Predictive Handover Optimization.

In future networks, cell densification is envisioned to provide higher data rates. In such dense deployments, moving user equipments (UEs) may experience handovers more frequently. The increased number of handovers causes more signaling overhead for the network. Furthermore, it might also result in more frequent handover failures (HOFs).The advances in artificial intelligence research open up the door for prediction techniques of channel state information (CSI). This paper focuses on a handover optimization scheme that relies on predicted CSI to reduce the number of HOFs and unnecessary handovers while maintaining the signal quality as high as possible. The proposed scheme relies on solutions to an integer programming problem to assign the UE to a base station at every time instance. To reduce the computational cost, we propose an approach that uses solutions to a linear relaxation of that problem, and we compare the performance of the two schemes.