Learning Radio Resource Management in 5G Networks: Framework, Opportunities and Challenges

The fifth generation (5G) of mobile broadband shall be a far more complex system compared to earlier generations due to advancements in radio and network technology, increased densification and heterogeneity of network and user equipment, larger number of operating bands, as well as more stringent performance requirement. To cope with the increased complexity of the Radio Resources Management (RRM) of 5G systems, this manuscript advocates the need for a clean slate design of the 5G RRM architecture. We propose to capitalize the large amount of data readily available in the network from measurements and system observations in combination with the most recent advances in the field of machine learning. The result is an RRM architecture based on general-purpose learning framework capable of deriving specific RRM control policies directly from data gathered in the network. The potential of this approach is verified in three case studies and future directions on application of machine learning to RRM are discussed.