Prioritized Traffic-Dependent Task Scheduling Approach in Embedded Operation System

Abstract

This paper proposes a PRIoritized trafFIC-dependent (PRIFIC) task scheduling scheme based on the traffic prediction approach. The PRIFIC manages network tasks by dynamically allocating system resources, so that it can not only prevent the network tasks from starvation state but also improve the utilization of system resources. The PRIFIC is constructed by employing an eigentraffic feature extraction model, a traffic prediction model, and an adaptive task scheduling model. Besides, an additional reference algorithm which is called principal component analysis (PCA) is adopted to reduce the dimensions of traffic features. It can efficiently reduce the computation complexity of the proposed PRIFIC algorithm. In summary, PRIFIC has the advantages of traffic prediction and QoS-enabling. According to the simulation results, the proposed PRIFIC exhibits excellent task scheduling performance in both constant traffic and random traffic applications.