Analyzing the Performance of Complex System Architectures Using Network Calculus

In an industrial context, performance analysis of complex system architectures is important for enhancing capability through architectural design and ensuring quality of service in both worst-case and best-case scenarios. Network calculus is founded on the principles of system theory and min-plus algebra, which employs min-plus functions to model and control constraints on service and arrival processes. It is a widely used analytical method for evaluating such architectures. However, most existing approaches often overlook the effects of buffer sizes and the number of virtual channels on system performance, leading to unsuitable performance for time-sensitive applications. In this paper, we highlight the limitations of existing queueing theory, scheduling theorems, and network calculus-based approaches. We introduce our approach, which accounts for the effects of the amount of buffers and virtual channels on throughput and is designed to support multiple virtual channels, improving the accuracy of performance analysis.