ML-based Analytics Framework for beyond 5G Mobile Communication Systems

Predictive analysis of cellular network behaviour has many advantages, and will be one of the major breakthroughs on 5G and beyond 5G networks. Knowing the parameters that have impact on reliable network operation and their use in network performance analysis and evaluation are very important. Thus, in case of degradation, intervention is possible before the fault actually occurs and exactly on the faulty equipment. This brings an increase in efficiency with respect to the scenario in which the enhancement solutions are brought subsequent to network events. This paper proposes tracing a predictive behaviour for certain performance indicators of an LTE RAN network and in presenting optimisation solutions of the undesired behaviour. Also, we propose an analytics framework for the access network that can be applied to 5G and beyond 5G RANs. A set of APIs is used in order to extract all the parameters from the LTE RAN Network and store them into a database. After that, the data is exported for further processing and analytics. Results show that the Gradient Boosting algorithm is the most suitable to be used in such a framework.