Resource-Aware Workload Orchestration for Edge Computing

Edge Computing is an emerging paradigm that improves the quality of experience when dealing with low latency applications. However, the problem of workload orchestration for the offloaded tasks still needs to be solved. In this work, we focus on the offloading decision problem in which execution locations for the incoming tasks are decided within an edge infrastructure or the central cloud. Particularly, we propose a resource-aware based approach to solve the offloading decision problem such that edge server resources are efficiently utilized before offloading to the central cloud. Numerical results from EdgeCloudSim show that this proposed policy outperforms other workload orchestration policies by exhibiting a better virtual machine utilization and fewer failed tasks due to wide area network link failures or congestion as the number of tasks increase.