Deep learning based context classification for cognitive network management

Self-Organizing Network (SON) is a crucial technology characterizing the behaviour of future mobile networks. Complex cellular networks' deployment, operation and maintenance are managed autonomously by multiple SON functions (SFs) with dedicated objectives. Designing appropriate configurations of SON is challenging as it requires comprehensive modelling of all their interactions. Previous works addressed this point and proposed cognitive management solutions to learn the optimal SON configurations through direct communication with the network. Some of these works have overcome SON's shortcomings, such as limited flexibility and adaptability to changing environments. This paper discusses a context-specific policy that uses machine learning (ML) to learn the network environment. This way, the cognitive management solution finds the convenient SON configuration for each network context that is identified with a higher value of automation.