Software-Defined Task Scheduling Strategy Towards Edge Computing

Abstract

Data processing has posed new challenges to transmission bandwidth and computing load under the existing cloud computing architectures. In this paper, a software-defined task cooperative scheduling structure is proposed towards edge computing. Specifically, a clustering algorithm based on two-way selection between idle users and overloaded users is firstly designed, which combines the historical information of idle users and the interest similarity of overloaded users to form the stable cooperative clusters. Then, a sub-task partitioning algorithm based on the optimal delay is presented to achieve the overall optimal delay with the guarantee that sub-tasks are completed simultaneously. Numerical results show that the proposed strategy is not only able to save the data transmission bandwidth significantly, but also achieve the optimal delay while ensuring the stability of cooperative clusters.