Rate-Variable-Latency service curve as an extension to Network Calculus

The objective of our work is to provide industrial communication based on IP networks with an increased level of determinism and reliability compared to the office IP networks. Our approach is based on the established methods of quality of service used in telecommunication networks and verification of our approach by both empirical and analytical means. In this paper, we propose an extension of network calculus in form of rate-variable-latency (RVL) service curve which is capable of modelling of step changes of latency during data congestions caused by flow aggregation. Furthermore, we introduce several properties of the RVL service curve. Finally, we provide a comparison of the outgoing port congestion simulation using the RVL service curve and our empirical measurements.