Multi-Granularity Decomposition based Task Scheduling for Migration Cost Minimization

With the development of mobile communication, network technology, and the continuous emergence of intelligent network applications, users' demand for network computing power has increased explosively, which promoted the formation of a multi-level computing power system composed of the end devices, mobile network edge cloud, and center clouds. The terminal and edge computing power resources are limited. The cloud computing power is rich, but the delay is high, so the computing power at all levels needs effective cooperation to meet the quality of service requirements of various ubiquitous computing services. In this trend, cloud computing and edge computing begin to evolve into networked collaborative computing. In this paper, a task scheduling heuristic algorithm based on task cost minimization is proposed for network computing services with a large amount of communication and computation and high delay cost. This method divides the computing tasks of network applications into multiple granularities and schedules the divided sub-tasks, which can improve the utilization of the distributed computing resources and enhance the collaborative scheduling capability of computing and network resources.