A Novel Traversal Search-Based D2D Collaborative Offloading Approach forWorkflow Application in Dynamic Edge Environment

Cheng Qian,Gansen Zhao,Haoyu Luo

DOI: https://doi.org/10.1007/978-981-19-4546-5_14

2022-01-01

Abstract:D2D (Device-to-Device) collaborative offloading is a kind of offloading paradigms in the Mobile Edge Computing (MEC) environment, which allows edge devices to offload workflow tasks to the adjacent mobile devices with idle resources. As a supplement to the End-Edge-Cloud collaborative offloading paradigms, the D2D collaborative offloading can relieve the MEC's performance bottleneck caused by the limited capacity of edge servers and increased offloading requirements. However, the mobility and capacity fluctuation of mobile devices makes it difficult to guarantee the quality of service of the D2D collaborative offloading of workflow application. For this issue, we propose a dynamic environment oriented D2D collaborative offloading approach to ensure that the workflow offloading requirements in edge devices can be responded with a minimum cost before the deadline. Specifically, we first encode the feasible offloading strategy in a rough order of cost, and then perform Fibonacci traversal to fast approximate the optimal offloading strategy. In the remaining iterations, a uniform traversal is performed in the neighborhood of the optimal offloading strategy. Finally, the best offloading strategy ever traversed is applied to practice. We do not pursue optimality but real-time performance, so as to adjust the offloading strategy online with the same procedure. Experimental results show that our proposed approach is highly cost-effective with low decision overhead and high success rate.