Automatic RRC Reconfigurations Based on User Mobility Patterns

In this paper we describe how a cellular network extracts information from users’ mobility patterns and applies that for automating and improving radio resource control (RRC). The proposed machine learning system localizes active users with sub-cell spatial resolution based on RRC mobility measurement reports and it matches segments of user trajectories to a database of commonly occurring trajectory segments, so that it can predict possible continuations in users’ movement. The system also includes a metric, which describes the rate of change of the radio environment. Such predictions and metric can automate the information content and timing of RRC messaging. The 3GPP has recently standardized procedures to make handovers faster and more robust. We illustrate by the example of conditional handover how machine learning can control handover messaging and the allocations/releases of radio resources at the network side.