. sensors, actuators, embedded computers, to perform detection of a newly arriving product, visual inspection for quality checks, and classification to decide on the final quality of the product. These devices generate different types of data flows depending on their function. Here, we model the use case of quality checks using visual inspection after production. We consider the industrial network to be based on Time Sensitive Networking (TSN) standards to meet the different data flow quality of services (QoS) requirements. We use OMNET ++ to model our use case. In addition, we extend an existing TSN module implementation to configure the network layer for our application model. We conduct a set of simulations while considering worst case analysis with infinite and finite queue sizes, and realistic data traffic models. Our simulation results show that using a combination of Time Aware (TAS) and Credit-based (CBS) shaping outperforms standard and priority Ethernet queuing strategies and achieves a high delivery ratio of 100% for critical data traffic and 94% for burst traffic, while meeting end-to-end latency requirements.