Research On Task Priority Assignment Configuration In Synchronous Reactive Model

With the rapid development of computer and information technology,the real-time embedded system is widely used in daily life.The existing embedded systems are usually described by corresponding task models.Synchronous reactive model,which may allow the communication among different tasks,is raising more and more attention.But among synchronous reactive models,zero-time semantics may cause data loss and data misread.In order to solve these problems,some buffering protocols are proposed,using the buffer to store the corresponding data and to achieve the goal of semantic protection.Because of the huge consumption of memory by using the buffer in buffering protocols and the limitation of memory in the embedded system,the study of optimization of buffer usage becomes crucial.The research shows that there are two factors that effect the buffer usage in the task set:The first one is the buffering protocol,namely how the task set protects the semantic by using the buffer.The second one is the relative priority relations in the task set.However,the existing fixed-priority assignment algorithms are lack of the consideration of the task inter-priority relationships when they assign the priority for the tasks.To solve this problem,this thesis firstly studies the different buffering protocols in the synchronous reactive model,such as the single reader buffering protocol and the dynamic buffering protocol.Secondly,this thesis proposes a more reliable method of buffer computation combined with the real-time scheduling theory.This method can compute the amount of buffer usage more accurately,avoiding the waste of buffer caused by the shortage of task scheduling analysis of traditional buffering protocols.The study of the dynamic buffering protocol for the synchronous reactive model shows the difference of buffer usage among different priority readers and analyses how the task relative priority relations affect the buffer usage in the task set,and this thesis gives the corresponding optimized configuration.At last,according to the optimized configuration,this thesis proposes a new heuristic writer-first assignment algorithm,which considers the influence of the inter-relationship and the relative priority,and chooses the priority for a task according to the rules.In the end,this thesis analyzes the correctness of the algorithm,and proves the optimality of algorithm rules.A great deal of experimental data approves that the algorithm in this thesis reduces the number of buffer usage in the task set observably.