Try to find a problem to solve in a specific application where Cloud may be a solution. Never start by the solution. Be careful not to find a problem to solve using Cloud that is like a “stone soup”, remember the tale? I see a lot of students choosing Cloud as subject because it is a trend topic but the most important contribution in the thesis was a mechanism useful in any other platform.

It is hard to find a subject in this area because, you can see, most of them was already explored until 2015. My suggestion is real-time problems which have specific requirements and constraints. Besides the basic requirements, you can focus on security and safety problems.

The thesis of Marcelo Dutra Ös, that I was the advisor, was presented in 2015 with the title “A community cloud architecture for real-time applications”. He had in mind real-time application for stock exchange market. The thesis in English is freely available at http://www.teses.usp.br/teses/disponiveis/3/3141/tde-14072016-151824/publico/MarceloDutraOs.pdf.

Another issue is to propose a SDN (Software Defined Network) solution to access the cloud where you can customize for a specific application with real-time, security, safety or QoS requirements.

Autonomous vehicular cloud applications is an area that has real-time requirements. It is not a recent subject but is still appealing. Maybe you can find a specific problem to solve or to improve.

Whatever problem you chose, have in mind how you will validate your proposal and the tool to be used. A proposal of a new architecture is very hard to validate mainly in cloud. Otherwise, if you chose a specific question, for instance safety, it is possible to model using some tool like Colored Petri-net, stochastic models or simulators and to show the performance improvement, for instance.

Pin point to your techniques by which you have solves the problem.

Usually, you first do a literature review in a specific field, then you discover the best players in that field (institutes, researchers, conferences, and journals) then from the publications of the pioneering people in the field you focus on their methodology and their suggestions for future work and start from there.

Good luck