Performance analysis of REM-based handover algorithm for multi-tier cellular networks

The advent of 5G networks, where a plethora of spectrum-sharing schemes are expected to be adopted as an answer to the ever-growing users' need for data traffic, will require addressing mobility ubiquitously. The trend initiated with the deployment of heterogeneous networks and past standards will give way to a multi-tiered network where different services will coexist, such as device-to-device, vehicle-to-vehicle or massive-machine communications. Because of the high variability in the cell sizes given the different transmit powers, the classical handover process, which relies solely on measurements, will lead to an unbearable network overhead as a consequence of the high number of handovers. The use of spatial databases, also known as radio environment maps (REM), was first introduced as a tool to detect opportunistic spectrum access opportunities in cognitive radio applications. Since then, REM usage has been widely expanded to cover deployment optimization, interference management or resource allocation to name a few. In this paper, we introduce a handover algorithm that can predict the best network connection for the current user's trajectory from a radio environment map. We consider a geometric approach to derive the handover and handover-failure regions and compare the current handover algorithm used in Long-Term Evolution with our proposed one. Results show a drastic reduction in the number of handovers while maintaining a trade-off between the ping-pong handover and the handover-failure probabilities.