Utility-Based Cloudlet Selection in Mobile Cloud Computing

Abstract

Mobile Cloud Computing (MCC) allows offloading compute-intensive applications execution from resource-constrained Mobile Devices (MDs) to more powerful computing resources, i.e., distant cloud servers or cloudlets. This is a challenging issue due to the situations that Mobile Users (MUs) may face in such context such as the inconstancy of wireless networks, mobility, server load or battery drain. Most cloudlet-based contributions rely on the selection of one resource for offloading that offers good network condition and high computing capabilities. The performance of the selected cloudlet may degrade over time thus leads to the re-selection of another one to perform a seamless service continuity. Our paper proposes the selection of two cloudlets, one ensuring good network connection and one ensuring high performance execution of the application in order to keep the MU "Always Best Connected and Served" (ABC&S) on both levels: network and cloud. We aim at predicting some MU situations and at finding the optimal solution in order to minimize unnecessary handoffs and unnecessary service migration.