Zero-touch coordination framework for Self-Organizing Functions in 5G

Abstract

Traditional mobile network services are built by chaining together multiple functional boxes on which creation of new services is rather static. With the advent of 5G technology the ability to offer agile on-demand services to the users is mandatory. Therefore lifecycle operations such as service initial deployment, configuration changes, upgrades, scale-out, scale-in, optimization, self-healing etc. should be fully automated steps. Self-Organized Networks Functions (SF) were proposed to provide self-adaptation capabilities to mobile networks on different fronts: configuration, optimization and healing and somehow reduce the error-prone human intervention.Nevertheless, conventional design of these SFs was based on single objective optimization approaches where SFs were considered as standalone agents aiming at one very specific local objective (e.g. reduce the interference or increase the coverage). Thus, complex inter-dependencies between SFs were at some extent unattended, so when more than one function is acting on the network, conflicts are inevitable. A well-studied conflict happens when Mobility Load Balancing (MLB) and Mobility Robustness optimization (MRO) functions are simultaneously set up: without coordination, performance degradation is expected because of the cross-dependencies between both SFs. To cope with these underlying conflicts, we propose a zero-touch coordination framework based on Machine Learning (ML) to automatically learn the dynamics between the selected SFs and assist the network optimization task.