Intelligent Management of 5G Mobile Network based on the Comprehensive Learning with deep learning

The deployment of heterogeneous networks, application of the dynamic spectrum and the use of unlicensed frequency spectrum are the basic technological solutions accompanying the deployment of 5th generation (5G) mobile communications networks. The synergy of the simultaneous use of these technology solutions will allow a significant increase in network capacity and quality of service (QoS), which is a prerequisite for a massive 5G deployment in the future. It is necessary to emphasize the fact that the mutual optimization of these aspects is a highly complex problem, which results in the fact that application of conventional optimization approaches are in most cases inefficient. A promising solution is the application of modern trends in machine and deep learning (e.g. generative adversial networks) that, due to their adaptability, have the potential to optimize complex dynamic and time-varying systems. The basic goals within the project are thus the design of new heuristic algorithms based on the application of machine and deep learning to optimize the performance of the 5G network, verification of these algorithms in realistic models and studies as well as experimentally validation of the proposed methods on functional USRP devices.