Holistic Scheduling Of Distributed Real-Time Systems

Distributed hard real-time system is the typical application of real-time system. The behavior of the system has strict deadlines, so tasks and messages of the system must be scheduled. The holistic scheduling of the system means that the processor scheduling analysis and communication scheduling analysis are combined into one framework, which is used to evaluate the time requirement of the system. Scheduling of the distributed hard real-time system is a NP hard problem.Some basic concepts of real-time system are introduced, as well as the scheduling and schedulability analysis of real-time system, and then the Rate Monotonic fixed priority scheduling is discussed. After that, the precedence constraints and release jitter of the end-to-end tasks are given. The response time analysis of arbitrary deadline tasks is discussed. On the base of single processor scheduling analysis and communication scheduling analysis, the holistic scheduling is introduced.CAN-bus is one of the Fieldbuses widely used in the industry field. On the base of the Worst-response time analysis of CAN message, the Best-response time of CAN message and response time analysis with probability bit-stuffing are introduced into the Holistic Scheduling of CAN-bus. An improved holistic scheduling method is provided which decreases the pessimism of the analysis. To prove the validity of the improved method a case of example is given.Time-Trigger and Event-Trigger are basic trigger types in the modern distributed real-time system. It has great significance to choose optimized and reasonable scheduling and analysis method for distributed real-time system. In this paper, after establishing the system architecture and application model of mixed-triggered system, schedulability analysis of Event-Triggered activities under the influence of static cyclic schedule is analyzed. Three alternative ways to compute the cost function, which is used to determine the start time of the Time-Triggered activities, are presented. All these works are the foundations for deeply research about the mixed Time-Triggered and Event-Triggered distributed hard real-time system.