AS2M4SoC: An Autonomic Service Scheduling Model and Strategy for SoC Based on Evidence Theory

Service-oriented resource management and application are important issues. The usability of selected resources for a service is a primary factor to computing performance. This paper proposes an architectural service management model (AS²M4 SoC), which realizes autonomic service efficiently through mapping a group of services to the user service registry via resource´s QoS appraisal and user´s credit appraisal strategy. Because of the uncertainty of heterogeneous resources, the resource usability appraisal strategy and the resources allocating method based on D-S evidence theory are established to realize the binding of service with high usability resources, the selecting of QoS-guaranteed resources and the guaranteeing of the users´ tasks QoS. Service scheduling algorithm is presented to realize that high QoS-guaranteed services can priorly map to user´s registry with high credit degree. Simulation experiment demonstrates that our service scheduling model can not only improve the QoS and service efficiency, but also effectively relieve the workload of GSM.