嵌入式系统流水线资源管理模型

Pipelined resource management model for embedded systems

针对嵌入式系统多任务多资源分配问题,提出了一种采用数据流水线的资源管理模型和基于模糊控制规则的自适应作业调度算法。该模型建立一组资源管理服务节点,每个服务节点管理一种类型资源,不同服务节点以流水线形式顺序处理任务作业。当多个不同任务的作业进入同一服务节点,该节点执行自适应作业调度算法进行多任务资源分配。该算法以保证不同任务作业实时性为目标,采用当前任务队列长度为输入,基于模糊控制规则自适应调整任务队列带宽。实验结果表明该模型使任务作业处理速度得到约1.4倍流水线加速提升,并能自适应分配资源,确保多任务作业实时性。

A pipelined resource management proposed to deal with the multi-tasking model and an adaptive job scheduling algorithm based on fuzzy control are resource allocation problem in embedded systems. The model creates a set of resources management service nodes which process task jobs by pipeline. Each node manages a type of resource and the adaptive job scheduling algorithm is used to allocation the muhitask resources. This algorithm＇ target is to en- sure the real-time characteristics of jobs. It takes the lengths of task queue as inputs, and adjusts the resources allo- cation by fuzzy control. The experimental result showed that the model improved about 1.4 times rate of job processing based on pipeline acceleration and ensured the real-time characteristics of jobs by effective adaptive adjustments.