Real-time Heterogeneous Systems, Integration Of Dynamic Scheduling Algorithm

Real-time heterogenoous systems are used widely in the field of fligh controland avionics, processing control, telecommunication, imagine processing andinternet applications. In this paper, a new intelligent bramework for integrateddyamic scheduling of real-time heterogeneous systems, called FIDSH, ispresented. Basing on FIDSH, a new adaptive integrated dynamic schedulingalgorithm for real-time heterogeneous systems, called IADSH, is proposed.In FIDSH, an integrated task model for real-time heterogeneous systems ispresented. This task model reveals the heterogeneity of systems and uniformlycharacterizes hard and soft real-thime in real-time heterogeneous systems byusing imprecise computing model and introducing the factor of task importance andvalue. The new task provides an efficient way to improve the scheduling successratio and system benefit obtained by schedule. Moreover FIDSH firstly introducesthe intelligent control mechanism in the integrated dynamic scheduling for real-timeheterogeneous systems by embedding an intelligent controller in it.Basing on FIDSH, a new adaptive integratd dynamic scheduling algorithm forreal-time hererogenous systems, called IADSH, is proposed. This algorithmincludes two parts that are called basic schedule and intelligent control respectivelyIn the part of basic schedule, a new methd for integratd dynamic schedulingof hard and soft real-time tasks in real-time heterogeneous systems, called IDS, ispresented. IDS is based on heuristic searching and implements the integratedschedule in real-time heterogeneous systems with uniform form successfully IDSimproves the scheduling success ratio by introducing a new task assignment policyand a Qos (Quality of Service) degradation policy for soft real-time taSks. Moreover,a new metric called a metric is proposed to trade off the effeC of taSk's dcadline andimportance on the schedule decision, whereas oPtimizing system benefit obtalned bythe schedule without decreasing the scheduling success ratio.To make the integrated dynamic scheduling obtaln stable schedulingperformance, we introduce inielligeni control to IADSH. In this part, feedbackcondol theory is maPped to illtegrated dyntalc schedule for real-time heterogeneoussystems. Fuhaermore, the new intelligent colltrOl policy, taSk flow adjustment policyand parameter adjustment policy are preselltCd to make IADSH achieve the stablescheduling pefformance.TO evaluae its performance, we haVe conducted extensive simulation studies fOrIADSH. SAnulation resultS demonstrate:l) IADSH highly imProves the scheduling success ratio. ItS schedu1ingsuccess ratio is aVeragely 30% higher than that of other related algorithms,such as myopic algorithm.2) IADSH dramatically improves the system benefit successfully withoutdecreasing the scheduling success ratio.3) IADSH shows fairly stable schedule performance when workflow changesin the syStCm.