Machine learning in radio connection management

An apparatus for a source network node of a radio access network, comprising:

• at least one processor;
• and
• at least one non-transitory memory storing instructions that, when executed with the at least one processor, cause the apparatus to:
• monitor one or more Key Performance Indicators (KPIs) of radio connections managed by the source network node;
• compare the one or more KPIs to a threshold;
• based on the comparison, detect:
• a handover problem in a predefined area of a cell managed by the source network node, an outage in the cell, or a need to modify a handover execution policy of the source network node;
• based on the detection, determine to perform an explorative handover;
• in response to determining to perform the explorative handover, transmit, within a Radio Resource Control (RRC) release message to a terminal device, configuration information to configure the terminal device to enter, in a state of not having a need to transfer data, an exploration mode, wherein the configuration information comprises:
• a location of the terminal device, a range of a parameter measured by the terminal device, a time of day, and a parameter indicative of a received signal strength measured by the terminal device;
• in response to the terminal device entering the exploration mode, establish a first explorative radio connection with the terminal device, wherein the first explorative radio connection comprises a control plane connection without a user plane connection and operates with a reduced set of radio connection functions compared to a default operating mode;
• receive from the terminal device, via the first explorative radio connection, one or more measurement reports of one or more neighboring cells of the cell, wherein the one or more measurement reports comprise Reference Signal Reception Power (RSRP) of the one or more neighboring cells;
• based on the one or more received measurement reports, determine, from the one or more neighboring cells, a target cell corresponding to a target network node for the explorative handover, wherein determining the target cell deviates from the handover execution policy by selecting the target cell associated with an RSRP value, indicated in the one or more measurement reports, that is lower than a minimum threshold defined in the handover execution policy;
• transmit, to the target network node, a handover request comprising information indicating the explorative handover;
• in response to transmitting the handover request, receive, from the target network node, a handover response message accepting the explorative handover;
• in response to the handover response message:
• transmit, to the terminal device, a handover command to configure the terminal device to perform the explorative handover to the target network node;
• and
• release the first explorative radio connection with the terminal device;
• after the explorative handover, receive, from the target network node, measurement data associated with a second explorative radio connection established between the terminal device and the target network node as a result of the explorative handover;
• and
• based on the received measurement data indicating that a performance of the second explorative radio connection exceeds a performance threshold defined by the handover execution policy, to modify the handover execution policy of the source network node to include conditions corresponding to the explorative handover.