Arrival and Service Curve Measurement-Based Estimation Methods to Analyze and Design Soft Real-Time Streaming Systems with Network Calculus

This paper reviews various methods for estimating arrival and service curves for simple linear functions from experimental data. We experiment with methods found in the literature and discuss the advantages and disadvantages of these methods. We propose two self-developed simple methods, inspired by the literature, to estimate rate-latency service curves that fit our use case to analyze a soft real-time streaming system with network calculus (NC). We also apply three methods to a novel field for performance evaluation with NC: hardware-in-the-loop (HIL) test benches. The performance metrics playback buffer and minimum required pre-buffer time are dimensioned using NC solutions for streaming devices from our previous paper. Linear NC elements are generated in three different ways based on measurements of software processing and network latencies. Playback buffer backlog and delay bounds are calculated based on the three service curve estimation methods using NC. These computed bounds are compared and discussed, and further verified by a discrete-event simulation model with trace-driven input and with distributional fits of processing latencies. The simulation model uses a variation of the workload to prove the system while operating at a soft real-time operating point, with the mean arrival-rate near the mean service-rate and above the minimum service-rate. We recommend using the NC delay metrics with a safety-factor to design the pre-buffer time.