Research On Predictability Of Timing Parameters Of The Distributed Real-Time Tasks Online

With rapid development of computer technology and network technology, the applications of distributed real-time systems are much popular in many fields such as information application, medical instrument, intelligent vehicle, industrial process control, and communication device. At all times, real-time systems have to take all kinds of methods to keep the stable and normal operation, like complicate algorithm, for the real-time tasks have the special request of time. These methods affected the performance of real-time system more or less. For this reason, people hope badly to prove the performance of real-time system right after satisfy the real-time request. If the runtime of the real-time task could be predicted, it will affect real-time system in system design, tasks schedule and resource distribution. Correct predict results do their jobs quite well.The purpose of this paper is to implement the prediction of real-time tasks' runtime based on schedulable analysis in distribute online environment. Since there are the factors such as task preemption, resource competition, timer interrupts, I/O collision, some real-time tasks with random behavior and so on. There are a number of schedulable analysis algorithms are used to predict that whether a task could be scheduled on a fixed computer and the runtime of the task if available. This paper will catch the execute information of CPU, and then predict the runtime of real-time tasks with the information and schedulable analysis.The runtime of real-time tasks on a fixed computer is related to the execution of the computer. Based on this fact, this paper will contribute the CPU's execution and schedulable algorithm, and predict the runtime of real-time tasks with the result of schedulable analysis. In this way, whether a real-time task's time constraint can be satisfied on a fixed computer could be detected before the task is scheduled. For distribute applications, the benefit of this method is that a new arrival task always can be appoint to a proper computer or the optimal computer. The correct prediction of the runtime of real-time tasks will prove the designation, resource utilization and performance of a real-time system, and offer some reference for similar researchs.