Evidence-aware mobile cloud architectures

Abstract

The potential of mobile offloading has contributed towards the flurry of recent research activity known as mobile cloud computing. By instrumenting the mobile applications with offloading mechanisms, a mobile device can save its energy and increase its performance. However, existing offloading mechanisms lack from efficient decision models for augmenting the mobile device with cloud resources on the fly. This problem is caused by the large amount of system’s parameters and their scattered values that need to be considered and characterized merely by the device depending on its contextual needs. Thus, the offloading process still suffers from deficiencies that do not allow a device to maximize the advantages of going cloud-aware. In this chapter, we explore the challenges and opportunities of a new kind of mobile architecture, namely evidence-aware mobile cloud architecture, which relies on crowdsensing to diagnose the optimal configuration for migrating mobile functionality to cloud. The key insight is that by using the massive parallel infrastructure of the cloud to process big data, it is possible to collect offloading evidence from large amount of devices that is later analyzed in conjunction to infer an efficient configuration to execute a smartphone app for a particular device.