Current students

Network Function Virtualization is leading-edge technology that drives the evolution of telecommunication-provider infrastructure, enabling important advances in networks, and integrating network and Cloud services. An important aspect is optimizing control systems and computing resources. In this dissertation, control and management infrastructures will be investigated, focusing in enhancing the service orchestration based on historical data with machine learning techniques to optimally place VNFs in metro-area networks.
The first part will be dedicated to reviewing the state-of-the-art. After having a solid analysis of the literature, the network planning will be performed to define the reference architecture, the requirements and the cost model. The next step involves the theoretical approach to the problem, in which linear programming models will be developed to provide the optimal solution to be used as benchmark. Moreover, machine learning algorithms will be selected to dynamically determine VNF placement and resource assignment. Then, the solution will be coded to be tested and used in possible demos in realistic scenarios. The demo will be developed in the context of the Metro-Haul European project in cooperation with other partners. The last step will comprise simulations to test the solution in different scenarios and possibly the analysis of results from the demo.