The method of stages simulator (MOSS) for modeling variance in soft real-time systems

Soft real-time systems such as those used for streaming video, weather forecasting, and flight plan analysis are becoming more pervasive as technology advances. Tasks found in such systems often exhibit unpredictability in their arrival, service, and deadline behavior. Due to this unpredictable nature, an important characteristic of soft real-time systems is the variance within the arrival, service, and deadline parameters of tasks. With traditional task scheduling algorithms such as Rate Monotonic (RM), Earliest Deadline First (EDF), and Least Laxity First (LLF), the variance of task parameters is typically ignored. However, this variance can directly influence the choice of the best scheduling algorithm, particularly under varying system loads. This paper presents a discrete-event simulation tool that uses the method of stages technique to model system variance in soft real-time systems. This Method of Stages Simulator (MOSS) can be used to estimate performance metrics such as met and missed deadline percentages, system utilizations, and response times. MOSS can also be used to: (1) conduct sensitivity analysis on input parameters such as inter-arrival times, service times, and deadline times, (2) compare the performance of various scheduling algorithms, and (3) discover and evaluate new scheduling algorithms. As an illustrative and motivating example, a pharmacy scenario is presented.