Study On Critical Technology Of Embedded Real-Time Systems

Embedded Real-time System has been wildly used in multimedia communicate,online- business-process, manufacture control,traffic control and so on in these yeas. So, more and more people attach importance to it now. The policy of process Scheduling is very important for Real-time System, so I research the real-process Scheduling.This article is based on reading source code of Linux, reference many books and articles. The idea of this article is useful for the Real-time process Scheduling Algorithms of other Operating System.Process management is the most pivotal part of Operating System, its design and implement will directly affect the performance of the whole system. Though there are a lot kinds of Real-time process Scheduling Algorithms,but many of them have some disadvantage. The main kind of process Scheduling Algorithms are FIFO (First In and First Out),RM (Rate Monotonic),EDF (Earliest Deadline First),HVF (Highest Value First) and so on.The biggest difference between Real-time System and other system is that it must contents enough time to finish the process. In Real-time System,the correctness of output not only depends on the logical result of compute but also depends on the time which it finish the result. The important require of Real-time System is it must respond to and deal with affair which come from outside and inside in defined time slice beforehand, we call the pre-defined time as Deadline.This article analyses the process Scheduling of Linux,and draw a conclusion : Its PRI does not deal with the deadline of real-time process and the pressure of process, to improve Linux's process method for real-time process; this article gives a kind of optimized and ameliorated Scheduling Algorithms. The Algorithm based on the importance of process; consider the process's pressure in deadline at the same time. It builds the queue of executable process, which based on priority.In the original Scheduling Algorithm of Linux (we call it Algorithm before optimized below), the main gist to build process PRI is the processes worth (also be the importance of the process), then system selects the process who's PRI is the highest to enter CPU and run. The implement of Linux's Scheduling Algorithm falls under the speed of CPU and the status of load. As Real-time System Scheduling Algorithm, it should consider the two concepts both: the power-value of process and deadline, to avoid the illogical use to CPU's ability. So the article gives us a new Scheduling Algorithm, in the new Algorithm, the expression to account process's priority is (1):Definitions of each Parameters in the expression are: vi denotes process's priority; wi means process's worth( also be the importance of the process) , Pi denotes process's estimating execute time; di denotes absolute deadline ; Ti denotes the process's submit time ;(d-Ti) is relative deadline, also be the biggest waiting time of the process, in the relatively deadline, the process should been executed, or else the process abortion; Pi/(d-Ti) we called pressure extent,this value is more bigger, it shows the task is more pressing , k is coefficient gene.Main jobs innovations of this paper are as the following. His paper makes innovations in the following aspects: (1)The article study priority reverse except studying and discussing process Scheduling Algorithms and analyses before process Scheduling Algorithm's fault and then put forward my own resolving. The resolving which we put forward is easy to be realized and can resolve priority reverse on condition that think over process priority in critical section and thus it has pretty high practicality value. For task' node, we should add the task's deadline, affair's estimating execute time, tasks' submitting time, the time tasks need to use, remained time of the tasks, weight -value and so on. (2)The important role of knowledge sharing in Modern enterprise production is analyzed in this paper, the method of utilizing knowledge resource efficiently in dynamic distributing sy...