| Real-time tasks are divided into 3 main types:periodic tasks,sporadic tasks and aperiodic tasks.The analysis of real-time tasks is mainly about its response time and s chedulability.To analyze the response time of real-time tasks,the main trend now is to model real-time task models with corresponding parameters,and compute response ti me.Response time stands for the time interval between the release and finish time of a task based on some priorities.The schedulability can be judged by computing respo nse time.If a task with complicated execution needs must be completed by its deadline,cho osing an appropriate model is of great importance for analysis.A good model needs to be both expressive and efficient,but high expressiveness may cause low efficiency du ring tests.Liu&Layland task model is the earliest real-time model.It has high analysis efficiency but low expressiveness.For time automata,it has high expressiveness but costs large amount of time in the process of analyzing.Computing the response time of multiple task sets with static priority on uni-processor is a coNp hard.It faces the problem of space explosion since each task is possible to occupy the processor.As a result,general analyzing techniques are not suitable for analyzing task sets like that.The thesis introduced a modification to analyze multiple task sets.For large numb ers of tasks with complex job sequences,our method can shorten the analysis time to a large extent.The reason is that it can simplify the Cartesian product and reduce the d anger of combinatorial explosion.Our method shows good performance for both stati c and EDF schedulers.For modeling and analyzing tools,our thesis proposed a customer-friendly tool to model systems with directed task graphs.It allows users to establish graph models based on real-time tasks,and compute response time of tasks in any tasks set.Directed task graph balances expressiveness and feasibility,and improves the corresponding level. |
PDF Full Text Request