Dynamic Mobile Cloudlet Clustering for Fog Computing

Although mobile devices have obtained more and more capabilities such as CPU power, memory, and battery, they still cannot satisfy the requirement of high resource consuming applications. Fog Computing is one of the solutions for offloading the task of a mobile. However the capability of fog server is still limited due to the high deployment cost. In this paper, we propose a dynamic mobile cloudlet cluster policy (DMCCP) to use cloudlets as a supplement for the fog server for offloading. The main idea is that by monitoring each mobile device resource amount, the DMCCP system clusters the optimal cloudlet to meet the requests of different tasks from the local mobile device. In simulation results, we demonstrated that our proposed policy can cluster the approximated optimal cloudlets for different resource demands of a task in a real-time processing time compared to the optimal method.