Smartphone application developers intensively use cloud services to cope with thestorage, processing, and energy restrictions of mobile devices. Here, Cloud Computingoffers them cost-efficient and dynamic data processing and storage. To this end,Cloud Computing is based on a layered architecture, providing access to resourcesat different layers of abstraction. A cloud provider can offer access to physical infrastructure,called Infrastructure as a Service (IaaS), as well as hire infrastructure froman IaaS provider, enhance it, and subsequently offer it as Platform as a Service (PaaS)or Software as a Service (SaaS). An example of a SaaS using third party infrastructureis Dropbox, which is partly based on Amazon‘s EC2 infrastructure.
However, users are often not aware of the utilization of cloud services. This is furtherexacerbated by the layered architecture of Cloud Computing. As a consequence, usersdo not know about the storage location of their data and the companies involved.This raises users‘ uncertainty about potential privacy invasions, which can cause anaversion to using this technology. Nevertheless, Cloud Computing has many potentialadvantages, so a full denial of Cloud Computing is disadvantageous.
To meet user concerns, we developed a mobile application capable of detecting utilizedcloud services and thus enable the user to make an informed decision regardingthe use of a specific service. We accomplished this by developing an Android applicationthat observes network traffic and detects utilized cloud services. To protectthe user‘s privacy, we gather and analyze data solely on her device. Since usability iscrucial for such a tool, we enabled network capturing without the need for any systemmodification. If an app, such as Dropbox, starts to communicate with the cloud infrastructure,we intercept all network packets, analyze them based on patterns, and thusreveal the hidden cloud infrastructure provider. We evaluated our application andconducted a small-scale user study. Our results show that our approach is feasibleand reveals that different user behaviour results in significantly different cloud usage.