Improving mobile cloud performance using offloading techniques and stochastic models

Abstract

Resource scarcity is a major obstacle for many mobile applications, since devices have limited battery and processing power. The use of cloud computing has been shown to be a feasible alternative to process demanding mobile devices workloads, leading to the research field called mobile cloud computing (MCC). By using the cloud, mobile devices may offload computation to resourceful servers. Many issues related to such a process have been investigated in the past decade, but those related to offloading process still remain. This PhD research has developed a smart MCC offloading strategy for mobile applications. The approach have considered an innovative balanced infrastructure parameters strategy. Another MCC challenge is related to the process of infrastructure evaluation and planning. Evaluating the MCC infrastructure in a deep level of detail may provide to software engineers precise information, guiding their decisions. Instead of evaluating the MCC infrastructure as a black-box, this work proposes to analyze the application at source-code level. This PhD research proposes providing a way for representing method-calls and evaluating mobile cloud applications by using stochastic petri nets (SPNs). The SPNs in this work allow software engineers to understand their applications through a statistic report. Case studies have showed that the proposed techniques are helpful for guiding cloud systems designers and administrators in the decision-making process.