Revolution of Self-Organizing Network for 5G MmWave Small Cell Management: From Reactive to Proactive

Correlation-Based Feature Mapping of Crowdsourced LTE Data

“…For example, some challenges arise from the fact that in future networks end-users may have their own femtocells (small and low-power cells) deployed which in turn requires better automated coordination and self-optimization from the network management layer [9]. Another challenge is that traditional SON functions are reactively detecting and recovering from network faults instead of proactively prevent faults occurring [11,159]. In addition to the limitations associated with current SON management use cases, more automation is expected to be needed in many network management tasks in emerging 5G scenarios [101,3].…”

“…On the one hand, the ongoing work in improving machine learning capabilities of SON control loops for both the existing and upcoming use cases will have a great emphasis also in the future research. [95,1,159] On the other hand, the interplay and interaction of management functionalities in a multi-domain environment is also seen as a key component in the virtualized and service-oriented 5G network management [111,3,1,122].…”

“…From the perspective of utilizing SON-related machine learning models effectively, some fundamental issues need to be addressed. First, the increasing Introduction number of machine learning models for SON use cases implies a crucial need to define how to select the suitable algorithms with respect to different objectives [159]. Second, the cross-domain (e.g., across different technologies and industries) adaptation of machine learning models needs to be addressed in the future SON research [95].…”

“…These contributions can be used to discover SON functions based on their metadata and verify their functionality in various contexts. The future perspectives of autonomic network management consider SON management as an important building block [95,1,159]. Characterization of SON functions has been studied in view of the dynamic mapping of operator goals for SON function configurations [55,85].…”

“…The results would benefit SON function and context modelling tasks in cross-platform management systems as semantic definitions of context attributes require effort and collaboration in order to ensure seamless information exchange. Moreover, in the 5G environment, it is likely that the number of SON use cases increase significantly [138,110] and more competing algorithms are developed for existing use cases [159,95]. These two aspects make the context-specific evaluation of the higher number of SON function operations even more important.…”