Big data analysis on beam spectrum for handover optimization in massive-MIMO cellular systems

In 5G cellular systems with massive multiple-input multiple-output (MIMO), the process of acquiring the suitable beam for the target cell upon handover would greatly extend handover latency and cause service interruption. This paper proposes a self-optimization network (SON) function to efficiently acquire the target beam with short latency, based on big data analysis of beam spectrum. First, in the training phase, the big data storage of beam spectrum is initially built and dynamically updated by historical handover footprints. Then, in the execution phase, when a user enters the handover region, the suitable target beam is instantly derived by beam spectrum matching between this user and the previous users in the storage, without taking time in target beams field measurement prior to handover. The relevant big-data storage, management and processing in this function are illustrated. Simulation results demonstrate that the beamforming gain of the selected target beam by the proposed function can be greatly improved in non-line-of-sight (NLOS) channels compared with the prior-art approach that solely relies on the strongest beam in the source cell, and the loss compared with the result by field measurement is acceptable.