Abstract

Edge computing enables devices with insufficient computing resources to offload their tasks to the edge for computing, to improve the service experience. Some existing work has noticed that the data size of offloaded tasks played a role in resource allocation shares but has not delved further into how the data size of an offloaded task affects resource allocation. Among offloaded tasks, those with larger data sizes often consume a larger share of system resources, potentially even monopolizing system resources if the data size is large enough. As a result, tasks with small or regular sizes lose the opportunity to be offloaded to the edge due to their limited data size. To address this issue, we introduce the concept of an emergency factor to penalize tasks with immense sizes for monopolizing system resources, while supporting tasks with small sizes to contend for system resources. The joint offloading decision and resource allocation problem is formulated as a mixed-integer nonlinear programming (MINLP) problem and further decomposed into an offloading decision subproblem and a resource allocation subproblem. Using the KKT conditions, we design a bisection search-based algorithm to find the optimal resource allocation scheme. Additionally, we propose a linear-search-based coordinate descent (CD) algorithm to identify the optimal offloading decision. Numerical results show that our proposed algorithm converges to the optimal scheme (for the minimal delay) when tasks are of regular size. Moreover, when tasks of immense, small and regular sizes coexist in the system, our scheme can exclude tasks of immense size from edge resource allocation, while still enabling tasks of small size to be offloaded.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call