.png){kind=icon}
.png){kind=icon}
Return
Cloud-Edge Collaborative Task Offloading and Resource Allocation Based on Mobile Computility
.png){kind=icon}
PRE
AI
DOI：10.1016/j.future.2025.108266.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In the cloud-edge collaborative environment, task offloading and resource allocation on the edge-side are primarily based on fixed edge infrastructures, adopting a model of ”cloud-edge vertical integration and horizontal collaboration among edge nodes”. However, the fixed edge infrastructures commonly face the issue of spatiotemporal resource supply and demand mismatch, which to some extent restricts the development of cloud-edge collaboration. This paper considers the on-demand introduction of idle computility resources from increasingly powerful mobile devices and proposes a dynamic task offloading and resource allocation method based on mobile computility within a cloud-edge collaborative framework. Firstly, a task offloading unit architecture for a base station edge system collaborating with the cloud is developed and refined from the holistic architecture of the cloud-edge collaborative IoT application system. By jointly optimizing the offloading decisions and the allocation of computing and communication resources, the problem of task offloading and resource allocation based on mobile computility is modeled as a mixed integer nonlinear programming (MINLP) with time-slot constraints, multi-dimensional resource constraints, and energy constraints of mobile devices. Secondly, in response to the complex interplay among resources, energy consumption and time slots, a resource allocation strategy for l-length task window is introduced. To extend the computility supply time of mobile devices and prevent rapid energy depletion, a mobile device selection strategy based on the concept of opportunity factors is incorporated. Lastly, an online task offloading and resource allocation algorithm that prioritizes tasks with the earliest deadline is designed. Experimental results demonstrate that the proposed method not only significantly surpasses the optimal solution in decision-making efficiency, but also achieves at least a 3% reduction in total energy consumption and a 4% increase in edge offloading ratio compared to other online approaches.