Multi-dimensional optimization for collaborative task scheduling in cloud-edge-end system

Abstract

As we all know, Mobile Edge Computing (MEC) can effectively reduce data transmission delay by scheduling tasks to edge servers. However, current research often fails to comprehensively evaluate the joint impact of key factors such as server location, service placement decision, caching ratio, computing power, computation offloading ratio, and offloading location on the effectiveness of task scheduling, which to a certain extent limits the comprehensiveness and effectiveness of task scheduling strategies. Moreover, in practical engineering applications, it is particularly crucial to comprehensively consider the key factors for the placement of edge devices. In view of this, this paper proposes a multi-dimensional optimization model for task scheduling that jointly considers factors such as Server placement, Service placement, Caching placement, Resource allocation, and Computation offloading (SSCRC) in a cloud-edge-end collaborative system. This model transforms the task scheduling multi-dimensional optimization problem into a Mixed Integer Nonlinear Programming (MINLP) problem to high-quality feasible solutions. To address this complex problem, we adopt a Branch-and-Bound with Parallel Interior Point (BBPIP) algorithm to obtain the optimal solution. Simulation results show that compared with several other schemes, the proposed scheme SSCRC exhibits significant performance improvements in terms of average delay, energy consumption and load balancing.